9Th International Conference on Social Science Methodology Book of Abstracts 11-16 September 2016, Leicester, United Kingdom Hosted by Leicester Sociology

9th International Conference on Social Science Methodology Book of abstracts 11-16 September 2016, Leicester, United Kingdom hosted by Leicester Sociology

TABLE OF CONTENTS: BOOK OF ABSTRACTS

Contents A ...... 26 Abramczuk, K...... 26 Survey and Behavioural Measures of Social Trust ...... 26 Alexandrakis, S...... 28 A software suite of tools to assist with complex survey authoring, testing and enactment ...... 28 Amati, V., Schoenenberger, F. and Snijders, T...... 30 The Generalized Method of Moments for Estimating the Parameters of Stochastic Actor- oriented Models...... 30 Amaturo, E. and Punziano, G...... 31 Blurry boundaries: Internet, Big-New Data and Mixed Methods Approach ...... 31 Andersen, H. and Mayerl, J...... 34 Analyzing socially desirable response behavior using paradata ...... 34 Angel, S., Heuberger, R., and Lamei, N...... 36 How (and why) does the poverty headcount change if income data from registers are used? Evidence from the Austrian SILC ...... 36 Anguera, M.T. and Sánchez-Algarra, P...... 39 Valorizing direct and indirect observational methodology: A challenge of Mixed Methods Data Analysis and Interpretation ...... 39 B ...... 41 Bakker, J...... 41 Breakoffs in general population web surveys and its effect on data quality: a comparison between smartphone, tablet and desktop users...... 41 Barlas, F., Thomas, R., Fahimi, M. and Buttermore N...... 42 Can Survey Data be Too Clean?: Data Cleaning and Bias Reduction ...... 42 Barnat, M. and Bosse, E...... 43

The challenge of creating meta-inferences: Combining data representing institutional and individual perspectives on first-year support in higher education ...... 43 Baroni, W.S...... 45 A Genealogy of the Autobiography in Contemporary Society and Social Sciences. Autobiographical Discourse, Communism and Feminism ...... 45 Barth, A. and Schmitz, A...... 48 Towards a relational understanding of data quality ...... 48 Beatty, P. and Lineback, J. F...... 49 Developing a Program on Emerging Methods for Improving Survey Measurement at the ...... 49 Becker, J., and Witte, N...... 50 Analysing places with reconstructive biographical methods...... 50 Bednarowska, Z...... 52 Why sociologists should follow political scientists and introduce spatial econometrics to their research? ...... 52 Belli, S...... 55 Manage Social Conflict for a Slow Internet Connection in Online Meetings ...... 55 Bengali, M...... 56 Seeking God's Assistance to Govern: A Comparative analysis of Islamization Policies in Pakistan and Egypt ....56 Bergmann, M., Hanemann, F. and Scherpenzeel, A...... 57 A responsive monitoring design using paradata to improve data collection in the Survey of Health, Ageing and Retirement in Europe (SHARE) ...... 57 Bergner, B. and Spellerberg, A...... 59 Monitoring Real-Time Stress Experiences in Space – Data Collection, Operability and Methodical Challenges of User-Generated Emotion Data research? ...... 59 Bertolini, S. and Musumeci, R...... 61 Changing preferences on work in the transition to parenthood? Evidence from qualitative longitudinal research on Italian couples ...... 61 Beullens, K...... 63 Interviewer effects among the elderly ...... 63 Bhattacharyya, A...... 64 A methodological quest set forth for Indian software Industry ...... 64 Blasius, J., Thiessen, V. and Nenadic, O...... 65 The dirty data index – assessing the quality of survey data in international comparison ...... 65 Brinkgreve, C. and van Daalen, R...... 66 A Sociological reading of Diary Fragments during wartime...... 66

Bolli, T., Egg, M. E. and Rageth, L...... 67 Does Vocational Education and Training (VET) Improve the Youth Labour Market Performance? ...... 67 Borgs, C., and Schnell, R...... 69 Linking against all odds: Using Administrative data with modern Privacy Preserving Record Linkage Techniques ...... 69 Brandes, U. and Schoch, D...... 71 When the centrality of actors is most obvious, centrality indices explain nothing at all...... 71 Budde, E., Knill, C. and Preidel, C ...... 71 A Matter of Timing: Religion and the Governance of Moral Issues ...... 71 Burzan, N. and Eickelmann, J...... 72 Event-oriented processes in museums: challenges of a multi-methodical approach ...... 72 C ...... 74 Calderwood, L., Gaia, A., Wood, M. and Alvarez, P. C...... 74 Using para-data for real-time trade-offs between fieldwork cost, response bias and response rates: experiences from Next Steps ...... 74 Campagnolo, G. M., Williams, R., Alex, B. and Acerbi, A...... 76 Social data science and the sociology of professions: reflections on methodological synthesis ...... 76 Carty, A. M...... 78 Conversations across the Valley: Using narrative therapy in film to create safe spaces for dialogue between fractured rural communities ...... 78 Čehovin, G., Manfreda, K. L., Bosnjak, M...... 80 Meta-analyses in survey research: A systematic review ...... 80 Cernat, A. and Oberski, D...... 81 The multi-trait multi-error approach to estimating measurement error ...... 81 Cernat, A. and Sakshaug, J...... 82 Nurse effects in survey biomarkers ...... 82 Cheung, G. and Johnson, J...... 83 Use Cases by capturing structured metadata to document complex CAI instruments using DDI3 ...... 83 Cirucci, A...... 84 Narrative Demographics ...... 84 Class, F.* ...... 87 Do “Grievance Statistics” measure quality of life in the GDR? Investigation of the validity of a new administrative data source by means of individual data ...... 87 Comanaru, R.-S. and d’Ardenne, Jo ...... 89

Young people and harmonised questions – findings from cognitive interviews ...... 89 Connelly, R., and Gayle, V...... 91 Administrative Data Quality and Measurement: A case study cross-validating UK Birth Records ...... 91 Cottingham, M. and Erickson, R...... 93 Using Audio Diaries to Investigate Emotion ...... 93 Coulton, J...... 94 The Art of Noise: Using Audio Soundscapes in Spatial Research ...... 94 Cusatelli, C. and Giacalone, M...... 95 Information and Communication Technology for Judiciary Big Data Management ...... 95 Czepek, J...... 97 The history of the return on human capital investment: Opportunities and challenges of multiple administrative data ...... 97 Czepek, J. and Tisch, A...... 99 Analysing public administrative data from two different organisations: A historical analyses of mothers’ pension credits in Germany ...... 99 D ...... 101 D'Antonio, V ...... 101 Privacy, publicity and self disclosure in the Big Data Era: From the «right to be alone» to Oversharing ...... 101 Danciu, A., Le Corgne, S. and Mairot, A...... 103 An overview of metadata and data management at the French Center for Socio-Political Data (CDSP) ...... 103 Das, B. and Subudhi, R...... 105 Influence of socio-economic factors in DSMM ...... 105 Davis, Peter ...... 106 The health effects of patients switching medicines after subsidy changes...... 106 de Leeuw, E., ...... 107 To gamify or not: an experiment into the effectiveness of gamification in an online survey ...... 107 Decataldo, A., Amico, A. and D'Alessandro, G...... 108 Longitudinal research on students careers. A quasi-experimental design on Sapienza University of Rome ...108 Decataldo, A., Amico, A. and D'Alessandro, G...... 110 Management and Analysis Strategies of Administrative Databases: Transforming Data into Longitudinal Vectors ...... 110 Decataldo, A. and Denti, F...... 112 Increasing participation rates and completeness of questionnaire compilation in web survey. An experimental design of research ...... 112

Dechant, A. and Rinklake A...... 114 Longitudinal qualitative interview data - comparisons within and between couples over time ...... 114 Delamont, S...... 116 Time to kill the Witch ...... 116 Demirkol, E...... 117 The Balance between Emotions and Research: A reflection on studying emotions among transnational migrant families ...... 117 Deviatko, I...... 119 “Virtual labs”, “social telescopes” and other metaphors of innovative online methods: new opportunities and new challenges to validity of social science research ...... 119 Dietrich, M. and Mey, G...... 121 Towards a Visual Grounded Theory Methodology: A framework for image analysis in culture studies ...... 121 Dluhopolskyi, Oleksandr ...... 122 Comparative studies of social capital Development in Ukraine Post-Revolution of Dignity 2014 ...... 122 Dorau, R...... 124 Comparability of job histories between datasets ...... 124 Dörfler, T., Brandt, S. and Klärner, A...... 126 Towards a Relational Analysis of Social Space – Methodological Implications for the Research on Space and Place ...... 126 Dunning, M...... 129 Social processes and the development of ‘jihadist terrorists’ ...... 129 Durand, C., Charest, A.-S., Valois, I. and Ibarra L. P...... 130 The use of longitudinal multilevel models to understand change in trust in institutions over time and between countries ...... 130 Durrant, G., Sturgis, P., Maslovskaya, O. and Brunton-Smith, I...... 131 Interviewer effects on response latencies in a face-to-face interview survey ...... 131 E ...... 132 Eberle, J., Heining, J., and Müller, D...... 132 A modern Job Submission Application to access IAB’s confidential administrative and survey research data...... 132 Edwards, D...... 134 Understanding Everyday Participation: Making the case for "short and dirty" multi-sited ethnographies .....134 Enzler, Heidi Bruderer ...... 136 Air Travel for Private Purposes. An Analysis of Survey Data Combined with Spatial Data...... 136

F ...... 138 Fischer-Neumann, M...... 138 Religious Affiliation and Regime Support among Turkish Immigrants in Germany ...... 138 Fearon, K...... 139 Reproductive decision making in women with Turner Syndrome: an exploratory study ...... 139 Fuchs-Sawert, L...... 141 Maps in Discourse Analysis: The Construction of Space in Socioecological Conflicts in Peru ...... 141 Fernandez, K...... 143 Sampling challenges in ethnography of homeless youth groups ...... 143 Fitzgerald, R...... 144 ‘Attempting to reduce comparison error across the survey life cycle. Illustrations from the European Social Survey (ESS) and future solutions from the Synergies for Europe’s Research Infrastructures in the Social Sciences initiative’...... 144 Furakhutdinov S. Sheludkov A. and Bystrov E...... 146 Maps and spatial analysis for visualization and interpretation of cluster analysis results...... 146 G ...... 148 Gaia, A...... 148 The influence on substantive results of the sampling strategies adopted in qualitative and quantitative paradigms: Evidence on men who pay for sex ...... 148 Garner, M...... 150 Do as I do, not just as I say: can students learn from their teachers’ research experience? ...... 150 Gartner, J...... 152 Motivational Effects of Gamified Surveys compared to Standard Online Surveys ...... 152 Gebel, T., Hense, A. and Schork, F...... 153 How to deal with convergent, complementary, and divergent findings in mixed methods research? ...... 153 Giehl, C. and Mayerl, J...... 155 Explaining systematic measurement errors using dual-process models of response behavior and the attitude towards surveys ...... 155 Gilbert, E., Stella, C., Fisher, K., Calderwood, L., Cleary, A., Huskinson, T. and Gershuny J...... 157 A mixed-mode approach to measuring young people’s time-use in the UK Millennium Cohort Study ...... 157 Girkontaitė, A...... 158 Agency-structure dialectics in discourse analysis: networks of discourses ...... 158 Gollini, I. and Murphy, B...... 160 Latent variable models for multiple network views ...... 160

Goodwin, J. and O’Connor, H...... 161 Jephcott's Unpublished Notebooks as Biographical Method ...... 161 Górnicka, B...... 162 Nudism, homines aperti and the methodology behind my study of emotions ...... 162 Graham, S. and Ravn, T...... 163 Ideals versus realities: exploring methodological influences on eliciting narratives of solo mothers ...... 163 Graeff, P. and Fuchs, S...... 165 Bridging the historical link between corruption and social trust: reproducing a contemporary effect with historical data ...... 165 Granda, P...... 167 Creating Structured Metadata: Quality Profiles and Ex-Post Data Harmonization ...... 167 Green, J. and Binsardi, B...... 168 Frontier Rescinding in Classic Grounded Theory ...... 168 Gupta, A...... 171 Time Versus Depth: challenges to conducting an ethnographic study into “the field” of education ...... 171 Gupta, P...... 173 Key Issues for better Academic Governance and Performance in Indian Higher Education Sector ...... 173 Guth, J. and Nelsen, B...... 174 The Changing Role of Religion in Public Support for the European Union...... 174 H ...... 175 Haan, M., Toepoel, V. and Lugtig, P...... 175 Device Use in Online Panel Surveys ...... 175 Haas, G.-C...... 176 How does asking for linkage consent in the beginning of the questionnaire affect data quality? ...... 176 Hájek, M...... 177 Analysing the controversial dimension of public discourse – combining corpus linguistic with L. Boltanski’s approach to the study of public controversies ...... 177 Halford, S. and Savage, M...... 179 Symphonic Social Science & the Challenge of Big Data ...... 179 Hameister, N., Klaus, D. and Engstler, H...... 181 Age and Data Quality? Lessons learned from the German Ageing Survey (DEAS) ...... 181 Hanna, E...... 182 Researching male infertility online: Bringing researcher emotions into the debate ...... 182

Haken, M...... 184 Videography of audience emotions in religious events...... 184 Hansell, A...... 185 Spatial analysis of administrative datasets at small area level ...... 185 Hering, L...... 187 How to trace the Intrinsic Logic of Cities using historical material ...... 187 Hlebec, V...... 188 Secondaru Quatitative Data in Explanatory Mixed Methods Research Design ...... 188 Hlebec, V., Kogovšek, T., Manfreda, K. J. and Lobe B...... 190 Usability of Offline and Online Focus Groups for Pretesting Sensitive Questionnaires ...... 190 Höfig, C...... 191 Different target groups, same error? - Investigating the robustness of mode differences regarding item nonresponse ...... 191 Höglinger, M. and Diekmann A...... 192 Do special techniques for surveying sensitive topics provide valid measurement? A validation design that detects false positives ...... 192 Hu, Y., ...... 194 Netnography, Forum Discussion and Focus Group: A Comparative Experiment on Three Modes of Qualitative Data Collection...... 194 Hülle, S., and Sakshaug J...... 195 Record Linkage Consent: Assessing its Determinants and Magnitude of Bias ...... 195 Hunter, Michael D...... 197 Revisiting Spinach and Ice Cream: A dynamical systems analysis of why social science is so challenging ...... 197 Hunter, M., Ou, L., Reddy, S., and Chow, S.-M...... 198 Comparing Numerical and Analytic Solutions to Stochastic Differential Equations ...... 198 I ...... 199 Ils, A. and Buchler, S...... 199 Changing Representations of Motherhood and Fatherhood in Germany’s Media ...... 199 Inna, B...... 201 How much time should we allow to participate? How fieldwork duration affects data quality...... 201 Israel, G. and Newberry, M...... 203 Effects of Stem and Response Order on Response Patterns in Satisfaction Ratings ...... 203 J ...... 205

Järv, O., and Masso, A...... 205 Validating the implementation of mobile phone data: an example of spatial segregation research ...... 205 Johnson, J...... 207 Discovery, provenance, harmonisation and re-use of longitudinal studies ...... 207 Jurado-Guerrero, T., Domínguez, M. and González, M...... 209 Analyzing couple’s life courses: (un)doing gender in dual-earner couples in Spain ...... 209 K ...... 211 Kandt, J...... 211 Socio-cultural space over time: the geography of population dynamics in Great Britain ...... 211 Kanu, W...... 213 Religiosity, Gender and Political Participation in Rural Areas of Imo State, Nigeria ...... 213 Kawulich, B. and D'Alba, A...... 215 Teaching Qualitative Research Methods Online with Second Life ...... 215 Keller, R...... 216 SKAD and the methodologies of sociological discourse research ...... 216 Kelly, G. A...... 217 An Examination of Underage and Adult Binge Drinking in Rural North Dakota ...... 217 Khavenson, T...... 218 How well School Age Children Report Family SES. Systematic Review ...... 218 Kim, J., Shin, I. and Yang, D...... 221 Identifying respondents who are very difficult to contact using interviewer’s reports ...... 221 Knappertsbusch, F...... 222 Qualitative Methods and Causality in Mixed Methods and Multimethod Research – A Methodological Comparison...... 222 Knight, V...... 223 Concept-formation, Complexity and Social Domains: Investigating Emotion(s) in a Prison Setting ...... 223 Krause, R.W., Huisman, M., Steglich, C.E.G. and Snijders, T.A.B...... 224 Multiple Imputation for Network Analyses ...... 224 Krebs, D...... 225 Agree-Disagree versus Item Specific Scales – The Impact of Different Response Formats on Response Behavior ...... 225 Kropp, P...... 227 Comparing regional delimitations by Newman’s modularity approach ...... 227

Krusell, S...... 228 Ethnic Cleavages in the labour market –Which concept to use for defining a target group? ...... 228 Kuiper, R...... 229 Combining Statistical Evidence from Multiple Studies Introduction ...... 229 Kunz, T., and Fuchs, M...... 231 Respondent burden and satisficing in grid questions: A comparison of early versus late positioning in a Web survey instrument ...... 231 Kumar, M. and Kumar, A...... 233 Changing socio-economic condition and livelihood of Aged society of tribal community in eastern Uttar Pradesh, India ...... 233 Kumar, A...... 235 Demographic and Physical Well Being of Elderly Tribal People in North India ...... 235 Kutschar, P...... 237 Data Quality of Survey Research among Older Populations – Item-Nonresponse and Interviewer Effects in a Survey examining Pain of 655 Nursing Home Residents ...... 237 L ...... 239 Lamprianou, I...... 239 Errors and Biases in mixed-mode administration...... 239 Lamprianou, T. A...... 240 The length of Likert response scale (number of options) and the quality of measurement: a double experimental design ...... 240 Landrock, U...... 242 How Real and Faked Interviews Differ in Explaining Political Participation ...... 242 Landrock, U. and Best, H...... 244 Selecting a Sample for Qualitative Interviews on Corruption...... 244 Lang, V...... 246 Scaling Sensitive Factorial Survey Analysis ...... 246 Lang, V. and Diewald, M...... 247 The added value of twin data analyses for causal inference in the social sciences ...... 247 Langer, W...... 249 Longitudinal analysis of repeated cross-sectional surveys: The usefulness of the multilevel model ...... 249 Law, C...... 250 Researching Reproduction with Men: Methodological Reflections ...... 250 Layne, L...... 252

‘A Bigger Picture’: The Ethnographic Case Study as an Under-utilized Method in Studies of Reproduction ....252 Lazarevic, P...... 254 Rating Your Health: An Examination of Non-Health-Related Factors and Differential Item Functioning in the Self-Rating of Health ...... 254 Lefeldt, J...... 255 Stories behind Pictures. The Photograph as a Medium of Memory ...... 255 Leguina, A. and Miles, A...... 256 The temporalities of cultural participation practices in the UK: life course dynamics and mobility effects .....256 Leszczensky, L...... 257 Can Panel Data Resolve Questions of Causal Ordering–and If So, How? ...... 257 Letterman, M. C...... 259 Use of Geographical Information Systems (GIS) for Sampling and Fieldwork Monitoring in Developing Countries ...... 259 Lin, J.-P...... 261 Data Science as a Foundation of Integrating and Enriching Administrative Data: The Case of Constructing Taiwan Indigenous Peoples open research Data (TIPD) Based on Taiwan Household Registration Administrative Data ...... 261 Lineback, J. F., Ashmead, R. and Slud, E...... 263 Using Propensity Score Weighting Adjustments to Examine Survey Mode Effects ...... 263 Linklater, A...... 264 Political Symbols and Civilizing Processes in Long-Term Perspective ...... 264 Liston, K. and Maguire, J...... 266 On Being ‘Irish’: negotiating identity and the Involvement-Detachment balance in archival Research...... 266 Lotto, R...... 267 Supporting the researcher in sensitive research – a case study of patient and public involvement ...... 267 Love, G...... 269 Is every abortion a tragedy? Vulnerability and resistance in abortion research ...... 269 Lübcke, T...... 271 Social research in the harm zone: How to follow in the wake of complex maritime search and rescue missions...... 271 Lübcke, T...... 274 Videography in dynamic environments – restrictions on and opportunities for multiplicative GoPro camera usage ...... 274 M ...... 276

Mack, A...... 276 The Process of Migration: Identity, Affect and Established Outsider Relations ...... 276 Mahapatra, B...... 277 Flexible Data Collection Methods and Data Quality In Survey Research ...... 277 Mahapatra, T...... 278 Qualitative research ...... 278 Manderscheid, K...... 280 A Field Approach to Spatial Structures: international comparisons of their structuring force on social inequality ...... 280 Manzoor, H., Mollona, E., and Rashid, M...... 282 Digging Deeper: Researcher’s emotions as a ‘methodological tool in understanding board dynamics ...... 282 Martini, N...... 283 Grasping the Multidimensionality of Relational Space. A Strategy for Data Collection...... 283 Marttila, H. T...... 285 Post-foundational Discourse Analysis: New Research Program in Relational Sociology ...... 285 Maslovskaya, O., Durrant, G., and Smith, P...... 287 Improving the Efficiency in Interviewer Call Scheduling: Analysing Final Outcome and Length of Call Sequences in a Longitudinal Context ...... 287 Matthes, B. and Unger, S...... 289 Determining recall errors in retrospective life course data – An approach using linked survey and administrative data ...... 289 Matjašič, M. and Vehovar, V...... 291 The Efficiency of Nonparametric Methods in the Detection of Response Time Outliers in Web Surveys .....291 Mavletova, A., Tholmogorova, T. and Gavrilov, K...... 292 Risk communication among children: The effect of a gamified design ...... 292 Mayerl, J. and Stocké, V...... 293 Moderation Effects of Response Latencies and Meta-Judgments on Response-Effects and Attitude-Behavior Consistency ...... 293 Mc Namara, C...... 295 The time-saving benefits of using Grounded Theory Methodology for observational studies in the social sciences ...... 295 Menold, N. and Bosnjak, M...... 296 Summarizing 60 years of Research on the Number of Response Categories in Rating Scales: First Result of a Meta-Analysis ...... 296

Menold, N., Landrock, U., Pellner, N. and Winker, P...... 297 The Impact of Payment and Participation on Interviewers’ Accuracy in Face-to-face Surveys: Investigations from a Field Experiment ...... 297 Mergener, A...... 298 Testing Validity of hypothetical Recruitment-Decisions using a Factorial Survey ...... 298 Metzler, A., and Fuchs, M...... 299 Using response time to predict survey break-off in Web surveys ...... 299 Meulemann, H...... 301 Makes religion happy – or makes happiness religious? Religiosity and life satisfaction between early and late mid-life ...... 301 Meyer, D...... 302 Social Network Analysis Meets Bibliometrics: Toward a Visualization of Academic Power in Economics, Political Science, and Sociology ...... 302 Mika, T...... 303 The first twenty years of employment trajectories in Germany: A comparison of the birth cohorts 1935 to 1974 ...... 303 Mikucka, M. and Valentova, M...... 304 Employed or inactive? Cross-national differences in coding parental leave beneficiaries in European Labour Force Survey data...... 304 Mohadjer, L...... 305 Minimizing Total Survey Error in a survey of out-of-school children in developing countries ...... 305 Mohr, A.H., Lindsay, T. and Sell, A...... 307 The Influence of Visual Design on Creative Divergent Thinking in an Online Survey ...... 307 Mondo, E...... 308 The Role of Religion in the European Parliament: An Analysis of MEPs’ Vote on Abortion and Stem Cell Research (1998-2013) ...... 308 Mueller, G...... 309 Charting the Pathways Through the Space of Welfare Regimes: On the Use of QCA for the Construction of Topological Social Maps ...... 309 Müller, J., and Garrido, N...... 311 Sensor Based Data for Social Sciences Research. Using Sociometric Data for Microlevel Insights ...... 311 Müller, S...... 313 Studying the influence of the direct living environment: examples with the georeferenced German General Social Survey 2014 ...... 313 Müller, S., Schweers, S., Kinder-Kurlanda, K., Siegers, P. and Zenk- Möltgen, W...... 315

Studying the influence of the direct living environment: Examples with the georeferenced German General Social Survey 2014 ...... 315 Munro, C. and Couch, P...... 316 Extending DDI to support the harmonisation of cohort variables: a new metadata registry ...... 316 Muraleedharan, S...... 318 Visualizing Democracy: Mapping ‘Dalit’ subjectivity in India...... 318 Murphy, P...... 319 A Re-appraisal of Elias’s ‘Scenes from the Life of a Knight’: A case study of the influence of heteronomous preoccupations on methodology...... 319 Musik, C...... 320 An Ethnography of Trade Fairs as Field-Configuring Events ...... 320 N ...... 322 Naldini, M., Santero, A. and Torrioni, P. M...... 322 Gender Dynamics within Couples using nested longitudinal qualitative data ...... 322 Nascimento, L...... 324 From codex to autocoding: reflections about CAQDAS, Digital and Figurational Sociology ...... 324 Nešporová, O...... 325 Qualitative panel data – three waves of interviews with parental couples ...... 325 Neumann, R. and von Hermanni, H...... 327 Charitable giving in the field - evidence from a quasi-experiment at bottle refund machines in Germany .....327 Nigel, F...... 328 Changes and continuities in online social research ...... 328 Nkolo, J. C...... 329 Using CAPI to Improve and Evaluate the Quality of Socio-Economic Surveys ...... 329 Nkolo, C et al...... 330 Experiences and Challenges in Data Collection Monitoring from Bukoba (Tanzania) to High Wycombe (UK) Experiences and Challenges in Data Collection Monitoring: From Bukoba (Tanzania) to High Wycombe (United Kingdom) ...... 330 Nosek, V. and Netrdova, P...... 332 Individual and regional determinants of unemployment in Czechia: multilevel analysis ...... 332 Nóżka, M...... 333 Mental maps of people experiencing exclusion: A study on the effect of subjective experience of exclusion on the way of valuing and mapping physical space...... 333 O ...... 335

O'Connor, H. and Goodwin, J...... 335 Fieldnotes, Marginalia and Paradata in Youth Employment Restudies, 1960-1985 ...... 335 O’Dwyer, A...... 336 CDA: Reflecting on problems associated with interpreting language & contextualizing discourse ...... 336 Odella, F...... 338 Longitudinal network data and time sequencing of structural changes. An analysis of serialization and competition in the organization of a cultural market...... 338 Ogresta, J. and Ružojčić, M...... 340 The Quality of Retrospective Reports about Occupational History in Web Survey ...... 340 Onaka, F...... 341 Examinations as a Pivot of Social Processes ...... 341 Ortmanns, V...... 342 Can survey characteristics help? - Trying to explain inconsistencies in the distributions of the education variable with survey characteristics ...... 342 Otte, G...... 344 Using mixed methods to investigate social inequalities and symbolic boundaries in a local club and discotheque market ...... 344 Oud, J., Voelkle ...... 346 SEM Based CARMA Time Series Models for Arbitrary N ...... 346 P ...... 347 Pennell, B. – E., Hibben, K. C., Lyberg, L. Mohler, P. and Worku, G...... 347 A Total Survey Error Perspective on Comparative Surveys ...... 347 Pereira, M. J...... 349 Capturing Emotion in the Statements of German Political Prisoners in Paraná during the Second World War ...... 349 Petschick, G...... 350 Researching social processes combining ethnography and interviews ...... 350 Petschick, Grit ...... 352 Problems arising by comparing two Case Studies ...... 352 Petzold, K. and Eifler, S...... 353 Presumed Behaviour and Real Behaviour in a Normative Everyday Situation ...... 353 Plugor, R...... 354 Opportunities and challenges in using administrative data for research purposes: two case studies ...... 354 Pollien, A. and Stähli, M. E...... 355

Promises and failures of nonresponse surveys ...... 355 Pretto, A...... 357 Bipolar photo elicitation ...... 357 Purwaningrum, F. and Fakhri, A...... 358 Q ...... 360 Quinlan, C...... 360 The Four Frameworks Approach: A new tool for teaching research methods ...... 360 Quinlan, C., Baldwin, L. and Knight, V...... 362 Methodological approaches to research on emotions in criminology and in criminal justice studies ...... 362 R ...... 363 Reeves, L. S...... 363 Challenging hierarchies: The role of visual research methods in participatory development communication 363 Redd, R. and Veltri, G...... 365 The Analysis of Cultural Processes and Social Identity Formation in the Re.Cri.Re Project ...... 365 Respi, C. and Sala, E...... 366 What is the impact of using personalized salutations on survey participation and data quality? Results from an experiment on Italian graduates...... 366 Rho, Eugenia ...... 368 Emergence of Membership, Naturalization, and Boundary Objects in the Personalized Social Networking Site ...... 368 Rivas, C., Tkacz, D. and Cruickshank, D...... 369 Automated ‘real time’ thematic analysis of large volume survey freetext responses: methodological issues in using an intelligent system to capture the human voice ...... 369 Reed, K. and Ellis, J...... 370 Creating maps of the mortuary: adopting mobile and sensory methods to examine the role of imaging in foetal and neonatal post-mortem ...... 370 Roberts, J...... 372 The ontology of early labour (and the difficulties of talking about it) ...... 372 Robinson, T...... 373 How do health researchers in New Zealand use quasi-experimental methods to evaluate new programs? A review of published studies...... 373 Robinson, T...... 374 The use of three methods to evaluate a health intervention to reduce readmissions to hospital in older people. An exploration of quasi-experimentation and managing threats to validity...... 374

Roche, M. and Dimbath, O...... 375 Sampling an urban district: lessons from Augsburg’s "Universitätsviertel"...... 375 Rosen, C...... 377 “Accessing the political field of Berlin and Madrid from a spatial perspective” ...... 377 Rüttenauer, T., and Best, H...... 379 The Causes of Environmental Inequality: Evidence from Spatial Time-Series Analysis in Germany ...... 379 Roberts, C...... 380 Do mode effects matter in cross-national surveys? An assessment of measurement invariance across data collection methods and countries ...... 380 Rhodes, B...... 383 Monitoring data collection from afar: Methods and lessons learned from data collection in Pakistan ...... 383 S ...... 385 Sakshaug, J...... 385 Quantifying the Prevalence of Analytic Errors in Secondary Analyses of Complex Sample Survey Data ...... 385 Sakshaug, J., Schmucker, A., Kreuter, F., Couper, M., and Singer, E...... 387 Evaluating Active versus Passive Consent Procedures to the Transfer of Federal Contact Data to a Third-Party Data Collection Agency ...... 387 Sand, M., and Kolb, J-P...... 389 Using Spatial Information for CATI-Samples ...... 389 Sanduk, M...... 390 Can the society be simulated as plasma fluid? ...... 390 Santos, F...... 392 Spatial Analysis at the Fringes of Europe: Field Notes from French Guiana ...... 392 Santorso, S...... 393 Researching Social Process in Prison: Consideration on Mixed-method and Analytical Dualism ...... 393 Saramago, A...... 395 Sarracino, F., and Mikucka, M...... 396 Estimation bias due to duplicated observations: a Monte Carlo simulation ...... 396 Savinskaya, O...... 397

Exploratory sequential design: toward the inductive and deductive transfer of knowledge1 ...... 397 Sauerborn, E...... 399 Understanding Emotions in Hermeneutic Processes ...... 399 Sawert, T...... 401

A ‘Dead Language’ as Profitable Investment? A Quasi-experimental Study on the Influence of Learning Classical Languages at School at Point of Career Entry ...... 401 Schmucker, A...... 403 Quality, analytic potential and accessibility of linked administrative, survey and publicly available data ...... 403 Schnable, S. A...... 405 Pregnant with Possibility: The Importance of Visual Data in (Re)Presenting Queer Experiences of Reproduction ...... 405 Schneider, S...... 407 The shortcomings of long list showcards: measuring education with an adaptive open question and semi- automatic coding ...... 407 Schnell, R...... 409 Putting people on the map without revealing their location ...... 409 Schnell, R. and Thomas, K...... 410 The Effect of Formal Question Characteristics on Design Effects: A Cross-national Study Us- ing the ESS ...... 410 Schoonenbooma, J., Pullen, A., Griffioen, D., and de la Croix, A...... 412 Integrating qualitative observation data and personal characteristics: overcoming the dilemma of sample size ...... 412 Schuman, A...... 414 Punctuated Ethnography: using social time to plot fieldwork plans ...... 414 Schwarzenbach, A...... 416 Youth-Police interaction in a Multi-Ethnic Society ...... 416 Schwengler, B. and Kropp, P...... 417 Stability of functional labour market regions ...... 417 Sendelbah, A. and Vehovar, V...... 419 Effects of questionnaire characteristics on web survey page response times: analysis of an archive of web surveys ...... 419 Sendra, E...... 421 Illustrating festivals research through the still and moving image: Finding an academic platform for documentary films ...... 421 Sieber, S., and Schmidt, T...... 422 The Influence of an Up-Front Experiment on Respondents’ Recording Behaviour in Payment Diaries: Evidence from Germany ...... 422 Simpson, A...... 423 ‘Studying Up’: Themes of and issues of the inverted ethnographic gaze ...... 423 Simpson, M...... 425

Singer, H...... 426 Simulated Maximum Likelihood for Continuous-Discrete State Space Models using Langevin Importance Sampling ...... 426 Smietana, M...... 428 Good science, reflexivity and ethnography: Methodological choreography of a sociological sponsored project ...... 428 Smith, T...... 430 Using the Total Survey Error Paradigm to Maximize Comparability in Cross-National Surveys...... 430 Smuda, R...... 431 Tensions in field access – A reflection ...... 431 Snijders, T. and Ortmann, M ...... 433 Networks in Social Contexts: the Settings Model ...... 433 Spencer, N...... 434 Giving researchers confidence to use quantitative methods: overcoming the fear of choosing the wrong technique ...... 434 Springer, S...... 436 Cognitive testing using different interview modes in different cultural contexts – the role of age ...... 436 Stadtfeld, C...... 437 Relational Events: Comparing Actor-oriented and Tie-oriented Models ...... 437 Stähli, M., and colleagues of the ‘International Surveys’ team of FORS...... 437 Threats to comparability: beyond translation and wording ...... 438 Stamm, I., Eklund, L. and Liebermann, W...... 439 Crowdsourcing as research method: hype, hopes, and hazards...... 439 Stathopoulou, T., and Papageorgiou, H...... 440 Radicalized Orthodoxy? Civic discontent and religious identification in times of crisis...... 440 Steele, J., and Ferrer, E...... 441 A dynamic systems approach to relationship affect over time: An application of the Exact Discrete Model. .441 Stein, P. and Bekalarczyk, D...... 442 Longitudinal modeling of the future development of occupational status in the third generation of migrants in the context of demographic transition by means of a dynamic micro simulation ...... 442 Stevenson, A...... 444 Stolz, J...... 445 A theory of social games as a metatheory for mixed methods research ...... 445 Stoop, I., Denies, K., Douhou, S., Halbherr, V., Kappelhof, J. and ...... 446

Minimising unnecessary variation and optimising quality and comparability in the European Social Survey ..446 Strey, J. P...... 449 “You’re a white chick, what are you researching?”: Reflections on using creative methods to counter researcher fatigue ...... 449 Struminskaya, B...... 450 Systematic review and a meta-analysis of studies on panel conditioning ...... 450 Szabó, J...... 452 Youngsters' leisure time and social stratification of cultural consumption at an Eastern European festival ...452 T ...... 454 Tayactac, J B., Gerardo A S., and Guerrero, L...... 454 Quality Control Challenges to Modernized Scientific Social Weather Reporting in the Philippines ...... 454 Teti, A. and Gross, C...... 456 Consistency of Vignette Judgments among Older Respondents ...... 456 Thomas, R K., Barlas, F M., Buttermore, N R., and Smyth, J D...... 457 Acquiescence Bias vs. Salience in Response: An Experimental Test ...... 457 Thomsen, U...... 459 Filling up the gap: Imputation of missing Working Time Information in the German Employment Data ...... 459 Thomas, R., Barlas F M., and Buttermore, N M...... 461 Thomas, R K., Barlas F M., and Buttermore, N M...... 463 Toward Better Grid Formats for Online and Mobile Surveys ...... 463 Toepoel, V...... 464 Adapting Surveys to the Mobile World: Data Chunking in the Dutch Probability-Based LISS Panel ...... 464 Tome, R S., Roberts, C., and Joye, D...... 465 The Impact of Mode Effects on Measures of Subjective Well-being ...... 465 Tormos, R., Rudnev, M., and Bartolomé, E...... 466 Justifiability of Euthanasia in OECD Countries: Patterns of Change and Age-Period- Cohort Analysis ...... 466 Tranmer, M., Marcum, C., Morton, F., Croft, D., and de Kort, S...... 467 Using the Relational Event Model (REM) to investigate the temporal dynamics of social networks...... 467 Trübner, M...... 468 Header Images in Web Surveys – Curse or Blessing? ...... 468 Tuvikene, T., Neves Alves, S., and Hilbrandt, H...... 469 Turhan, Z...... 470 Conceptual framework and reflexivity for the examination of the engagement process in a sensitive

qualitative research ...... 470 U ...... 471 Ueblacker, J...... 471 Mixed Methods-Mixed Research Synthesis in Urban Sociology: The Example of Gentrification Research in Germany ...... 471 Unfried, M., and Buder, F...... 473 Increasing Respondents’ Engagement with Incentivized and Implicit Interviews: Learnings from Behavioral Economics ...... 473 Unger, S...... 475 Measurement invariance between employed and unemployed persons - An example using a multi-item health measure ...... 475 V ...... 476 Vacchiano M., Martì J. and Yepes, L...... 476 The use of social networks in the age of uncertainty. Young adults’ strategies along the job search...... 476 Valentova , M...... 478 Did the escalation of the crisis of 2008 affect the perception of immigration-related threats? A natural experiment...... 478 Valér, V...... 479 The Quality of Online Survey Data Using Prize Draws ...... 479 van Egmond, M., and van de Vaart, W...... 481 Data quality in asynchronous online focus groups: the directive versus the non-directive moderator style ...481 van de Vaart, W...... 483 Exploring flexible data collection methods for vulnerable, older populations ...... 483 van Houten, G., Schwarzer, S. and Simmons, K...... 485 Using Timestamps for the Evaluation of Data Quality ...... 485 van Vlimmeren, E., Moors, G., and Gelissen, J...... 486 Developing a diagnostic tool for detecting response styles in survey research ...... 486 Vandenplas, C., and Loosveldt, G...... 488 Monitoring the evolution of the fieldwork power: Illustration based on the seventh round of the European Social Survey ...... 488 Vidovićová, L., and Petrová Kafková, M...... 489 Who should interview older respondents about ageing-related topics? The effect of interviewer and interviewee characteristics on the subjective indicator of interview quality ...... 489 von Scheve, C...... 491

Core relational themes as a method for reconstructing emotion from text ...... 491 Vogl, S. and Fuchs, M...... 493 Item-Nonresponse in surveys with children and adolescence...... 493 W ...... 494 Wan, A...... 494 Model Averaging in Sociological Research ...... 494 Wagner M., and Bergmann, M...... 495 Cultural response styles in the evaluation of people´s social contacts in 11 European countries ...... 495 Wagner, C., du Toit, J. and Venter, S...... 497 Wang, C...... 498 Factors Influencing Non-Response in Surveying Taiwan’s Bureaucrats ...... 498 Warner, U. and Hoffmeyer-Zlotnik, J. H.P...... 499 International Harmonized Labor Status Concepts and Measures for the Analysis of Social Structures across countries and cultures: A questionnaire module for comparative social surveys ...... 499 Weidenhaus, G...... 501 Methodological Approaches to Space ...... 501 Weis, C...... 503 Wearing out my welcome? Negotiating authority and access when recruiting participants with the help of gatekeepers...... 503 Weller, A...... 505 The visual as modus operandi ...... 505 Wengrzik, J...... 507 Web Surveys versus Other Survey Modes: An Updated and Extended Meta-Analysis Comparing Response Rates ...... 507 Weßling, K...... 509 Wetzels, M...... 511 The problems of practicing video graphic methods at the example of religious and sportive mass events ....511 Wijnant, A. and Elshout, S...... 512 Mobile devices in web panels: a case study of how to upgrade an existing panel to be ready for mobile device respondents ...... 512 Wolf, C. and Roßteutscher, S...... 513 Wulfe, M...... 514 Using the World Bank's Survey Solutions to Maintain Data Quality during Fieldwork: Examples in South Africa and Tonga ...... 514

Y ...... 515 Yam, S. - C. J...... 515 Econopolitpsychoanthroculturalsociological Reality: Hidden Discourses of the Early 21st- Century Social Sciences ...... 515 Zaccaria, D., Sala, E. and Guaita, A...... 516 Non-response, attrition and age in a cohort study of older people in Italy: The case of InveCe.Ab study ...... 516 Zajc, N...... 518 Current experience with mixed mode including web mode in opinion surveys in official statistic ...... 518 Zangger, C...... 519 Bringing space into the equation. Using spatial econometrics to untangle neighbourhood effects on educational outcomes ...... 519 Zifonun, D. and Naglo, K...... 521 Core Activity, Event and Crisis: Ethnographies in Football Worlds ...... 521 Zimmermann, T...... 523 Assessing the causal relationship between students’ aspirations and school success – a comparison of analytical procedures ...... 523 Züll, C. and Scholz, E...... 525 Effects of Left-right Scale Design in a Cross-national Perspective ...... 525

CONFERENCE ABSTRACTS

Alphabetically by Surname (A – Z)

A G M S Y

B H N T Z

C I O U

D J P V

E K Q W

F L R X

CONFERENCE ABSTRACTS

Alphabetically by Surname (A – Z) A

Abramczuk, K. (University of Warsaw, Institute of Sociology, Poland) Survey and Behavioural Measures of Social Trust The proposed paper focuses on the relationship between experimental and survey measures of social trust. It discusses their applicability in the context of trust radius and the number of trust recipients. It also briefly discusses the role of both experimental and survey measures of trust in the analysis of determinants of social action and the need to construct measures of risk attitudes suitable for social context. The analysis is based on a small scale study consisting of a short survey and several experimental games. There are two parallel threads in the literature on measuring social trust. The first one is based on survey measures that are used in large scale national and transnational polls such as the European Social Survey. They typically adapt standardized sets of questions. The credibility of the resulting trust indeces has been discussed for a long time. Some researchers choose to rely solely on measures of civic engagement instead. Others try to augment traditional questionnaires in order to increase their accuracy. The second line of research on trust measurement originates from behavioural economy and uses experimental games such as the Trust Game or the Prisoner’s Dilemma. These measures are based on behaviour. Unfortunately, they also have many drawbacks, the primary one resulting from highly abstract game scenarios that seem to reduce their external validity. Both approaches to measuring trust levels are fallible but in different ways. It is therefore promising to analyse the relationship between them, seeking ways to extract the best of both of

them and apply mixed methodologies. Some efforts in this direction have already been made. One of the most interesting is research run within the German Socio-Economic Panel (SOEP). The current proposal is based on a small scale exploratory study. The study consists of two parts: a survey and experimental games. The emphasis is on two problems. First, I explore the problems of trust radius and construction of survey measures differentiating between in-group trust and generalized trust. I contrast them with experimental games played with in-group and out-group partners. Second, I analyse the potential differences in trust manifestations when the number of trust recipients varies. Specifically, separate measures for one-to-one trust and group trust are analysed and contrasted with behaviour in paired and group experimental games. There are two additional topics covered in the study. First, the generalizability of the derived survey measures is analysed in the context of their usefulness for predicting involvement in social action. Second, the theoretical problem of viewing trusting behaviour as a special case of risky decision is discussed. Here I introduce two different survey measures of risk attitudes and analyse their correspondence with experimental choices. Keywords: General trust, Survey measures, Experimental games

Alexandrakis, S. (Netelixis Solutions) A software suite of tools to assist with complex survey authoring, testing and enactment SURVANT (SURVey Authoring and eNactment Toolkit) is a suite of software tools that allows technical and non-technical users alike to collaboratively author a survey, test it during development, enact it and collect the data in a straightforward process without the need to understand or get involved in any software development. Researchers author their surveys in Microsoft Word using simple text, structured in tables. On demand and at the click of a button, SURVANT processes the text, checks its validity and enacts the survey, allowing data to be collected according to the survey specification. Collected data can be stored securely “offline” on the local computer in XML format, in a database server or in cloud storage. During survey authoring, because of the use of Microsoft Word as the editor, SURVANT does not require researchers to learn a new software tool or install additional applications on their system. Survey specifications, being Word documents can be easily edited, exchanged, marked and compared for changes, facilitating collaboration. Furthermore they can be easily versioned and stored in any document management system in order to maintain document history or survey version history, if required. During survey development, due to its one-click enactment capability, SURVANT allows automated testing by the authors/collaborators themselves. This removes the need for manually testing the question order, routing mechanism, branching and data input/output scenarios which is complicated, tedious and error-prone, potentially saving days or months’ worth of work. In addition, SURVANT enables one-click survey demonstration, where collaborators can visualise and run the actual end-product on their own computers as it is being developed, without the need to access a different system or re-install new software for example. At the end of the development iterations, once a survey has been finalised and released, modifying it or continuing a study with additional sweeps or waves of participants can be as simple as modifying the corresponding Word documents describing the surveys or adding new

ones. No further software creation or modifications are required as the SURVANT enactor software always remains the same and only the survey specifications change as appropriate. Alongside being easy to use, SURVANT provides powerful features such as multi-lingual survey authoring, rich metadata capture, complex routing and text substitution, custom reporting, survey documentation and extensive look-and-feel customisation. In addition, SURVANT enactor can be instructed to collect data from external systems such as hardware instruments and databases. For example, it may be used to automatically extract measurements from a blood pressure monitor device for a survey participant or read patient data from an EPR system in a clinical setting. SURVANT is designed to be flexible, extensible and interoperable with other systems in the survey lifecycle, facilitating reproducible research and component reuse. Towards this aim, adoption of a standardised survey metadata representation is being considered in the form of the Data Documentation Initiative’s “DDI Lifecycle”. This allows SURVANT to import and enact surveys created with different tools but also export structured, reusable metadata, compliant with emerging standards. Keywords: Metadata; Data Collection; CAI

Amati, V., Schoenenberger, F. and Snijders, T. (University of Konstanz, University of Konstanz, University of Oxford and University of Groningen)

The Generalized Method of Moments for Estimating the Parameters of Stochastic Actor- oriented Models. Stochastic actor-oriented models (SAOMs) are applied to examine the role of selection and influence processes given network panel data, collected by observing a network and a behaviour over time in a panel design. The parameters of SAOMs are estimated by a stochastic approximation version of the method of moments (MoM), where statistics defining the population moment conditions correspond to the parameters in a natural way. The standard MoM cannot incorporate more population moment conditions than parameters. Here, we propose to apply the generalized method of moments (GMoM), using more statistics than parameters, to account for more information when estimating the parameters of SAOMs. We concentrate on statistics depending jointly on the network and the behaviour, because of the special importance of their interdependence. For some examples we study the resulting gain in statistical efficiency and we discuss algorithmic issues that have to be solved in order to obtain a stable procedure.

Amaturo, E. and Punziano, G. (University of Naples Federico II, Italy and Gran Sasso Science Institute – GSSI, Italy) Blurry boundaries: Internet, Big-New Data and Mixed Methods Approach Nowadays the Internet, which more and more takes the form of a place of daily life, is configured as a varied scenario essential to understand many phenomena and complex social issues. If the research does not take into account this container of information its potential resulting limited. The reality that is generally investigated should be conceived as a whole and the web becomes an integrant part. Social researchers have to cope with an exponential growth of new data, plus or minus user-generated and freely available on line, in the inability to be anchored to the traditional research practices and conceptual categories. The web mediated research, in this manner, is already transforming the way in which researchers practice traditional research methods transposed on the web: for example it will be considered the online ethnographic research (ie. netnography) or the social network analysis that actually can leverage the most of its potential compared to relational data relying on information provided by social networks. The steady growth of interest in these new analytical frontiers has not been properly accompanied by a solid reflection on the implications that the new data available and their analysis can generate (Amaturo & Aragona, 2016). Concretely, it means to take into account a critical discussion of the quality of data and processes that involves data collection, organization, analysis and ethical issues. First of all, there are issues of data access to the constraints of privacy. As a matter of fact, the materials in the net, which assume the status of personal published documents (Amaturo, Punziano, 2013), is produced and used with different purposes than the analytical one (as it generally happen for every secondary data). This goes on ethical and value questions that characterize the data. In fact, the data can be considered either as simple information freely available or as the result of a contextual construction hardly to be investigated outside the scope in which these was produced. The ethical boundaries are becoming increasingly blurred, as line between public and private also becomes blurred. Secondly, questions arise on the nature of the data and the ability to analyze them. The development of the net and mobile devices, with their social applications, are making obsolete

the usual ways of dealing with those data (Gantz & Reinsel, 2012). This fades the boundaries of time (synchronous/ asynchronous) and space intended as boundaries between real and virtual. Consequently new types of data emerge which are classified as big data (user-generated content, streaming digital data, log data and information generated from any communicative and net processes). The problems, along with that of ethics, consist in archiving, processing and analyzing these new data. Over the next years, this domain will leave space to much efforts in social research including new methodological approaches such as the Mixed Methods (cfr. among the others Cresswell, 2003; Tashakkori & Teddlie, 2010; Amaturo & Punziano, 2016) that seems to make increasingly blurred the boundaries between qualitative and quantitative research. These represent one of the methodological frontiers, and in this track this paper aims to problematize future challenges for the approach, whit a particular emphasis on the advent of big/new data. In fact, it is an approach that has stopped fighting for his emergence, and currently is fighting for its consolidation and its specialization, in particular by identifying those holes concerning the thematic areas and issues which is missing yet a consolidated discussion in the current literature on Mixed Methods (as well as on research in general) including the challenge that carries with it the advent of big data (Hesse-Biber & Johnson, 2013; Dumbill, 2013). References Amaturo, E. & Aragona, B. (2016). La “rivoluzione” dei nuovi dati: quale metodo per il futuro, quale futuro per il metodo?. In F. Corbisiero, E. Ruspini. Sociologia del futuro. Studiare la società del ventunesimo secolo. Milano: CEDAM, 25-50. Amaturo, E., & Punziano, G. (2016). I Mixed Methods nella ricerca sociale. Roma: Carocci Editore. Amaturo, E., & Punziano, G. (2013). Content Analysis: tra comunicazione e politica. Milano: Ledizioni. Cresswell, J.W. (2003). Research Design: Qualitative, Quantitative and Mixed Method Approaches. London: Sage. Dumbill, E. (2013). Making sense of big data. Big Data, 1(1), 1-2. Gantz, J., & Reinsel, D. (2012). The digital universe in 2020: Big data, bigger digital shadows, and biggest growth in the far east. IDC iView: IDC Analyze the future, 1-16. Hesse-Biber, S.N. & Griffin, A.J. (2013). Internet-mediated technologies and mixed methods research problems and prospects. Journal of Mixed Methods Research, 7(1), pp. 43-61.

Hesse-Biber, S.N. & Johnson, R.B. (2013). Coming at Things Differently Future Directions of Possible Engagement With Mixed Methods Research. Journal of Mixed Methods Research, 7(2), 103-9. Tashakkori, A., Teddlie, C. (2010). Handbook of Mixed Methods in Social and Behavioral Research. London: Sage.

Andersen, H. and Mayerl, J. (Technische Universität Kaiserslautern) Analyzing socially desirable response behavior using paradata The research addresses the difficulties involved with sensitive questions and perceived social desirability in survey research using paradata as a non-reactive method. Paradata can include response-latencies, response changes as well as patterns of item non-response. At a theoretical level, the research attempts to gain a better understanding of socially desirable response behavior. At a methodological level the attempt is made to identify and explain response effects by analyzing response behavior using paradata. The goal of the research is to examine respondents’ behavior when faced with socially desirable questions by analyzing what takes place between opening the survey web-page and logging in the final answer. Doing so provides insight into answers of first choice, ambivalence as well as retroactive changes to previous questions. This data can then be used to better model socially desirable answer behavior to improve survey measurement quality. The ongoing research involves modelling a dual process model of socially desirable respondent behavior involving the respondent’s motivation and opportunity to respond either spontaneously or through effortful reasoning while incorporating individual tendencies (“need for social approval”), the desirability of the topic itself (“trait desirability”), as well as social and contextual effects. While some research by other authors has suggested that the perceived social desirability of an answer should lead to longer response latencies, our preliminary findings show that sensitive questions illicit more decisive behavior from respondents (i.e. quicker response latencies and less changing of answers). This may show that socially desirable responses can be both automatic-spontaneous as well as controlled modes of behavior. Methodologically, the research focusses on questions involving personality traits relevant to the workplace; the assumption being that a self-assessment of the respondent’s suitability for their own field of work will be biased by social desirability. In order to establish the perceived level of social desirability for the personality trait-items, an external survey was conducted at the outset. In addition to the gathered paradata, we compare the self-reports of the main respondents with assessments by the respondents’ instructors/superiors. Thus it is possible to spot whether the

respondents perceived themselves more positively or negatively than their instructors/superiors and, more generally, whether certain questions illicit more disparity than others. Moderator variables such as the respondent’s need for social approval and the level of social desirability related to the individual items are also assessed in the questionnaire. To combine and analyze these datasets we use multi-level structural equation modelling. The research presented here should provide a deeper understanding of response behavior involving sensitive questions and contribute to the generation of methods for achieving better measurement validity, as well as higher levels of explanative performance in statistical modelling, without increasing the burden put on respondents using paradata as a non-reactive method.

Angel, S., Heuberger, R., and Lamei, N. (WU Vienna University of Economics and Business) How (and why) does the poverty headcount change if income data from registers are used? Evidence from the Austrian SILC In this paper, we take advantage of the fact that for the Austrian EU-SILC (European Community Statistics on Income and Living Conditions) of 2008 to 2011 both register- and questionnaire- based income data is available for the same observational units. We aim at providing explanations for changes in different aggregate poverty indicators by investigating the underlying changes in the distribution of household income as a consequence of using register data. First, it is asked which component (income type, weighting) contributes most to the change of the poverty headcount if register data is used instead of questionnaire data. Second, we look at overall and group-specific differences in measured household income and poverty headcounts between the two data sources. Third, by estimating multinomial logit and linear models with covariates referring to the interview situation (f.i. CATI vs. CAPI) and household characteristics we try to explain which house-holds tend to under-report respectively over-report their household income

The analysis proceeds in four steps. In the first step it is investigated in detail how the total and group-specific poverty rate and other income statistics alter if the data source for income compo-nents is changed. Second, separate multinomial logit models (with alternative-invariant repressors) at the household level are estimated in order to test which variables explain whether households over- or understate/-estimate their income. In a final step, negative and positive differences in household income are modelled separately, applying OLS with the same set of explanatory varia-bles as in step 2. The multivariate analysis also aims at investigating whether social desirability or cognitive error is more relevant for explaining differences in household income between question-naire and register data.

Results show that the at-risk-of-poverty rate in Austria increased by roughly two percentage points in each of the four years investigated. Social groups with a generally higher poverty risk also show a greater difference between register and questionnaire data. The at-risk-of-poverty

rate is based on the median of the total equivalised household income and thus dependent on the household income distribution (and of its components). This income distribution becomes more uneven when using register data. In general, small incomes become smaller and high incomes become higher. Consequently, the observed changes in the poverty rate are mainly driven by differences in em-ployment income (as compared to sampling weights and other income components). There is a higher number of households with a very low income as compared to questionnaire data. Taking together all regression model results for over-reporting and under-reporting several conclusions can be drawn. First, the magnitude of under-reporting rises with increasing income whereas an opposite correlation can be found for over-reporting. However, the latter relation is less robust over time (sometimes insignificant) and weaker in magnitude as compared to under-reporting. Second, variables related to the interview context do not seem to influence the likelihood of over-reporting variable at all. In contrast, CATI significantly reduces the likelihood and magnitude of error in the case of under-reporting in some years. Third, a generally higher model fit is observed for under-reporting.

Keywords: register data, household income, EU-SILC

Anguera, M.T. and Sánchez-Algarra, P. (University of Barcelona, Spain) Valorizing direct and indirect observational methodology: A challenge of Mixed Methods Data Analysis and Interpretation The growing body of scientific literature in psychological interventions shows the complementary nature of qualitative and quantitative data. The aim of this work is to highlight the enormous potential that direct and indirect observational methodology offers in this respect, and how both approaches produce qualitative data that can be quantified. Direct observational methodology preserves the naturalness of a situation or episode, guarantees perceptibility, permits a wide range of data recording formats, and is supported by an increasing number of specialized software programs. The organized processing of the array of information collected guarantees a high level of rigor in terms of the processing of quantitative data, whatever the application field is. Indirect observational methodology implies that, even several decades ago, researchers gathered and used textual information, such as life stories, self-reports, personal accounts, narrative registers or any other type of written document. However, in the considered purest areas of scientific psychology, this approach was given almost no opportunity to prosper. The explosion of information and communication technologies in recent decades has led to methodological innovations that offer new ways to quantify biopsychosocial aspects in everyday context. The conceptual framework is based on the work of several authors who have shown that quantification is an inherent part of observational methodology, as the second step of the process. Researchers have learnt to extract valid and consistent data from these kinds of documents, with a high degree of systematization. Also, there are many ways of processing all texts, ranging from narrative algebra or content analysis to the use of certain software for coding information or analyzing data. The conceptual framework is based on the work of several authors (Cook & Reichardt, 1982; Bakeman & Gottman, 1987; Sackett, 2000; Krippendorf, 2004; Anguera & Izquierdo, 2006; Sánchez-Algarra & Anguera, 2013) who have shown that quantification is an inherent part of observational methodology, as the first step of the process (recording of descriptive data, design

of non-standardized instruments, coding of data, and generation of parameters) is qualitative while the second step (data quality control and analysis) is quantitative. We add here that observational methodology allows the ‘qualitization’ of coded, quantitative data through their interpretation and their bearing on the new research process generated. Before transforming qualitative inputs from psychological interventions into quantitative data, it is first necessary to decide how to organize the heterogeneous information available. This process can be extremely complex as it is necessary to bring together data from very different sources, and very possibly, different points in time (Duran, McCarthy, Graesser & McNamara, 2007). The first step is to correctly record and code the data. As started by Bradley, Curry, and Devers (2007:1761), “coding provides the analyst with a formal system to organize the data, uncovering and documenting additional links within and between concepts and experiences described in the data.” In this presentation, we will illustrate how observational methodology produces qualitative data that can be quantified and then ‘re-qualitized’. Observational methodology could be a clear example of a mixed methods approach, as it involves gathering qualitative data that is then used to produce a matrix of codes that can be analyzed quantitatively (Sánchez-Algarra & Anguera, 2013). The processing of quantitative data in observational methodology differs in that it does not involve classical statistical analyses but rather focuses on lag-sequential analysis, detection of T-patterns, and vectorization of behavior via polar coordinates. It is thus methodologically feasible to transform the flow of several kind of episodes (communication between a psychologist and a patient, family discussion …) into a matrix of data. This is an important concept, as it means that we can process all potentially relevant information in a systematic, organized fashion, using a replicable procedure, and transform it into a format suited for quantitative analysis. In brief, the procedure used to collect and control the data fully meets the requirements of flexibility and objectivity.

Bakker, J. Statistics Netherlands Breakoffs in general population web surveys and its effect on data quality: a comparison between smartphone, tablet and desktop users. It frequently occurs that surveys are only partly completed, which could negatively influence the representativeness of the data. In this study, we analysed several general population sur-veys of Statistics Netherlands to determine where in the questionnaire respondents breakoff. Furthermore we explored several causes of breakoffs, like the type of device used (i.e. smartphones, tablets and desktops), completion time, type of questions asked and several background characteristics of the respondents. Our preliminary findings show that smartphone users broke off earlier and more often than desktop users. Several suggestions will be given on how to reduce breakoffs.

Barlas, F., Thomas, R., Fahimi, M. and Buttermore N. (GfK Custom Research) Can Survey Data be Too Clean?: Data Cleaning and Bias Reduction Many researchers have argued that, to improve accuracy, we should clean our data by excluding cases from analyses when participants have demonstrated sub-optimal behaviors, such as speeding, egregious non-differentiation on grid questions, or leaving many questions unanswered. Some researchers have gone so far as incorporating ‘red herring’ or ‘trap’ questions in their surveys in an attempt to catch such participants. Increasingly, some researchers are suggesting aggressive cleaning criteria that often identify a large portion of respondents for removal and replacement, sometimes up to 20 percent of participants. This not only raises questions about the validity of the survey results, but also has cost implications as replacement sample is often requested. While Thomas (2014) showed that eliminating small proportions of such respondents (less than 5%) did not significantly affect survey results, a question arises: How much data cleaning is too much? For this research, we analyzed data from three different surveys that contained items which allowed us to estimate bias, including items for which external benchmarks existed from reputable sample surveys, and with one study, actual election outcomes. Survey 1 included a probability-based sample of 1,297 participants from GfK’s KnowledgePanel and 2,563 participants from non-probability online samples. This survey included 36 questions with available external benchmarks to investigate bias. Survey 2, had over 1,860 participants from KnowledgePanel and 3,329 from non-probability online samples. This survey predicted election outcomes in the 2016 Florida Presidential Primary and incorporated 10 additional state-level benchmarks to investigate bias. Survey 3 was the Foundations of Quality (FoQ2) study sponsored by the Advertising Research Foundation, which included 57,104 participants from 17 different non-probability sample providers. This survey included 24 national benchmarks used to investigate bias. We examined how broadening the data cleaning criteria by excluding an increasing number of participants affected bias and external validity of survey estimates, including among key subgroups. Keywords: data cleaning, bias reduction, sub-optimal response

Barnat, M. and Bosse, E. (University of Hamburg, Germany) The challenge of creating meta-inferences: Combining data representing institutional and individual perspectives on first-year support in higher education Integrating qualitative and quantitative data during the course of a research project often involves challenges too specific to be addressed in handbooks. As an example for particular challenges and potential solutions, this paper draws on a study applying a “complex design with integration in several phases“(Kuckartz, 2014, p. 94). This study seeks to integrate multiple types of data as it involves different samples and methods to investigate the transition to higher education. Drawing on the first findings of a document analysis, expert interviews and a longitudinal student survey, this paper focuses on challenges regarding the interpretation and integration of our data. In particular, we aim to discuss the combination of the data representing different perspectives on first-year student support initiatives. The present paper is part of a larger research project that investigates the individual and institutional factors contributing to students’ transition to higher education. After exploring first- year challenges in terms of critical requirements students experience at the beginning of their studies in a pilot study (Bosse & Trautwein, 2014), we currently focus on the following question: (How) does institutional support seeking to address first-year challenges match individual students’ needs? Thus, we intend to combine insights into support initiatives with findings depicting the student perspective. Regarding the qualitative strand of our mixed-method approach, the initial phase of our study included data collection by means of a document analysis and semi-structured interviews. First we investigated the first-year support provided by German universities (n=80) using the qualitative document analysis to examine the design, aims and target groups of first-year interventions. The analysis included identifying the first-year challenges that the interventions seek to address. Second, we complemented the results with information from within the university context conducting interviews with experts at four HE institutions. Apart from the general university profile, the interviews focused on the implementation and integration of first-

year support into the university system and policy. Thus, the qualitative data provides detailed insights into the institutional context of the transition to higher education. A quantitative survey among students at the beginning of their studies (n=1.840) complemented the first phase of our study. Based on the category system developed in our qualitative document analysis, the students from the four selected HE institutions were asked for the first- year support they considered the most beneficial and to assess its effects. In addition, we inquired in what kind of first-year support they were interested in in the future. Finally, we also included items representing the previously identified critical requirements and asked the participants to assess how difficult they would find it to deal with the related challenges. The complex design of our study results in the methodological challenge to create meta- inferences integrating the different samples, methods and (the resulting) perspectives on first- year support. In the session we present the analytical grid we developed to integrate the different perspectives on first year support in order to discuss the legitimacy of conclusions. Keywords: Mixed-Methods, Meta-Inferences, Legitimacy

Baroni, W.S. (University of Manchester, England) A Genealogy of the Autobiography in Contemporary Society and Social Sciences. Autobiographical Discourse, Communism and Feminism The paper analyses two different channels through which autobiography has spread during the XXth century, becoming an integral part of the contemporary culture of the self and an extraordinary tool for social sciences. The mass political use of autobiography by the Communist movement and by feminism in the seventies – particularly in France and Italy – are key chapters in the history of its success. The importance of feminism in the promotion of self-narration, both as part of a new culture of the self and as an element of a renewed methodological sensitivity within social sciences, has been generally acknowledged. On the contrary, the communist use of autobiography is less well-known – it is only studied from a historical point of view, receiving little or no attention from social scientists. I will start my presentation by examining the connection between party, narration and the self, established by the communist movement, before addressing the differences between the communist and feminist autobiography. I will also use Italy as a case study. On the one hand, the communist and feminist cultures of self-narration – the former spanning from the end of the Second World War to the sixties, the latter taking place in the seventies – were strong and widespread in this country. On the other, both Italian communism and feminism are closely tied to their European counterparts – the communists are linked to the Soviet Union and the feminists to their French counterparts. As such, the Italian experience can be regarded as providing an insight into a broader, transnational phenomenon. The communist party autobiography materialized under Leninism and was later refined during the Stalin era. This practice formed part of the general requirement to engage in self-narration that characterised Russian society in the wake of the October Revolution. The population was required to produce narrative curriculum vitae and to answer personal questionnaires, both for surveillance reasons and also as a means of controlling their induction as Soviet Citizens. At the same time, from 1920 onwards, the Komintern – that coordinated the international communist movement from Moscow – directed Western parties to adopt the Bolshevik cadre party model,

which relied on order and the absolute self-discipline of members. To this end, the Communist International made it mandatory for members both to write their political autobiographies and to subject themselves to public sessions of criticism and self-criticism, to ensure their absolute transparency and adherence to the party. After the Second World War, this model was applied to its members by the Italian Communist Party. In Italy, the aforementioned practice went into gradual decline following the crisis of 1956 and during the sixties. At the end of that decade, however, the tradition was revived by Italian feminists, through the emergence of female consciousness-raising groups. Autobiographical practice was not restricted to oral accounts provided during face-to-face sessions; it also extended to the written form. Such writings appeared in journals published directly by feminist groups, along with diaries and autobiographies written by women belonging to the Italian movement. Communist and feminist experiences shed light on some very relevant characteristics of the contemporary culture of autobiography. Firstly, its association with the experience of confession. Autobiography is neither, primarily, a literary experience nor a document. From a Foucauldian perspective, it is essentially a technology of the self whose features can be understood only in connection with an institutional setting – a party or a small group, in our case. In such settings, the voice of the master, asking the individual to tell his/her truth, always resonates. From this point of view, the present success of autobiography should be problematized in light of the relationship between power, truth and narration of the self. On whose behalf does the subject tells his/her story when he recounts his/her biography? Secondly, communism and feminism represent two different ways of using the autobiographical mechanism. The former regards it, above all, as the party member’s curriculum vitae, anticipating the strategies of modern human resources departments. The latter, particularly in Italy, pushes the experience of self-reflection through self-writing to the point where the world vanishes and the subject remains alone with her body and its humours, prefiguring contemporary striptease culture. In short, Italian communist and feminist experiences of self-narration enable us to shed some light on the precarious positioning of the modern autobiography – between the need to control

the population and the postmodern obsession with the body as a surface upon which the history of the subject can be written.

Barth, A. and Schmitz, A. (University of Bonn, Department of Sociology) Towards a relational understanding of data quality In survey data, response styles such as acquiescence tendency, extreme responding or straightlining as well as other non-substantial patterns (item non-response, refusal, or socially desirable responding) are usually regarded as source of errors. The implication is that in order to ensure high data quality, such response patterns need to be corrected. From a cross-cultural perspective, different response styles, in contrast, are seen as societal emanations of respondents’ cognitive and personal traits, thus conveying important information on the respondent and his/her cultural and ideological background. However, this approach implies the reduction of particular societies to an artificial unity, thus ignoring variation within these societies (Bourdieu 1993). Altogether, both research streams tend to restrict themselves to isolated response styles as well as to singular explanatory variables. In our presentation, we suggest a relational understanding of data quality by way of (1) assessing relations between different response behaviours (2) within the structure of a single society. Using the example of the US, we examine data from the General Social Survey (GSS; 2010-2014) and demonstrate that differential response patterns are systematically related to respondents’ socio-structural position and their political-ideological dispositions. For this purpose, we assess the interrelations of multiple indicators of non-substantial responding (acquiescence, item nonresponse, midpoint-responding etc.) by applying finite mixture analysis. We then relate the resulting classes to spatial positions via geometric data analysis, a technique that allows to empirically construct the structure of a society and its sub-spheres such as the political field. Our findings imply that there are (1) different combinations of response patterns and thus different empirical types and (2) that these types systematically vary by the respondents’ (dis)positions. We conclude with a reflection on the widespread assumption that survey instruments are equally valid for all population strata. Reference: Bourdieu, Pierre (1993). Sociology in question. London: Sage Publications. Keywords: Data quality; geometric data analysis; finite mixture model

Beatty, P. and Lineback, J. F. (U.S. Census Bureau)

Developing a Program on Emerging Methods for Improving Survey Measurement at the U.S. Census Bureau Measurement error can be a significant threat to the validity of survey data. It is also difficult to address for many reasons: survey data collection relies upon imperfect processes of comprehension, recall, and judgment; minimizing one type of error can inadvertently increase another; and determining quality of measurement can be challenging in the absence of any particular "gold standard." A variety of methods have been available for developing and evaluating the quality of survey measurement, both qualitative (in-depth and cognitive interviews, focus groups, ethnography) and quantitative (experiments, latent class analysis, item response theory). However, methodologists continue to apply new data sources and analytic approaches to these problems. This paper outlines the rationale and initial activities of a new program at the U.S. Census Bureau designed to push the frontiers of these methods. Many approaches entail new analyses of existing data—for example, paradata (i.e. data about the data collection process itself) have been used to monitor field performance and applied to adaptive design, but may also be used to diagnose comprehension problems and difficulty navigating survey instruments. In addition, administrative data, social media data, and passively collected measures may important applications for evaluating survey measurement. Other approaches may focus on identifying generalizable insights from previous evaluations, and quantifying the statistical consequences of errors that have been identified through qualitative methods. The Census program in Emerging Methods complements other methodological units in questionnaire evaluation and usability research, in an effort to not only extend the efforts into new areas but also to build bridges between groups that have previously functioned independently. Keywords: survey measurement error, quantitative methods, qualitative methods, Paradata

Becker, J., and Witte, N. (University of Göttingen, Germany) Analysing places with reconstructive biographical methods In this presentation, we address questions of how to combine the analysis of places and biographical studies on the basis of our empirical research about processes of ‘emplacement’ in the context of a research project in Palestine/Israel. It is a common assumption that biographical research is the typical way of analysing time whereas the sociology of space is supposed to be concerned with spaces and places (Martina Löw, 2001: 10). However, challenging this separation of ‘time’ and ‘space’, it has been highlighted that the two categories cannot be viewed entirely independent of each other (e.g. by Doreen Massey [1995] or Edward Casey [1996]), but that spaces/places should also be analysed as processes. Furthermore, it has been argued that places are not only constituted by mobile or immobile things, but also by actors. This entails that there are no ‘places without people’ and that they cannot be analysed without taking into account the actors, their interactions in a place as well as their experiences, memories and narrations about it. At the same time, people also belong to and are determined by places: “we are always already in a place, never not emplaced in one way or another” (Casey 1996: 17). Thus, when taking into account that places and actors constitute each other mutually and that places (also) have a processual character, biographical studies become one avenue to analyse places and their changes. Reconstructing lived lives and life-stories enables not only empirical results about ‘emplacements’ or ‘localisations’, but also about the corresponding places in which people localise or ‘emplace’ themselves (or are emplaced). The relevance of such an analysis of places becomes obvious in our empirical research in city quarters of Haifa and Jerusalem where we have conducted biographical-narrative interviews and participant observations. The Old City of Jerusalem, for example, is a dense ideological and geographical space in which the belonging to certain place-based communities and perceived safe retreats e.g. in family houses gains a special importance. We will highlight these issues by discussing our small-scale study in a Palestinian Old City neighbourhood. Introducing the biographical case reconstruction of the interview with Hafez, a middle-aged Palestinian in the

Old City, we will show how people experience relevant places from their primary socialisation onwards and how relevant places in a biography might shift from the neighbourhood to the family house and also to a ‘longing to be somewhere else’. Such an analysis has, however, only proven fruitful if it is closely linked to the analysis of changing societal power relations and the socio-economic situation on the ground. Reference: Casey, Edward S. (1996): How to get from space to place in a fairly short stretch of time. Phenomenological prolegomena. In: Steven Feld, Keith Basso (eds.): Senses of place. Santa Fe, 13-52. Löw, Martina (2001): Raumsoziologie. Frankfurt a.M. Massey, Doreen (1995): Places and their pasts. In History Workshop Journal 39(1), 182-192.

Keywords: spatial analysis, biographical research, urban sociology

Bednarowska, Z. (Department of Sociology of Economy, Education, and Social Research Methods, Institute of Sociology, Jagiellonian University in Krakow, Poland) Why sociologists should follow political scientists and introduce spatial econometrics to their research? INTRODUCTION

Spatial data interest is growing rapidly. Geomapping, geomarketing, geospatially oriented social media, geotagging, development of Volunteered Geographic Information are called „the geospatial revolution” (Robinson et al. 2015). “Mapped” data is more and more popular among data analysts, market research executives, probably due to the fact that maps’ visual aspects are more appealing when presenting data. However, oftentimes it is limited to applying space in a basic exploratory spatial data analysis.

SPATIAL ECONOMETRICS

Political scientists, economists, and geographers go further than exploratory spatial analysis and introduce spatial econometrics (statistics) (Ward & Gleditsch 2008). Here space serve as a part of regression equation, an independent variable that may contribute to higher explained variance. There are many topics already covered with spatial statistics, just to mention spatial relation between democracy and GDP or turnouts (Ward & Gleditsch 2008), World War II alliances (Franzese, Hays, & Kachi, 2012).

RESEARCH PROBLEM

My research project on consumption space will only serve as an example to introduce and discuss the method.

One can observe a paradox of an increasing number of shopping centres and anchor stores, which are neighbours of dead malls. This impacts social and economic transformation of urban and suburban areas. The research aim is to identify main spatial factors influencing development of these two contrary processes. The geographical scope of the project focuses on Poland. However, the research problem is explained in the international comparative context – due to

analysis of the international data on shopping malls and research conducted in USA on dead malls. In order to explain research problem consumer behaviour theories, Lefebvre’s social production of urban space, as well as mobility paradigm (Urry 2007) were applied.

METHODOLOGY

The research design follows mixed-methods, including secondary data analysis, non-participant observations, based-on-regression case study and qualitative interviews with consumers. However, the major method applied in the project is spatial econometrics model, where a list of the economic indicators was enriched by the spatial autocorrelation in order to see if there is a relation between increasing consumption space, economic condition and spatial proximity (Bednarowska 2015).

CONCLUSIONS AND OPEN PROBLEMS

As a follow up to my work with a spatial regression model of retail chains in Poland, I will present key aspects of application of spatial econometrics in sociological research:

• examples of the sociological theoretical problems that are best suited for such analysis

• spatial data collection challenges

• requirements regarding type of data that represent space

• possible types of models introduction (SLDV, SEM)

• data preparation, mapping data, data conversion

• spatial data analysis process overview using R and STATA

• a short overview of the software available for spatial analysis for social scientists

• possible data visualisation solutions for spatial data results

• a list of challenges and limitations for sociologists, as well as possible solutions.

The presentation aims at proving that an implementation of spatial econometrics has great potential in sociological research. However, it is a time-consuming and demanding type of

analysis, therefore ideas how we can work on improving it, will serve as an opening for discussion.

KEY REFERENCES Bednarowska Z. (2015). A spatial regression model of retail chains development in Poland. Ekonometria (Econometrics), 3(49). Franzese, R.J., Hays, J.C. & Kachi, A., 2012. Modeling history dependence in network-behavior coevolution. Political Analysis, 20(2), pp.175–190. Robinson, A.C. et al., 2015. Maps and the geospatial revolution: teaching a massive open online course (MOOC) in geography. Journal of Geography in Higher Education, (April), pp.1–18. Ward, M.D. & Gleditsch, K.S., 2008. Spatial Regression Models, Thousand Oaks, CA: Sage. Urry, J. (2007). Mobilities. Cambridge: Polity.

Keywords: Spatial Analysis, spatial econometrics

Belli, S. (Yachay Tech, Equador) Manage Social Conflict for a Slow Internet Connection in Online Meetings In online meetings often Internet connection is not fast enough, and users experiment critical situations. The purpose of this paper is to identify how participants manage this situation during online meetings. We analyze the participation framework in a corpus composed of 30 hours of online meetings between students of an Andean university, their teacher and international experts. In these online meetings, students present their projects based on Innovation and Entrepreneurship to specialists in this field headquartered in a UK university. The internet-based IT platform used was ZOOM platform.

Emotions regulation between users are social constructed and managed to improve the effectiveness of the online meeting. We present a multimodal interaction (Goodwin, 1981; Stivers, Sidnell, 2005; Becvar, Hollan, Hutchins, 2008) of verbal and body language in collaborative activity following the Chovil & Bavelas-approach (1997) to the analysis of moment- by-moment evolving social interaction. Also using conversation analysis (Heritage, Perakula, 2004; Rossano, 2012; Tiitinen and Ruusuvuori, 2014), we focus on the ways in which participants interact with their words and their non-lexical expression (gaze, gestures, prosody). Thanks to this methodology, we describe the moment-by- moment interactional work performed in collaborative activity (Perakyla & Ruusuvuori, 2006; Goodwin, 1981, 2007).

We have observed in this study, how technical difficulties, like a slow internet connection, generate social unrest and negative emotions shared between participants. In many cases, these difficulties produce conflicts among participants. We describe how negative emotions are shown in mixed contexts, and how users solved these during an online meeting.

Bengali, M. (Texas A&M University, Qatar) Seeking God's Assistance to Govern: A Comparative analysis of Islamization Policies in Pakistan and Egypt The uneasy relationship between religion and politics is a dominant feature of many post colonial Muslim majority states. Two most prominent cases in this regard are Egypt and Pakistan. This paper does a comparative analysis of the two countries under the regimes of Anwer Sadat in Egypt (1970 - 1981) and Zia-ul-Haq in Pakistan (1978 - 1988). The analysis takes into account the changed role and status of the state in context of public and private life, construction of the new and modification of existing institutional structures to implement Islamic visions of the two leaders. It is noteworthy that even tough both leaders tried to Islamize the polity the trajectory followed was radically different. The paper argues that this difference can be attributed to the lack of a universal understanding of the notion of an Islamic State. In most cases, Islam is used as a tool to gain legitimacy by the ruling regime. The conclusion emphasizes the current situation in both countries.

Bergmann, M., Hanemann, F. and Scherpenzeel, A. (Chair for the Economics of Aging), Munich Center for the Economics of Aging (MEA) A responsive monitoring design using paradata to improve data collection in the Survey of Health, Ageing and Retirement in Europe (SHARE) The Survey of Health, Ageing and Retirement in Europe (SHARE) is a multidisciplinary and cross- national face-to face panel study of the process of population ageing. It studies the different ways in which people aged 50 and older live in 20 European countries from Sweden to Greece and Portugal to Estonia. Data are collected every two years using a harmonized core questionnaire in all countries as well as objective health measures, such as hand grip strength and dried blot spots.

For the sixth wave of data collection, a responsive fieldwork design was implemented in the German sub-study of SHARE. The aims of this design were, firstly, to improve the overall response/retention rate in the German panel that was considerable lower than in other countries and, secondly, to decrease nonresponse bias – both with the overarching goal of improving data quality in mind. Recently, responsive designs have been given a lot of attention in the survey methodology literature.

These designs often make use of background information (paradata), which is captured during the data collection process, to more efficiently allocate fieldwork resources to specific sample units. SHARE is especially suitable for such a strategy, because it already conducts a high level of fieldwork monitoring, has in place an advanced system across countries to register fieldwork results, and possesses extensive background information about the interviewer performance as well as the panel members and their response behavior in previous waves.

Against this background, we implemented a comprehensive responsive monitoring design for the German sub-study, which uses this information in order to improve the survey data collection process. At the same time, it served as a “dashboard” of response probabilities during fieldwork associated with measures of relevant respondent characteristics and behaviors. This allowed for immediate feedback to the survey agency and focused actions with regard to specific groups of respondents. We will present the responsive design that was used focusing on the

fieldwork interventions, which resulted from the monitoring, and the results on the German response/retention rates.

Furthermore, we evaluate the effectiveness of such a design with respect to data quality as well as nonresponse bias for different groups of panel members and give practical advises for researchers in directing their efforts by using this kind of additional information.

Bergner, B. and Spellerberg, A. (University of Kaiserslautern, Germany) Monitoring Real-Time Stress Experiences in Space – Data Collection, Operability and Methodical Challenges of User-Generated Emotion Data research? Open air events like festivals, sport events or concerts are regarded as essential for liveable communities. In terms of collective experience, events are characterised as social, non-routine, emotional and singular phenomena. They allow the experience of community, interaction and corporate identity for people with different backgrounds.

Our research focuses on well-being and emotions of event attendees at open air festivals, primarily situations, in which people feel stressed or unsafe. These emotions are important as they influence social behaviour. Positive emotions may lead to positive mass behaviour. But especially in risky situations, stress can lead to avoidance behaviour or aggression and thus endanger the course of an event. Typically, unsafe areas are identified by means of classical empirical methods (quantitative and qualitative interviews). All these methods are based on cognitive self-perception, experiences and social desirability. To overcome these influences, nowadays it makes sense to measure and track physiological vital data in real-time. This is possible with the help of innovative wearable devices combined with GPS sensors for spatial localisations. Social scientists are able to get objective information, which can be added to verbal statements on emotions and affections.

With the rise of wearable sensor technologies the human being itself becomes an emotion sensor. It is questionable to what extent this data can be used for identification of spatial deficits, group processes or external effects. For decades, ambulatory assessment methods have been used to identify humans’ emotions linked to distinctive stimuli in laboratories under controlled conditions. Knowledge transfer to real-world applications is still rare due to many uncontrollable environmental issues. Psychophysiological monitoring, however, shows high potential of transferability to real-world studies.

In order to measure well-being in our studies, classical methods like standardised questionnaires (n=746), observations and interviews (n=20) are conducted, but complemented by monitoring

real- time stress experiences. Here, a sensor wristband was used, which allows to collect location- independently peripherally-physiological data as indicators for emotions/stress in real- time. The city festival “Altstadtfest 2014” in Kaiserslautern and the “Back to the Woods Open Air 2014 & 2015” in Munich/Garching are the chosen events with a total number of 25 study participants equipped with sensor wristbands.

The usage of user-generated emotion data in specific socio-spatial analyses still have to meet many challenges, concerning data acquisition, operability, validation and visualisation. In order to identify reasons for stress at this state of our research, further data sources were taken into account. For example, standardized questionnaires before and after visiting the event give insight in the study participants’ personality, actual state of anxiety/fear, event affinity, perceived sense of safety and socio-demographic data. Our mixed-method approach combines classic empirical methods, psychophysiological monitoring, spatial analyses as well as GIS-based visualization techniques.

Results: There is a significant connection between the three different events and the standardised stress values of the participants. With the help of density and sectoral spatial maps of stress events, specific neuralgic areas can be identified. They were validated by qualitative interviews with event organisers and security organisations and by standardised questionnaires with event attendees.

Keywords: Spatial Analysis, Emotion and Stress Detection, Event Monitoring

Bertolini, S. and Musumeci, R. (University of Turin, Italy) Changing preferences on work in the transition to parenthood? Evidence from qualitative longitudinal research on Italian couples This paper investigates first-time fathers’ and mothers’ preferences and decisions on work and the participation in the labor market during the transition to parenthood by analyzing interview data on 22 Italian couples. Each partner of each couple had been interviewed separately before and after the born of the first child, for a total of 88 interviews. The qualitative longitudinal analysis of all these interviews allow us to observe the changes in expectations and behaviors in different phases of the life course (before and after the baby’s birth) within the couple comparing him and her.

The advantages of this methodological approach is to admit individuals that are dynamic and interactive, having preferences that can change over time, with regard to structural, institutional and economic constraints, but also in relation to the different phases of the life course and in the interaction with partners.

This approach calls into question economic and sociological theories on labour market assuming that individuals’ preferences are stable such as human capital theory (Becker 1964), which assume for example that the level of education structures the preferences, or Hakim (2000) theory, or sociological theories emphasizing the effect of the education in terms of emancipation (Reyneri 2011). Both perspectives do not explain the changes in preferences along the life course, but presuppose a static actor over time. In our opinion, the birth of the first child may be a turning point around which individuals redefine priorities, preferences and roles regarding participation and commitment in the labor market. First time fathers’ and mothers’ discourses are analyzed both at a descriptive level and at a socio-cognitive level; to this aim we refer to the cognitive dissonance theory (Festinger 1957; Schober and Scott 2012; O'Reilly et al. 2014) and to the adaptive preferences (Elster 1989), which can lead individuals to change their opinions to reduce the conflict between attitudes and behaviors. The four interviews per couple allowed us to reconstruct attitudes, preferences and behaviors on work in three stages of mothers’ and fathers’ life course: 1) pre-pregnancy period, which is reconstructed retrospectively during the antenatal interview, 2) pregnancy, 3) and finally when the son/daughter is about a year and a half. We have analysed the interviews by the software Atlas.ti7, using codes on nine thematic areas created intersubjectively by a multidisciplinary research group. We focused our attention to the first time parents’ narratives related to the previous working career, current job and orientation towards work. We filtered these codes for the following families of interviews: wave of 61

interviews (antenatal and post-natal), gender, and type of employment contract (self-employed, permanent or temporary).

This in order to evaluate the existence of specificities and/or similarities between groups of employees in different work situations that for Italian case means also to have different rights and possibilities to use parental leave. In addition to the queries so constructed, we also examined in their entirety the 88 interviews by 22 longitudinal synopses, one for each of the couples interviewed.

Keywords: parenthood, work, transitions

Beullens, K. (Centre for Sociological Research - KU Leuven) Interviewer effects among the elderly In the European Social Survey, we observe strong indications of interviewer effects. In some countries, the proportion of variance attributable to interviewers (ICC) may amount to 20%.

One of the possible explanations for interviewer effects in surveys is that interviewers play a more dominant role than they are expected to (e.g. asking questions in a suggestive way) or induce or provoke satisficing behavior from the respondents, such as straightlining.

We expect that elderly people (as well as the lower educated) are less prone to adequately play their role as a respondent and therefore leave the interviewers to play a more prominent role during the interview. This might become more obvious when respondents tend to ask for more clarification or when respondents give the impression not entirely to understand the questions.

Therefore, we will assess in this paper whether interviewer effects among the elderly are greater than those measured among younger respondents. We also seek to relate this to a short questionnaire interviewers fill out at the end of the interview about how well the respondents play their role.

Keywords: European Social Survey, interviewer effects, elderly

Bhattacharyya, A. (Vidyasagar University, West Bengal, India) A methodological quest set forth for Indian software Industry Post-globalised society is synonymous to all invasive spell of ‘Information Technology Industry’. The new industry has aggressively replaced the existing conventional notion of nature of work and workplace and keeping the door ajar for a lot of women to work in this sector. This sector has immense potential for emerging researchable areas. Its restrictive and prohibitive policies debarring entry for the outsiders/researchers to its workspaces for interaction with its substantive employee-respondents for necessary inputs from the field by applying probabilistic sampling procedures turns out to be a difficult task for unbiased research. Most studies, therefore, suggest for non-probabilistic way of data collection with fewer sample base. This study delves into this sector by evolving the method so as to garner maximum sample base of about 250 respondents by inducing probabilistic approach in non-probabilistic procedure of sample selection. Preceding this, the present study looks into the methodological standpoints to understand gender and work perspectives of the said industry. It explores the contemporary feminist methodological debates on espousing articulation of women’s voices and their experiences vis-a vis western epistemology legitimising male monopoly. Feminist methodological juxtaposition is meant to focus beyond women’s productive role i.e. Women in Development (WID) with that of deconstructing male biases operating at the workplace by Gender and Development (GAD) analysis. Further, women’s engagement to Information Communication Technology (ICT) led to development approaches like ‘Gender Evaluation Methodology (GEM)’, or the other approaches such as Gender in IT (GIT) and Gender and IT (GAIT). The breakthrough of these approaches unveil a convergent methodological middle ground by bridging the gap between agency vs structure, representativeness vs generalisability, contextual vs universalism, macro vs micro realities, positivistic vs interpretative methodologies and qualitative vs qualitative methods. Ultimately, we opt for the ‘mixed-methods’ of triangulating both the quantitative and qualitative data in its applicability.

Keywords: Women studies, Gender, Mixed method

Blasius, J., Thiessen, V. and Nenadic, O. (University of Bonn, Germany; Dalhousie University, Halifax, Canada; University of Gottingen, Germany) The dirty data index – assessing the quality of survey data in international comparison Responses to survey items are associated not only with socio-demographic characteristics such as age, gender, and educational level, but also with various sources of impurities. These include, among others, response styles, such as acquiescence response style, extreme response style, and midpoint responding, cumbersome wording of questions, arbitrary responses, and respondent fatigue, all of which reduce the quality of data. In general, when analysing a battery of items, responses are related to the substantive concept, which we are mainly interested in, and to methodologically-induced effects. The higher the proportion of methodological effects is in a given data set, the dirtier are these data. Applying categorical principal component analysis (CatPCA) to an item battery of survey data allows us to explore what part of the responses is due to substantive relationships and what part is attributable to methodological artefacts. As is true for PCA, CatPCA quantification values are standardized to a mean of zero and standard deviation of one. The closer the data are to being metric, the smaller are the differences between PCA and CatPCA, and the closer the distribution of the CatPCA quantification values is to the standard normal distribution. Using these features we created an index measuring the quality of a given item battery that is based on the quantification values, an index we call the dirty data index (DDI). In the presentation we will briefly explain how to construct the index and we will demonstrate the usefulness of the DDI for comparing the quality of survey data in countries participating in the European Social Survey. As we could show with previous research (Blasius and Thiessen 2012) using other data and other methods, some countries provide high quality data, some with medium quality and others with low quality.

Reference: Blasius, Jörg and Victor Thiessen, 2012: Assessing the quality of survey data. London: Sage.

Keywords: categorical principal component analysis, quality of data, ordered categorical data, dirty data

Brinkgreve, C. and van Daalen, R. A Sociological reading of Diary Fragments during wartime. During wartime people have to adapt to new situations. They see themselves confronted with new dilemma’s, they have to make new choices and decisions, in a context of changed power relations. How do people deal with these new experiences, how do they behave and what do they tell us about their feelings and behaviour?

We did research on these issues by studying diaries of people from different social layers and positions, written in the Netherlands during the Second World War, and stored in the Netherlands Institute of War Documentation (NIOD). We restricted our research mainly to the selection of the diary fragments as published in Dagboekfragmenten 1940-1945, with a focus on the early years.

We read these fragments in a sociological way. That means that we used sociological sensitizing concepts as 'situations', 'interactions', 'identification', 'disidentification', ‘compartmentalization’, ‘exclusion', 'orientation', 'confusion'. We tried to connect the inner world of the writers with the context of their networks and the situations in which they are involved. In this way we tried to link the micro- and the macroworld.

In this paper we want to make a comparison between our sociological perspective and other perspectives. When reading the diaries, a psychoanalyst will use other concepts, give other accents and other interpretations, and the same will be the case when an historian looks at these fragments. How do these approaches differ, what is the value of the different perspectives? What do these different perspectives see and neglect, what do they keep on the background, what remains seen but unnoticed? And can these different ways of looking and conceptualizing fruitfully be integrated?

Bolli, T., Egg, M. E. and Rageth, L. (ETH Zurich, KOF Swiss Economic Institute, Switzerland) Does Vocational Education and Training (VET) Improve the Youth Labour Market Performance? Vocational Education and Training, Apprenticeship, Youth Labour Market Due to spectacularly high youth unemployment rates in countries strongly affected by the 2008 financial crisis, the media has drawn great attention to the labour market performance of young people. However, not all countries experienced a decline in their youth labour market performance. The question is why. Education is responsible for providing young people with skills that are needed in the labour market. Theory reasons that VET matches employees and employers better than general education as it teaches more vocational skills and prepares for specific occupations. In addition, they argue that combined school and work-based (dual) VET is more successful than school- based VET as the combination of the different learning and training locations ensures the labour market orientation of the acquired skills. Hence, we analyse whether VET affects the labour market performance of young people. Using a rich panel data set of over 40 countries for the years 2001 to 2014, we investigate the impact of enrolment rates in upper secondary education programmes (general education, school-based VET, dual VET) on the labour market performance of the 15 to 24-year-olds. We thereby depart from prior work by three contributions. First, we apply a comprehensive set of outcome indicators by including not only the integration of young people in the labour market (e.g. youth unemployment rate) but also the quality of their jobs (e.g. involuntary part-time rate). Second, we are the first ones looking at the impact of school- based VET relative to general education on the youth labour market performance. Third, we address unobserved heterogeneity by not sticking to pooled OLS but exploiting the time variation of our data. Our main specification is a fixed-effects model with country-specific intercepts to account for the heterogeneity across countries. However, we additionally run random-effects regressions to compare our results with previous literature. To further check for our results’ robustness, we apply the generalized method of moments (GMM) which accounts for the dependent variable being dynamic and for independent variables not being strictly exogenous. In addition, we lag our explanatory variables by three years, which is the average

duration of these education programmes. With that, we account for the expected impact of enrolment patterns taking place after its completion. Moreover, the lagging of the enrolment rates and the application of GMM helps us to address reversed causality. Our results show a positive impact of dual VET enrolment rates on youth labour market integration, which is in line with previous research. In addition, we find no disadvantage of dual VET regarding job quality. For school-based VET, our results reveal no impact on labour market integration but a negative effect of it on job quality. As the coefficients of all three estimation models are similar with regard to their direction, size, and significance, we conclude that our findings are robust. In addition, a t-test shows that the impact of school-based VET and dual VET significantly differ for labour market integration. Hence, the two programmes are not equally appropriate in improving the youth labour market performance.

Keywords: Vocational Education and Training, Apprenticeship, Youth Labour Market

Borgs, C., and Schnell, R. (City University, London) Linking against all odds: Using Administrative data with modern Privacy Preserving Record Linkage Techniques The use of administrative data is often hampered by the existence of bureaucratic silos within companies and governments. One main argument against record linkage across organisations used by data hosting agencies are privacy regulations. Erroneously, it is widely assumed that record linkage is permitted by law only if persons explicitly permit the use of administrative data for research purposes. In many countries (including even Germany) this is not true. Most countries permit the use of administrative data if public interest outweighs personal privacy considerations. While balancing these conflicting interests, data protecting agencies are more inclined to support research, if the risk of re-identification of research subjects is very low. If a unique person identification number (PID) is available, encrypting this PID for a given research purpose with a project specific encryption is widely considered as a secure method for linking administrative data. Some countries like France rely on this linking method. However, if such a PID is not available as in most other European countries, linking of administrative databases requires the use of encrypted personal identifiers such as names, date of birth or place of birth. In practice, these identifiers show error rates up to 20% per identifier, therefore linking on encrypted identifiers usually implies the loss of large subsets of the databases. In many applications, this loss of cases is related to variables of interest for the subject matter of the study. Therefore, this kind of record-linkage will generate biased estimates. These problems gave rise to a new subfield within Computer Science: Privacy Preserving Record Linkage (PPRL). These developments have hardly been noticed in social research. Therefore, the presentation will give a short non-technical introduction to PPRL methods. Many different PPRL techniques have been suggested within the last 10 years, very few of them are suitable for practical applications with large database containing millions of records as they are typical for administrative databases. The presentation will focus on technical solutions for PPRL, which have proven their usability for large scale datasets. Pre-processing of identifiers,

encryption techniques, record-linkage options and the handling of very large datasets will be demonstrated. Of course, security issues of the procedures are discussed in a non-technical way. Finally, the application of PPRL for national databases on births and deaths and a national economic elite study will be illustrated.

Brandes, U. and Schoch, D. (University of Konstanz) When the centrality of actors is most obvious, centrality indices explain nothing at all

Determining the relative centrality of actors, or the degree to which they are structurally important, is a most common technique in social network analysis. Many indices have been proposed to measure a variety of centrality conceptions, and choosing one that is most appropriate for the particular research question and data at hand proves a challenge in many empirical studies.

Although each centrality index represents a particular interpretation of centrality, and in very precise quantitative terms, their selection is usually justified on a substantially more informal level. This vagueness in justification may be one reason why studies regularly determine and compare the results of several centrality indices on the same network.

If multiple centrality indices rank the nodes of the same network similarly, the joint ranking is interpreted as a robust finding or as an indication that the indices measure similar notions of centrality. If, in turn, rankings disagree, the indices said to measure different aspects of centrality.

We recently showed that all common centrality indices preserve neighborhood inclusion, i.e., they rank an actor at least as high as another if the former has ties to the same and maybe other actors as the latter. As a consequence, it can be shown that there is an infinite family of networks on which all these indices necessarily produce the same centrality ranking. Whatever the theoretical interpretation associated with a centrality index, it thus appears to provide a fitting explanation. The more clearly the structure of a network determines the centrality ranking independent from the particular index, there more competing explanations there are.

This poses a major methodological challenge: How do we determine whether the ranking obtained from a centrality index can actually be attributed to the interpretation associated with the index?

Our proposed solution is to avoid quantitative indices as long as possible and start from discrete comparisons of relative centrality instead. The approach is based on recently introduced concepts of network positions and positional dominance, and facilitates the application of successively more specific explanations until a ranking is sufficiently complete to test a hypothesized centrality effect. In addition to more direct theorizing, this also provides means to include multiple relations and attributes in the definition of centrality.

Keywords: social networks; centrality; core-periphery structure

Budde, E., Knill, C. and Preidel, C (Ludwig Maximilians University of Munich, Department of Political Science, Germany) A Matter of Timing: Religion and the Governance of Moral Issues The question of whether or not religion accounts for variance in the governance of moral issues between and within countries over time has long been debated but never conclusively answered. A novel dataset encompassing innovative measurements of state regulation of “life and death” issues and of the religious stratification of society enables us to conclude this debate and answer why previous studies reached contradictory results. The time-series cross- sectional analysis of 26 countries over 50 years reveals that the stance of the dominant religious denominations in society on abortion and euthanasia indeed influences state governance approaches regarding those issues. This denominational effect is shown to be contingent upon the religiosity of a country`s population, but independent from the formal state-church relationship. Lastly, it is shown that the religious effect has an inverse U-shaped relationship with time, exposing the time frame of analysis as decisive for inferences drawn in the study of morality policy.

Burzan, N. and Eickelmann, J. (Technical University Dortmund, Institute of Sociology, Germany) Event-oriented processes in museums: challenges of a multi-methodical approach It’s remarkable that (hand)books dealing with mixed methods often emphasize their practical challenges, for example time, money and other resources or chances for publications and funding (e.g. Creswell/Plano Clark 2011). However, there are many challenges concerning the comparability of results and other problems in order to the integration of methods/data, and there is in fact the risk of combining the weak instead of the strong points of different methods.

We will present some of those challenges and also some solutions with regard to a specific research project (funded by the German Research Foundation, 2014-2017) that analyzes event- oriented processes and their effects in museums. We ask for example, if and how museum’s orientation to entertainment and emotionality has increased and if those processes have an effect to distinctive options and behavior of museum’s publics.

We use different qualitative and quantitative methods: non-standardized observations of expositions and visitors, guided interviews, standardized observations and visitor surveys. Thus we analyze the interplay of providing and acquiring culture in different genres of museums. Detailed observation notes are conducted in several museums in order to gain a widespread perspective. In some selected museums in Germany we use all listed methods (partly in a sequential, partly in a parallel way) – thus it is one of the advantages of our design that we can refer data and results to one museum in each case and – in a second step – by comparing different museums. Our objectives of the mixed methods design are: 1) a reflection of the researcher’s role and 2) a more complex coverage of the research object by relating perspectives to each other. However, we do not require to validate our results automatically.

In our presentation we will show some concrete challenges and strategies of solution:

 advantages and disadvantages of a (partly) sequential/circular research practice of qualitative and quantitative methods: amongst others we test special hypotheses by quantitative methods which are explored by qualitative methods before.

 integration of (qualitative) observation and guided interview data (with museum’s authorities): while there is need to reflect researcher’s presumptions with respect to the observation data, we also have to consider that museum’s representatives provide a ‘proscenia’ presentation of their exposition concept.  integration of different standardized data (observation and interviews): it is a challenge to combine reliable observation data (with many situational effects) with interviews that are preferably not characterized by social desirability.  integration of qualitative and quantitative methods: we will present examples for options and limits of comparability, e.g. qualitative observations can motivate a standardized observation, but the standardized observation cannot consider the totality of relevant situational factors and thus cannot falsify the qualitative observation ultimately. At least some apparent ‘contradictions’ can be resolved in this way.  finally: options to analyze ‘distinction’ – even by a mixed methods approach In all these cases we exemplify our methodical concept of ‘looking through one another’ in order to optimize the chances to integrate different methods without neglecting their challenges and limits.

Keywords: Mixed Methods; Museum; Event

Calderwood, L., Gaia, A., Wood, M. and Alvarez, P. C. (Centre for Longitudinal Studies, NatCen) Using para-data for real-time trade-offs between fieldwork cost, response bias and response rates: experiences from Next Steps Responsive or adaptive designs involve making adjustments to survey design protocols during data collection in order to actively manage in real-time the trade-off between fieldwork costs, response bias and response rates (Groves and Heeringa, 2006; Couper and Wagner, 2011). These approaches are facilitated by technological developments such as electronic sample management systems allowing the real-time collection and analysis of para-data. Schouten et al. (2011) recommend the use of the ‘R’ indicator to evaluate representivity in real-time. Longitudinal studies are generally better-able than cross-sectional surveys to evaluate response bias as they are able to use prior wave data.

Although there has been a growing interest in this area, relatively few large-scale surveys have implemented these approaches. This paper will report experiences from the eighth wave of Next Steps (previously known as the Longitudinal Study of Young People in England) which implemented a responsive design approach to managing the trade-off between fieldwork cost, response bias and response rate using ‘R’ indicators.

Next Steps is a cohort study following around 16,000 young people in England born in 1989- 1990. The sample was recruited in 2004 when the study members were aged 13/14. By wave 7, the achieved sample size had fallen to around 8,600, and this attrition was strongly related to baseline socio-demographic characteristics. The study was funded and managed by the Department of Education until 2013 when the Centre for Longitudinal Studies secured funding to re-start the study and conduct another wave of the survey in 2015/2016 (Wave 8), for which all participants who have ever taken part will be approached again. The survey is being carried out by National Centre for Social Research (NatCen).

The survey design for Wave 8 consists of a sequential mixed mode approach involving web, telephone and face-to-face. Due to length of time, and variability in time, since the study members were last interviewed there is considerable uncertainty about what the response rates will be, both overall and to each of the different survey modes, and therefore what the cost of the fieldwork will be. In order to help manage this uncertainty, we are conducting a ‘soft-launch’ of the survey involving a random sub-sample of 2,000 cases which will provide more reliable estimates of response rates and fieldwork costs. Electronic real-time capture of para-data on fieldwork effort, such as contact and tracing attempts, will also allow us to evaluate the impact of enhanced fieldwork effort, particularly re-issuing and ‘enhanced tracing’ on response rates and response bias. Overall, our aim is to maximise the overall response rate, and hence the achieved sample size, and minimise non-response bias, within the context of a fixed fieldwork budget. We estimate response models in order to compute a response probability for each participant and use R-indictors to monitor the representativeness of the achieved sample at key points during fieldwork and to evaluate the impact of fieldwork effort on sample representivity. In the event that it is not possible to issue all of the cases to face-to-face fieldwork due to the budget constraint, we will use the R-indictors to help prioritise cases to be issued.

Campagnolo, G. M., Williams, R., Alex, B. and Acerbi, A. (University of Edinburgh, Eindhoven University of Technology) Social data science and the sociology of professions: reflections on methodological synthesis This paper exemplifies how existing insights from empirical sociology can contribute to computational approaches for the analysis of social media data. By addressing the case of influencer detection, this study demonstrates how ideas from the sociology of expertise (Turner, 2001) and the sociology of professions (Abbott, 1988; 1990) can help offer directions towards a more adequate approach to influencer identification. First, the paper will offer a sociologically informed criticism to current computational approaches aimed at detecting influencers by measuring influence. Considering sociological insights suggesting that identification based on career data. This paper presents early results from sequential analysis of a sub-sample of analysts’ population, including those reported in the database as the 100 most influential analysts. In conclusion, discussed will be lesson learnt concerning role of methodological synthesis to reconfigure relations between data science and social scientific inquiry.

The study of influencer professions (Blank, 2007; Fine, 2001) and more recently e-fluentials – i.e. people that influence using social media, has attracted much attention across disciplines. Capability of anticipating the rise of an influencer - which is key to marketing as well as public awareness campaigns – is believed to be more cost-effective than other more resource-intensive information diffusion strategies (Bakshy et al., 2011). However, agreement on how influence can be quantified, and influencers identified, is far from being achieved.

Computational measures based on centrality of a node in a network (Kitsak, 2012) are criticised for being applicable only to certain types of content spreading (Centola, 2010). Computational network analysis approaches also underestimate social factors affecting spread including homophily and community structure (Weng et al., 2013). Indeed, as demonstrated by empirical studies, the majority of influentials are "monomorphic": that is to say that their opinion is decisive only on certain matters (Katz and Lazarsfeld, 1955; Merton, 1968). In the likely absence

of a general-purpose definition of influencer, what social network analyses provide might just be ad-hoc explanations.

Meanwhile, empirical sociology beganEdinburgh-based developments in the study of new professions, including the first extended academic study of the industry analyst professions (Pollock & Williams, 2016).

Presented will be main findings of pilot study contrasting top analysts with a random sample of the professional population, confirming career data as a viable source of insights for influencer identification.

Carty, A. M. (University of Westminster, England) Conversations across the Valley: Using narrative therapy in film to create safe spaces for dialogue between fractured rural communities Although boundaries are blurred, crudely speaking there are four distinct local and incomer communities in Machynlleth and the Dyfi Valley in mid Wales. Over forty years the Centre for Alternative Technology (CAT) has brought in significant numbers of mainly English staff and volunteers, many of whom stayed in the area. It has been argued locally that CAT and these incomers largely failed to engage with the existing communities, and that this influx contributed to the loss of Welsh language and culture in the area. Previous video-based ethnographic research in the area had alerted the team to a degree of existing mistrust between the communities, and particular resentment towards CAT from those not directly involved in the organisation. It was felt that a reflexive filmmaking methodology, guided by principles and techniques from narrative therapy, could act as an ethically sensitive approach to overcome mutual suspicion and create a recognition and understanding between the different groups.

At CAT’s invitation we used their audio-visual archive to develop a reflexive and iterative process involving screenings, discussion and feedback in order to create a conversation between these different social groups and CAT. In order to engage people who might otherwise not be willing to take part we framed the project as an opportunity to see archive film of the area and engaged nearly two hundred people across the valley in filmed discussions about CAT’s local and global impact. We then edited these comments into a new version of the film, which we returned to CAT staff to increase their awareness of local opinion and feeling, eliciting their responses and incorporating them into a new iteration of the film. Narrative therapy techniques were used to manage negative emotions arising from the reflexive film process and to guide the overall safe trajectory of the project. This is an ongoing exploration where we aim to take CAT’s views back to the local communities in a subsequent stage of the project but the work so far allows for reflection on uses of both narrative therapy and reflexive film techniques as a tool for community engagement and ethnographic research. We would suggest that participants’ highly positive responses to the process and requests to continue the work demonstrate the success of

the method as an engagement process, but that a successful outcome as a “product” of visual ethnography for a wider audience has yet to be assessed.

Keywords: Visual methodology, community engagement, narrative therapy

Čehovin, G., Manfreda, K. L., Bosnjak, M. (Faculty of Social Sciences, University of Ljubljana) Meta-analyses in survey research: A systematic review The primary goal of this contribution is to present a systematic review of meta-analyses that were previously conducted in the field of survey methodology. First, we systematically identify such meta-analyses, then classify them in terms of the core thematic areas they addressed, and finally we identify gaps in research, pointing to avenues for possible future meta-analyses in the area of survey methodology.

To this end, we will summarize (a) which research problems and questions could have been addressed in principle, (b) which were actually addressed, (c) how the respective research questions were translated into inclusion criteria, (d) how the procedure of identifying and selecting primary studies were performed, (e) which data on effect sizes and potential moderators was extracted, (f) which analysis procedure was pursued, (e) and how the findings were interpreted.

With this presentation, we contribute to further popularize meta-analysis in the area of survey methodology, which has evolved as a well-respected approach for deriving evidence-based conclusions about causal and correlational relationships in the social, behavioral, health and economic sciences.

Cernat, A. and Oberski, D. (University of Manchester; Tilburg University) The multi-trait multi-error approach to estimating measurement error Measurement error is a pervasive issue in surveys. One of the most common approaches used to measure and correct for systematic errors in this context is the Multi-Trait Multi-Method approach. Thus, it is possible to sep-arate method, random error and “true” score using an experimental design that combines multiple traits (i.e. questions) with multiple methods (i.e. answer scales). As with other statistical approaches that tackle measure-ment error the results of this model are biased if any other types of systematic error (such as social desirability) are present. In this paper we present an extension of this model, which we name Multi-Trait Multi- Error, that manipulates multiple characteristics of the question format using a within factorial design. Thus, it is be possible to estimate simultaneously: social desirability, acquiescence, method, random error and "true" score. We will illustrate how to implement the design and show initial results using measures of attitudes towards immigration in the 7th wave of the Understanding Society Innovation Sample.

Keywords: Measurement error, experimental design, Understanding Society, latent variable modelling

Cernat, A. and Sakshaug, J. (University of Manchester, England) Nurse effects in survey biomarkers Surveys are extending the types of data they collect. An important new source is biological data such as blood or saliva collected from respondents. Nevertheless, this exciting new data source brings with it also a number of methodological issues. One of the most important ones is selection due to non-response and lack of consent to biological data collection. Our research aims to tackle this issue by looking at how nurses, respondents and areas influence 1) the likelihood that individuals will respond to the survey, 2) consent to biological data collection, and 3) the quality of the biological samples collected in two major British surveys: Understanding Society and English Longitudinal Study of Aging. This research will inform how nurses can have an impact on selection and data quality and how to model these effects as part of substantive research. It will also inform possible ways in which data collection could minimize nurse effects.

Keywords: longitudinal surveys, biomarkers, non-response, nurse effects

Cheung, G. and Johnson, J. (Survey Research Center, The University of Michigan & Centre for Longitudinal Studies, UCL) Use Cases by capturing structured metadata to document complex CAI instruments using DDI3 The collection and dissemination of large scale studies often involve multiple organizations. The projects are often complex and managing the flow of information from study design to dissemination through to archiving requires careful thought. Primarily for reasons of efficiency and data quality it is important to have systems that manage this information in a consistent and transparent manner. But as the information passes through multiple hands and computer systems information it is often lost or degraded or its provenance is not well connected requiring manual intervention to restore this. This is often a significant cost and a source of potential error. Additionally, the scale and complexity of the CAI instruments is a significant barrier to making the survey collection transparent and comprehensible to survey managers, researchers and analysts and for its data management.

In this paper, we will discuss two complex large scale studies one in the US, the other in the UK which used different data collection systems for the instrument programming (Blaise and IBM (now UNICOM) Data Collection). We will also discuss how we developed systems using the DDI3 metadata standard to provide computer-process-able metadata like web-based information systems, traditional codebooks, and command setups for statistical packages, question banks, which assist in the use or interpretation of the data. We will discuss in depth how structured metadata in a standardized form can be used across the survey life-cycle e.g. for instrument design, data entry, data processing, maintaining data documentation, and capturing and storing the metadata within a repository for later reuse independent of data collection implementation.

In this paper we will also present some challenges we faced in capturing metadata during the survey life-cycle in particular as complex projects move into multi-mode data collection and how the use of structured metadata enables reproducible research and reuse of survey components for other waves or surveys.

Keywords: Metadata; Data Collection; CAI

Cirucci, A. (Kutztown University of Pennsylvania, US) Narrative Demographics INTRODUCTION

The goal of this study is to introduce and assess a less violent method of interpreting socio- cultural issues. In addition to completing traditional demographics surveys, participants will be asked to share their stories through writing autobiographies, thus capturing their perceptions and interpretations of lived experiences. Because, as a society, we are fixated on categories such as race and gender, autobiographies will likely still capture these trends. As such, this method will still work to capture real-world, wide-spread trends regarding racial, gender, and other identifications, just as traditional demographic surveys sometimes do. At the same time, however, participants are invited to craft their own narratives, free from drop-down options, check boxes, and suggested, or preferred, labels. Arguably, this new method will more accurately represent the human experience as an open-ended identification space as well as providing study participants with a more enjoyable, empowering, and reflective experience.

BRIEF REVIEW

Demographics are often restrictive. While researchers utilize socially validated human categories, they also reify notions of differentiation and division. Traditional demographic breakdowns highlight points of division, such as race, gender, age, and socio-economic status. These differences only thrive through their constant use in comparing, ranking, and judging social groups.

Indeed, a cyclical process exists wherein customs and cultural “truths” are constantly repeated and eventually understood as intrinsic or natural. To be “masculine” or “feminine,” for example, are only culturally and socially created notions that are reified through performativity, and thus accepted as “normal” (Butler, 2006). Racial minorities, as another example, are generalized to shallow understandings (Hall, 1996), and these racial stereotypes, like gender’s, are violent— they force a person into a macro version of the self that rejects other identifications (Perinbanayagam, 2000).

With the mass adoption of social media platforms in the US, categories and discrete, quantifiable modes of identifying the self have been magnified. As digital users, people are constantly asked to input finite information that clusters them into categories, easily identified and archived by computers. Limiting categories, made particularly apparent through social media spaces, have recently been contested. Facebook users, for instance, fought for more fluid gender options beyond the binary “female” and “male” options. This became particularly relevant when, in 2008, the site required that Facebookers choose one or the other (Gleit, 2008). In short, our new culture of digital, visible, and “authentic” selves has worked to further naturalize human difference and reify that haphazard division is a standard, even proper, practice.

METHOD

This study will employ Matthew Joackers’ Syuzhet package for R to create simple narrative arcs that act as new vehicles with which to compare variables. Goals are two-fold. The first is to introduce a new “identification” standard that no longer promotes violent categories. The second relates exploring the ways in which simple narrative arcs can be used to better appreciate acts on, and through, social media spaces, including concepts relating to: identification performances, comprehension of privacy agreements, and cyberbullying.

I will recruit 500 participants and have already received a grant to assist with this process. Participants will be asked to write an autobiography of about 500 words. They will then be asked to complete a short survey that asks them about (1) their social media usage and (2) traditional demographics information. Each narrative will be uploaded to Syuzhet, and the software will perform sentiment analyses, outputting a simple narrative arc for each. The Syuzhet software will also cluster the narrative arcs. At this time, it is not known how many types there will be since they are dependent on my sample.

I will compare demographic trends to simple arc clusters in an attempt to understand the ways in which simple narrative arcs and demographics data are similar and different. Comparing the arc types to demographics percentages will assess if simple narrative arcs can represent social groups as they experience the world through socially created, and culturally mandated, identifications. Lastly, I will compare the simple narrative arcs to my social media usage survey

data by comparing the simple arc clusters to the trends presented regarding the ways in which participants noted they use social media spaces. I will discuss the ways in which all three subsets of my study are similar and different, and I will conclude with a final assessment regarding the viability of replacing traditional demographics with narrative demographics.

Class, F.* (University of Potsdam, Germany) Do “Grievance Statistics” measure quality of life in the GDR? Investigation of the validity of a new administrative data source by means of individual data This presentation introduces “Grievance Statistics” – an entirely new administrative data source from the German Democratic Republic (GDR) – and a strategy to validate the meaning of these statistics using data from the German Socio-Economic Panel (GSOEP). Grievance statistics are accounts of the GDR’s county authorities on the number and contents of “Eingaben” – written or verbal petitions and complaints – which were a dominant mode to air discontent in the GDR. An exhaustive source of these statistics are the protocols of the county councils of the GDR, which are available in East German county archives. The German Research Foundation (DFG) funded project “Spontaneous Revolution or Long Turn” (SRLT; KO 2239/3-1) has recently collected these data and entered them into a machine readable dataset which will become publicly available in due time. The data cover the years 1970-1989 in around 200 counties, including almost all counties covered in the GSOEP. One aim of the SRLT project is to describe trends of subjective quality of life in the GDR between 1970-1989. As the decision to write a petition in a socialist state such as the GDR is affected not only by the level of discontent, but also by the amount of repression, and the probability to successfully alter the source of the discontent, it is an open question whether grievance statistics provide valid evidence on subjective quality of life. It is thus necessary to investigate this validity, or, in other words to study the meaning of the grievance statistics. This is the aim of the presentation. The presentation is structured in three parts. In the first part of the presentation “Eingaben” and “Eingabenstatistiken” are introduced. It is shown how “Eingaben” were processed by the GDR’s administration and compiled into grievance statistics.

*This presentation is based on research funded by the German Research Foundation (DFG KO 2239/3-1

The second part focuses on problems encountered in the data entry step. These problems include amongst others the fairly heterogeneous formats and varying report periods of the presented statistics. Furthermore, the categorization of the topics of the grievances did frequently not follow GDR’s legal regulations. The dataset itself is organized as spell dataset. The third part of the presentation deals with the above mentioned validation of grievance statistics. The grievance statistics collected at the county level by the SRLT project include almost all counties to which the start addresses of the GSOEP sample C belong. The East German GSOEP data of wave 1990 contain data on actual “domain satisfactions” and – retrospectively – of the year 1985. For cross-validation, all these domain satisfactions will be aggregated to the county level and gradually correlated with corresponding grievance frequencies of the years 1985-1989. If the grievance statistics contain valid information regarding quality of life, the following hypotheses should hold. • The correlation between grievance frequency and quality of life is negative. • The correlation between data on the same topic is higher than the correlation between different topics. • The correlation decreases as the time lag between two measurements increases. Keywords: German Democratic Republic (GDR), quality of life, German Socio-Economic Panel.

Comanaru, R.-S. and d’Ardenne, Jo (NatCen Social Research, Questionnaire Development and Testing Hub) Young people and harmonised questions – findings from cognitive interviews Today, there are many governmental surveys aimed at retrieving information related to the social and economic aspects of everyday life in Britain. Many of these surveys have evolved separately from each other and currently use a variety of measurements to quantify the same concepts. For this reason, the Office of National Statistics UK has established a programme to standardise these measurements across departments. The “harmonised” concepts and questions are now becoming the norm in collecting certain categories of data across national surveys in Britain. These questions can now be compared across years and across surveys, and thus provide robust national statistics data to inform social and economic policy.

The harmonised questions have now been included on many large-scale surveys and allow for a standardised set of inputs and outputs, which make comparison more appropriate. They have also been used on surveys aimed at various groups within the population, including young people.

The harmonised questions include measurements related to demographic information, ethnic group belongingness, general health and long-term health conditions, illnesses, impairments and disability, as well as questions about well-being, gender and sexual identity and educational attainment. The Office of National Statistics is currently undertaking the task of developing more harmonised questions to be included on future surveys, which will allow for a better comparison of data.

The present study looks at the appropriateness of using the harmonised questions with young people, between the ages of 11 and 19. Using cognitive interviews, a subset of the harmonised questions are investigated for clarity and comprehensibility with a group of young people, who vary in terms of geographical location, gender and age. The sets of harmonised questions tested are related to ethnicity, religion, long term health conditions and al identity, and wellbeing. The aims of the cognitive interviews include ensuring that the questions are easily understood by all participants. and identifying potential comprehension, judgement and recall errors pertaining to these measurements as understood by young

participants. Using observations, think-aloud and probing techniques, the harmonised questions were discussed with the young participants in great detail. Our results suggest that some questions are easily understood by the young people regardless of their age. Cognitive testing also showed that other harmonised questions require a brief explanation or definition of certain terms in order to aid younger people’s comprehension. .

These results are considered significant in light of the ONS’ attempt to standardise and harmonise these measurements. Many surveys today collect data from adults over the age of 17, and some are addressed even to younger audiences. Ensuring that the harmonised questions are clear and comprehensible to all age groups is thus an essential part of “harmonising” the measurements. This study investigates the appropriateness of the ONS harmonised questions for people aged as young as 11 years old, and makes recommendations for each set of questions in light of the cognitive interviews results.

Keywords: harmonised questions, cognitive interviews, young people

Connelly, R., and Gayle, V. (University of Warwick, England and University of Edinburgh, Scotland) Administrative Data Quality and Measurement: A case study cross-validating UK Birth Records The use of administrative data is becoming increasingly common in sociological research, and has the potential to provide new and rich sources of information. There is a long track record of using administrative data for social science research in the Nordic countries, but in other nations (such as the UK) until recently access to administrative data have been much more restricted. Administrative data are originally collected to organise, manage, deliver or monitor services and are not primarily collected for social science research. Social surveys predominantly collect data for social research and there is a now a large methodological literature on survey data quality. The challenges associated with ensuring that high quality survey data are collected and the strategies for optimising survey data analysis are widely understood. Methodological work on data quality in administrative social science datasets is some way behind, this paper makes a unique contribution to that field.

In this paper we present a case study using administrative data to showcase a number of issues associated with the validity and reliability of administrative data. The case study examines official data from UK birth records on parental occupations. These measures provide a potentially important source of information on the social circumstances of children (e.g. socioeconomic position) and could be operationalised in a wide range of sociological studies (e.g. on social mobility). We are in the inimitable position of being able to cross- validate a set of official birth records using social survey interview data collected as part of the UK Millennium Cohort Study.

The results indicate that there are only moderate levels of agreement between the (official) administrative records and the survey data which is collected by trained interviewers. There was a large amount of missing information in the administrative records. Mothers are frequently classified as ‘unemployed’ or undertaking ‘full time care of home/relative’ in the administrative records, but report in the survey that they have a valid occupation. Mothers with lower levels of education and mothers from less advantaged social classes are more likely to have missing occupational information in the administrative records.

Administrative social science data offers exciting and unparalleled possibilities for sociological research. We conclude that sociologists should be cautious however when using administrative data. Sociologists should also be mindful of the methodological lessons on data quality that have emerged from survey data collection and analysis. We provide some recommendations for sociologists working with administrative social science data.

Cottingham, M. and Erickson, R. (University of Amsterdam, Netherlands and University of Akron, United States) Using Audio Diaries to Investigate Emotion Research on emotion is fraught with methodological limitations, as feelings can have non- discrete, ephemeral, and ineffable qualities. Audio diaries offer a method for capturing the sequential and varied experience of emotions as they emerge from everyday life. Following theory and methodological development in the sociology of emotion, we examine how audio diaries might be used to capture (1) spontaneous emotions, including those linked to microaggressions, (2) processes of emotional reflexivity, and (3) the sequential and simultaneous nature of diverse emotional experiences and their management. We further explore how audio visualization might be meaningfully integrated into qualitative analysis of audio diary data to illustrate emotional changes. We draw on audio diaries collected from 48 nurses from two U.S. hospital systems to explore the possibilities and limitations of using diaries in emotion research.

Coulton, J. (University of Leicester, England) The Art of Noise: Using Audio Soundscapes in Spatial Research Conducting ethnographic study often affords the researcher a wealth of opportunity to employ a variety of data gathering tools. When the study extends beyond the observation of people or specific processes, particularly into explorations of space and place, the chance to be ever more innovative becomes not simply an option, but almost a necessity. This paper outlines how the recording of audio soundscapes was employed during an ethnographic investigation into the management and manipulation of space in a British airport terminal building. This was not the sole, or even primary method for gathering qualitative data, but one which supplemented the observation field notes, photographic visual images, interview responses, and focus group data; analysis of the audio data, did however, contextualise and help to make a great deal of sense of all the other material gathered as part of this year-long period of intensive fieldwork. Many aspects of using audio soundscapes are explored here, from the issues surrounding gatekeepers and gaining access, through the challenges of negotiating ethical approval, issues surrounding the practicalities of recording ambient sounds in a voluminous interior space, to the analysis of lengthy recordings. The key areas of interest that are drawn from this study, are just how the background sounds that are produced by spaces, change so frequently in such a relatively short time frame, and how these enable the ethnographer to drawn on Lefebvre’s (2004) early work on ‘Rythmanalysis’, and develop a much more transparent ‘picture’ of precisely what is going on in the space at any particular moment in time.

Reference: Lefebvre, H (2004) Rhythmanalysis: Space, Time and Everyday Life. London : Continuum

Cusatelli, C. and Giacalone, M. (University of Bari “Aldo Moro”, Italy and University of Naples “Federico II,” Italy)

Information and Communication Technology for Judiciary Big Data Management The need for strong regulatory and valid instruments is clear in contemporary society to resolve disputes in a timely manner and penalize wrongdoing. The judicial function must be measured against this complexity, in order to ensure the safety of citizens, social cohesion, and economic competitiveness at an international level. The Information and Communication Technology (ICT), that facilitates knowledge and exchange of data and information through an analytical approach to problem solving, is of paramount help in tackling this complexity.

The new technologies have become essential, thanks to the enormous possibilities that they can offer, as the immediate transfer of a document from one continent to another with just a click, saving both time and money and also allowing many simultaneous interpersonal exchanges, that considerably speed up the decision-making processes which involve numerous individuals located in various places. Furthermore, we have witnessed over a very short period of time, that most of the human activities which were carried out manually have given space to much more efficient digital implementations. For instance, we can consider the serious problems that the vast documental archives have created in its management, and how centralized computer databases helped to solve most of these problems, speeding up and optimizing all research operations and data mining. This natural easiness of data exchange is still being expanded and facilitated by the development of computer networks, and in particular by the internet.

The Big Data Analysis allows the management of large amounts of data of different nature and from various sources, having great importance also in judicial proceedings. In investigative and judicial field, analysis of correlations, semantic enrichment and sentiment analysis have proved valuable tools for feedback of projective type: such a statistical analysis thus appears to be of fundamental support in judicial activity. Moreover, Big data require specific tools in the analyses that need large amounts of data to get reliable results. ICT in the administration of justice offers possible solutions, improving the administration of justice and helping to streamline procedures and reduce costs. The prior condition for the development of an information system in the field of justice capable of responding effectively to the needs of users consists of the availability of reliable basic data on the activity of judicial offices.

Despite progress made in recent years, the quality of the data is still one of the critical aspects of statistical production in the legal field and is partly due to the lack of accurate data provided by the peripheral structures, especially in the civil subsystem, where the measurements are still in course of automation. The quality of the production processes is largely linked to the degree of computerization of the detection process, which affects the feasibility of obtaining data with a level of detail sufficient to fully meet the needs of users. So, on the basis of data collected through the ultimate European Commission for the Efficiency of Justice (CEPEJ) evaluation scheme, we will describe some partial and global, respectively unweighted and weighted, composite ICT indicators.

Keywords: ICT, Big Data, Justice

Czepek, J. (Institute of Employment Research, Germany) The history of the return on human capital investment: Opportunities and challenges of multiple administrative data In Germany, debates in the 1960s led to extensive educational reforms. Unquestionable, these quantitative and qualitative changes in educational participation reduced inequalities. However, individual outcomes of an increasing educational participation are an ongoing touchstone for investigation: Assuming a limited number of qualified positions in the labour market, the expansion of high skilled workers should lead to a growing competition. Thus, many expect a relative devaluation of the economic value of these qualifications (educational inflation). In contrast, the driving force for investments in human capital was the demand for labour in high skilled positions. Therefore, the question whether a higher education leads to higher income in older ages or if an educational inflation occurs is still open. We compare four cohorts in the pre- and post-educational era to observe differences in the return of investment. In Germany, the individual pension income strongly depends on lifetime earnings. Additionally, individual pension entitlements reflect ones position in relation to the average wage. Both, the individual contribution to the pension scheme and the relative wage position make the entitlements in later pension income a perfect measurement for a cohort study focusing on differences in human capital outcomes. We used a unique set of data called BASiD (biographical data of selected social insurance agencies in Germany). Linked micro data is provided by the German National Pension Insurance and the Federal Employment Agency and allows us to investigate employment biographies of 568.468 persons. The surveyed cohorts were born between 1940 and 1977. Our results show that younger cohorts continuously lose pension entitlements in comparison to older cohorts. Unsurprising, results on the return on investment are more positive for persons with vocational training than for those without skills having longer periods of unemployment. High skilled workers started their career later but with higher incomes that resulted in higher pension entitlements. However, later birth cohorts of higher skilled workers still had lower pension entitlements compared to their earlier born counterparts. Women gained the most profits of human capital investments given their

increased rate of participation in both the educational system and on labour market. However, part-time work and periods of home caring limited their return on investment. The potential of this linked administrative data set is not exhausted yet. In the contribution, we may discuss the results of the empirical study with their limitations in interpretation depending on the provided administrative data. The discussion could offer even more potential applications of BASiD for research in social sciences. Keywords: “Administrative Data” + “Linked data” + “Longitudinal Analysis”

Czepek, J. and Tisch, A. (Institute of Employment Research, Germany) Analysing public administrative data from two different organisations: A historical analyses of mothers’ pension credits in Germany Debates on gender equality in the German pension system are ongoing. Additionally, the ageing of society challenges the national pay-as-you-go system and requires removing barriers for child care. Recent pension reforms mainly address mothers since legislative changes improve the recognition of child care for the primary carer. Another goal was to establish more social justice towards parents with children born before or later than 1992.

In Germany, national pension insurance is closely linked to employment because individual statutory pension entitlements are gained by employment subject to social security contributions. Pension levels are calculated on collected pension credits and proportional to lifetime earnings. However, mothers can collect additional pension credits during times of child care. Before 2014 mothers have only benefited with pension credits for one year per child born before 1992 and with three years of recognition per child born later than 1992. After the most recent pension reform, retroactively, pension credits are recognized for two instead of one year per child born later than 1992. To justify the retroactive recognition two arguments are stated: Firstly, one reason for legislative changes was the appreciation of child rearing as a common good. Secondly, proponents of the reform intended to additionally compensate mothers for their opportunity cost due to child rearing. Many argue that in the 1980s and at the beginning of the 1990s almost no public child care facilities were available and mothers had no choice but to interrupt their employment careers to raise their children. When child caring facilities became available during the 1990s, full-time child caring turned more to a voluntary decision.

In our study, we ask whether the reform provide an equal subsidy to mothers with employment interruptions due to child rearing, independent from the timing of their child births. We aim at testing whether social differences on the macro level of society can be moderated by providing support on the individual level and how ex-post chances regarding the pension entitlements contribute to future pension levels. This paper discusses methodological problems of our analyses. It particularly gives an overview on advantages and disadvantages of linking data from two different administrative processes.

We make use of the longitudinal individual-level data set BASiD (Biographical data of selected social insurance agencies in Germany). BASiD is a combined data set of the German National Pension Insurance and the Federal Employment Agency. Linking data from two different institutions offers the opportunity to amplify the information available. Our results show that the combination of two different administrative data sources can help to explain how macro phenomena such as pension reforms have impacts on the individual level. However, we also point at limitations which occur when administrative data is used to analyse social (in- )equality. We discuss advantages and disadvantages of both data sets as well as of the combination of both data resources with special regard to the evaluation of the most recent pension reform in Germany.

Key Words: Process-Oriented Methodology – Longitudinal Analysis –Historical Analysis

100

D'Antonio, V (Sapienza University of Rome) Privacy, publicity and self disclosure in the Big Data Era: From the «right to be alone» to Oversharing Part of a larger PhD thesis on big data, social media and information ethics, this contribution discusses the emerging concerns about privacy and the release of personal information on social network sites (SNSs). With increasing frequency and with particular impact on the youngest age brackets of the population, people share intimate moments and details of their personal and private lives on on SNSs. As social media usage permeates people’s lives, an increasing portion of their daily behaviour leaves digital traces online. Institutions, businesses and even single citizens get to know more about people through the information that the latter – more or less consciously – render public and available on the Web. This mere observation would indicate that the urgency to redefine the categories of privacy and publicity, and to read Big Data and the access to personal data as a renegotiation of what is be considered “public” and “private”, has emerged. The “right to be forgotten” (Mayer- Schönberger 2009; Rodotà 2014; Suárez Villegas 2014) is involved in a continuous oscillation with the tendency to overshare (Labrecque et al. 2011; Belk 2013; Agger 2015). Privacy online today is, more than ever, the object of a dynamic interdisciplinary debate involving Social Sciences, Psychology, Law, Philosophy, Computer Science, in particular concerning government surveillance and personal data management. What is at the center of this debate, however, is the way we not only conceive, but also problematise the theme of privacy since the rise of the Internet and, in particular, in the "Big Data Era," a period characterized by the most massive production and sharing of data that mankind has ever witnessed. In the proposed research, the focus is on the triadic relationship between privacy, publicity and self-disclosure in networked publics (Boyd 2010), paying particular attention to the Facebook platform. This research aims to investigate a field of study as yet little explored, especially in Italy, with the objective of understanding the youth culture in connection with the use of Facebook and privacy concerns. In particular, the attention is placed on how teens use the social media affordances and the resources that Facebook makes available to them to protect their personal data and lifes or, on the contrary, oversharing on Facebook their innermost thoughts, and why. The aim is to define 'how' and 'why' teens use (or not use) this particular medium to express their “digital self” and manage their self presentation online, and to understand the meaning that they give to their own choices.

References Agger, B. (2015), Oversharing: Presentations of Self in the Internet Age, Routledge, London Belk, R. W. (2013), Extended Self in a Digital World, Journal of Consumer Research, Vol. 40, No. 3, 477–500 boyd, d. (2010), Privacy and Publicity in the Context of Big Data, WWW10 Conference in Raleigh, North Carolina

101

Lauren I., Ereni Markos, and George R. Milne (2011), Online Personal Branding: Processes, Challenges, and Implications, Journal of Interactive Marketing, 25(1), 37–50. Mayer-Schönberger, V. (2009), DELETE: The Virtue of Forgetting in the Digital Age, Princeton University Press Rodotà, S. (2014), Il mondo nella rete: Quali i diritti, quali i vincoli, Roma-Bari, Laterza Suárez Villegas, J. C. (2014), El derecho al olvido, base de tutela de la intimidad, TELOS Cuadernos de Comunicación e Innovación, No. 97

102

Danciu, A., Le Corgne, S. and Mairot, A. Center for Socio-Political Data (CNRS/SciencesPo) An overview of metadata and data management at the French Center for Socio-Political Data (CDSP) Secondary analysis is gaining a central role in contemporary social research. This type of research uses existing materials. Finding the appropriate data is a crucial step of a research project. Researchers can contact the French Center for Socio-Political Data to obtain quantitative surveys.

The CDSP data dissemination system includes platforms that allow data discovery (a question bank, a data discovery webviewer…) and data download. The several data dissemination platforms are not inter-operable. When we publish a new study, we have to publish it on each of the dissemination platforms. The consequences are a risk of de-synchronisation of the versions of the files. Besides this, the metadata is not entirely harmonised across waves/studies. Data labels may be different for the same question asked across time. In this context, researchers may encounter difficulties in finding the data they need.

The presentation will address:

How can we make data more available and usable for the scientific community by determining how best to recognise continuities between data within the studies, including question continuity and methodological continuities.

How to create visualisations of all the waves including the question and download the ones of interest.

How through the use of concepts, we aim to make our catalogue more attractive for the scientific community.

Harmonisation of studies/waves over space and time and across studies, will allow researchers to gain an easier and improved access to the data and metadata.

The paper will present the programs, interfaces and procedures that are to be designed in order to accomplish this purpose, in the context of a workflow including limited human and material means

103

Keywords: Metadata; Repository; Harmonisation

104

Das, B. and Subudhi, R. (School of Management, KIIT University Bhubaneswar, India) Influence of socio-economic factors in DSMM Convenience, novelty and simplicity are the prominent characteristics of Digital & Social Media Marketing (DSMM). At the same time, intensity and extensity of marketing activities on social media is a big challenge for any business, particularly in Developing countries. This very challenge provokes management studies researchers to explore as much as possible in terms of gathering deeper knowledge and to attempt to make it useful for all stake holders. This paper covers various socio-economic aspects of ‘users’ with their ‘Engagement pattern’, within the framework of DSMM. The aim is to create engagement opportunity and build up the relationship value and maintain it with long term objective. It tries to capture as to, how users look at online reviews and online purchases. Finally, their behaviour in terms of sharing the online experiences within friends, family and social group is examined. It is found that DSMM users of all age groups demonstrate quite positive indulgence in checking for online reviews and seeking information in social media. But the inclination to, share online experiences and influence buying decisions of others, actually declines with advancing age groups. It attempts to find out the contributions of most influencing factors, in making DSMM more effective and sustainable.

Keywords: Socio-economic survey method, DSMM, Online Reviews

105

Davis, Peter (University of Auckland, New Zealand) The health effects of patients switching medicines after subsidy changes. For over 20 years New Zealand has been operating a rigorous reference pricing system in setting subsidy levels for medicines (drugs, pharmaceuticals) dispensed to patients under the country’s comprehensive, tax-payer-funded national health service. Under such a system a subsidy is set at the level of the cheapest available medicine within a given therapeutic group.

This system has reaped huge savings for the New Zealand taxpayer, but it can require patients to switch medicines (for example, from an originator to a generic drug) - in part because of the price differential that opens up once reference pricing is introduced to match the cheaper drug. Could this pressure to switch drugs possibly be detrimental to patient health?

We analysed the health effects of switching to generic medications for four medicines in the neurology or mental health field. Three studies used a controlled before and after design and one an uncontrolled before and after. In each we carried out a “difference-in-difference” analysis in cohorts constructed retrospectively either side of the introduction of generic drugs to the market. We compared changes in health events before and after the subsidy shift between patients who switched brands and those who stayed with the originator.

No differences in pattern of health events were detected for any of the drugs. While quasi- experimental designs can never be free of the risk of bias “difference-in-difference” methodology allows for the control of unobserved biases that do not change over time, and therefore enjoys advantages over standard before-and-after observational studies.

Keywords: pharmaceutical subsidy policy; difference-in-difference; health outcomes

106

de Leeuw, E., (Utrecht University)

Co-authors: Joop Hox, Peter Lugtig, Jolien Cremers, Rob Kessels, Merel Plat, Marielle Zwijnenburg, Timo Brakenhoff, Danielel McCool, Bente Oterman, Sanne Smid, Sandra Taal, Laura Boeschoten, Brett Ory, Pascal van Kooten, Anen Brons, kees Mulder, Kirsten Schutter, Eva Zijlmans (all Utrecht University)

To gamify or not: an experiment into the effectiveness of gamification in an online survey Already in 1987 Dillman advocated that researchers have to engage potential respondents and make the survey experience pleasant. Puleston (2012) advocated gamification, which emphasizes the importance of ease, interest, and fun for online questionnaires. Gamification soon became a buzzword in marketing research, but only a few empirical studies have been published. For an overview see Keusch & Zhang (2015). In an online experiment we investigated four different gamification principles and compared these with a control condition. After a friendly but neutral introduction, respondents were randomly assigned to five conditions. In the control condition a brief questionnaire on leisure activities was presented using common best methods for online research. In the ‘scenario’ method a cartoon professor asks for assistance and addresses the respondents in an encouraging way through the questionnaire; in the engaging version, the questions were reformulated to sound more game like; in the visual version pictures were added to response options, and in the direct reward version, respondent could earn a virtual coin for each answered question. No clear differences were found between the experimental conditions. Respondents did not differ in experienced enjoyment, subjective evaluation of completion time, or data quality. Keywords: Text gamifying, visual presentation, scenario, respondent-oriented

107

Decataldo, A., Amico, A. and D'Alessandro, G. (University of Milan Bicocca, Sapienza University of Roma, Italy) Longitudinal research on students careers. A quasi-experimental design on Sapienza University of Rome This paper presents the results of research that aims at the evaluation of Italian Higher Education reform policies and their outcomes. In the last fifteen years the Italian university system has been characterized by radical reforms introduced by the “Bologna process”. The most important Reform (DM 509/1999) aimed to solve three main aspects of the Italian university system: low number of graduates, excessively long careers, very high number of drop-outs. The AY 2001/2002 represents the transition from a university based on the single level degree to a stratification into two levels BA (3 years) and MA (2, 5 or 6 years).

Three main agencies have dealt with the university system evaluation: CNVSU, ANVUR and ALMALAUREA. Their analyses are cross sectional and repeated over time. Instead, our issue is to evaluate the impact of the reform considering career paths through longitudinal data. The research is a secondary analysis of longitudinal administrative data in order to analyse the phenomena of dropping out, perpetual students and low number of graduates. Sapienza University of Rome was identified as a representative case due to its dimensions and its variety in terms of scientific areas. Information concerns the cohorts of the enrolled students before (from AY 1991-92 to AY 2000-01) and after the reform (from AY 2001-02 to AY 2012- 13). The pre-2001 cohorts were monitored until March 2008 and the post-2001 cohorts until March 2014. The information is collected by Sapienza for administrative purposes, such as monitoring the fee payments and the exam registrations. However, the information is collected in different archives. Furthermore, the information is registered in a synchronic way (each information refers to a single moment). Starting from these archives, a unique dataset has been re-constructed in a longitudinal way, using the key codes (such as the identification numbers of the study course and of the students).

The final dataset has longitudinal information about all students without a previous career (544,025). The organization of this dataset allows us to follow students’ careers from first enrolment until graduation (or dropping-out). Through the adoption of a quasi-experimental research design, we are able to attribute to the introduction of the university reform (which we consider as the experimental variable), the differences between the outcomes of

108

academic careers of post-2001 cohorts (that are the post-test population) and the ones of the pre-2001 cohorts (the pre-test population). In particular, with reference to the classical experimental approach, the adopted research design represents a combination of the Time- Series and the Separate-Sample designs. In this way, we are able to provide an assessment of the reform with reference to its main aims. In other words, it is possible to compare the outcomes of the careers of the students that enrolled into pre-reform cohorts with the post- reform ones.

The research outcomes show that the transition from the old to the BA/MA university system had a greater impact on the regular graduation rate and a weaker impact on excessively long university careers, and on the dropping out. Overall, the reform was a partial success.

Key words: Reform Evaluation, Impact Analysis, Higher Education

109

Decataldo, A., Amico, A. and D'Alessandro, G. (University of Milan Bicocca, Sapienza University of Rome, Italy) Management and Analysis Strategies of Administrative Databases: Transforming Data into Longitudinal Vectors Nowadays in social research greater information availability and higher costs of data gathering make convenient the use of existing databases. In social sciences, even when the researchers have access to a large amount of information, it cannot be directly used for the research purpose. Indeed, the vast majority of the data gathered and recorded each day are not collected for the scopes of the social research, but for administrative, business or communicative purposes. It is the task of a researcher to choose the information of the greatest relevance and the highest quality, and to organize the database in an optimal way.

Firstly, the case study shows the construction of a longitudinal dataset of information extracted from the administrative archives of the Sapienza University which refer to the registered students. Starting from those archives the database has been re-constructed in a longitudinal way, using “keys” codes to join the data in a unique dataset. By positioning the events on a string, temporary ordered, it is possible to control the dynamics of each academic career and to highlight specific phenomena (dropping-out, stopping-out, mobility, degree rates, etc.). These data allow us to follow students’ careers from first enrolment until graduation (or dropping-out).

Secondly, this paper shows the analysis of this longitudinal dataset in order to highlight the utility of this kind of data structure and with an aim to: (1) prove that the Sequence Analysis (SA) tools are suitable for studying the complexity of Italian higher education students’ careers; (2) analyse the factors (also contextual) that have had an impact on the success/failure throughout the students’ academic careers (through Event History analysis, EH).

The SA is an efficient method that involves the examination of ordered social process. We use SA in order to (1) describe in detail the phenomena of late graduation and late performance (retention); (2) identify different kinds of students who are dropping out from the university (attrition); (3) evaluate other particular phenomena that could delay the careers (such as the mobility within and between faculties).

110

Through the implementation of EH models it is possible to study simultaneously the changes occurred over a particular period of time in different contexts and in different students’ careers. The EH allows precisely to relate the duration and the performance of the processes with the independent variables or covariates that are theoretically relevant. It is, therefore, possible to make inferences about the influence of covariates on the length of the careers and on the occurrence (or non-occurrence) of an event.

The main results of SA show the importance of the warming-up period (in terms of study course reorientation and first year performance) in determining the success/failure of each career. Furthermore, the EH models show the relevance of some external conditions, that can affect student's decision to use the university as a parking place.

Key Words: Process-Oriented Methodology – Longitudinal Analysis – Students’ Careers

111

Decataldo, A. and Denti, F. (Department of Sociology and Social research, University of Milan Bi-cocca) Increasing participation rates and completeness of questionnaire compilation in web survey. An experimental design of research Declining participation rates and data quality are serious problems in survey research. This is particularly true with reference to Internet surveys, where the absence of interviewers may pose an additional threat to interviewees’ propensity to respond and to complete questionnaire. In this paper we focus on the impact of: 1) e-mails and SMS reminders on survey participation; 2) data linkage on completeness of questionnaire. Our web survey aims at students registered during the AY 2015/16 in degree courses at the University of Milan Bicocca (young and educated per-sons, who are comfortable with internet use).

With reference to the first issue, in web surveys e-mails and SMS can be used as remindersto improve participation rates. Relative to the second issue, the evaluation of data completeness implies to analyze several indicators, such as the percentage of missing values and response set, the lack of personal data, etc.

Our study assesses the impact of different types of reminders on survey participation through an experimental design of research and it represents an interesting example of data linkage between university administrative data and survey data. In particular, this data linkage exemplifies an opportunity to improve the completeness of the questionnaire compilation (reducing the number of items), and to reduce the percentage of inconsistent data (i.e. the percentage of data that are different for the same subject in administrative database and in the responses to the question-naire). As a matter of fact, the procedure of data linkage has the advantage of making more sustaina-ble the compilation for an interviewee. The researcher releases the respondent from providing the information that are already available through administrative sources.

To check our hypothesis, we consider eight experimental groups:

- T1.1-SMS, that will receive SMS only as a reminder and a questionnaire with additional questions about administrative information;

- T1.2-SMS, that will receive SMS only as a reminder and a questionnaire without additional questions about administrative information (through the use of the data linkage only);

112

- T2.1-E-mail, that will receive an e-mail only as a reminder and a questionnaire with additional questions about administrative information;

- T2.2-E-mail, that will receive an e-mail only as a reminder and a questionnaire without additional questions about administrative information (through the use of the data linkage only);

- T3.1-E-mails and SMS, that will receive both SMS and an e-mail as a reminder and a questionnaire with additional questions about administrative information;

- T3.2-E-mail and SMS, that will receive both SMS and an e-mail as a reminder and a questionnaire without additional questions about administrative information (through the use of the data linkage only);

- C.1-no reminders, that is the control group and it will not receive any reminders, but a questionnaire with additional questions about administrative information;

- C.2-no reminders, that is the control group and it will not receive any reminders, but a questionnaire without additional questions about administrative information (through the use of the data linkage only).

Keywords: experimental design; reminder impact; response rate

113

Dechant, A. and Rinklake A. (State Institute for Family Research (ifb), Germany) Longitudinal qualitative interview data - comparisons within and between couples over time In order to understand change over time, it has become more widespread to collect and analyse longitudinal qualitative data. In this line, to gain insights in couple processes, the collection of qualitative longitudinal interview data on both partners of couples at least at two points in time becomes more frequent. This approach enables researchers to analyse how interviewees nested in couples experience, anticipate and respond to changes. The comparison of different couples who face similar changes in their lives allows to identify similarities and differences in ways of reacting to a similar situation and to find reasons for these similarities and differences.

Qualitative longitudinal data on couples offer a huge potential but also demand a complex strategy of analysis in order to take into consideration the various levels of interest. In this paper we concentrate on how to organize the analysis to take advantage of the specific nature of the data. This includes the analysis of the single interviews, a comparison of both interviews within a certain couple, a comparison by gender and over time. As an empirical illustration we use interviews with couples during their transition to parenthood. Both partners of these couples were interviewed during pregnancy and about one year afterwards. The main focus of the qualitative semi-structured interviews was the division of paid and unpaid work among the partners. Those aspects were evaluated from a present-day perspective as well as retro- and prospective points of view.

The interviews are analysed in an iterative process. First, the research question is analysed cross-sectionally for the first interview of each individual then, the descriptions of both partners in each couple are compared to each other focussing on similarities and differences. In addition to the within-couple analyses, the different couples are compared to each other. Furthermore to analyse the influence of gender, the interviews are also compared within the group of women and men, respectively. During these steps, it is likely that issues emerge in the interviews that were not foreseen in the initial coding scheme. If this is the case, an iterative process starts in order to check if the issue is also important for interviewees or couples that were already analysed.

114

Second, the research question is analysed cross-sectionally for the second interview in the described manner. In this process, it is possible that the informants discuss aspects that were not included in the coding scheme of the first interview. Therefore, another iteration of analysing the first interviews can occur after the analysis of the second interviews. Third, to use the information of the longitudinal data structure, the first and second interviews are compared to determine and explain changes regarding the research question over. This comparison is undertaken for each interviewee, for each couple, between the couples and between the women and men, respectively.

Keywords: qualitative research; longitudinal research; couple data

115

Delamont, S. (University of Cardiff) Time to kill the Witch It is a commonplace that ethnographers learn a great deal from critical encounters: from the ‘First Days in the Field’ to quote Blanche Geer’s famous (1964) Paper. Far less has been written about Leaving the Field, about insights gained from exit strategies. This paper focuses on what can be learnt from exits, drawing on a long ethnographic project on the ways in which the Brazilian martial art capoeira is taught and learnt in the UK. In capoeira serious students have nicknames, and the one I use in publications is Bruxa the Portuguese word for witch. I ask myself in the paper – ‘is it time to kill the witch?’

116

Demirkol, E. (University of Sussex)

The Balance between Emotions and Research: A reflection on studying emotions among transnational migrant families Emotions are an inevitable aspect of our everyday lives. They, both, shape and are shaped by our interactions with the other people, social institutions/structures and places, which make our emotions unique and different at the end of the day. They are powerful enough to control our decisions, which would lead an entire change in our lives. Although emotions are so powerful in our daily life, the focus on the study of emotions is a relatively new area in social sciences. It has recently attracted social scientists, and migration studies is one of those disciplines. However, it is still a grey area what kind of methodology we could apply to define, research, measure or conceptualise emotions in migration studies. Applying the same as theoretically is also another challenge while studying emotions. This is a significant issue both in migration studies and other social science disciplines. Also, personal emotions while doing interviews gain importance, too. Both as a researcher and researched face the complexity of diverse emotions emerged in stories of leaving home or the other way around, being left behind. Although, during the interview, the researcher tries to protect his/her reflexivity, it could not be easy to handle with the emotional dynamic that could arise between the researcher and the participant. What could the researcher do during the interview if an upsetting issue comes to the scene for the researcher as well, as a listener? How could the researcher control this situation? How can the researcher change the topic or actually should change the topic or not even he/she is also suffering from the conservation? What could be those ‘appropriate’ emotions for the researcher during the interview to continue to interview under the scientific conditions? And what about afterwards? How could you approach your transcripts with the massive, tiring bag of emotions to protect your professionalism? While a reflexive approach is a solution for the situations, still how can we be that sure about our reflexive position? Therefore, the aim of this paper is to discuss the difficulty of studying emotions as a part of my PhD study, which focuses on the impact of transnational migration on family life in the general framework. Specifically, its purpose is to gain a better understanding of how transnational migrants and their non-migrant relatives build relationships across nation-

117

states. To do this, the present research proposes to focus on the construction of family relations among Turkish transnational migrants, particularly labour migrants, in Japan and their non-migrant family members in Turkey and how they face their emotions, how those emotional relations change between the wife and the husband, the father and the child and the child and the mother. And what kind of difficulties emerged during the interviews, what kind of emotional stages I have faced in myself and between with my participants while talking about their emotions, their feelings

118

Deviatko, I. (National Research University Higher School of Economics, Moscow) “Virtual labs”, “social telescopes” and other metaphors of innovative online methods: new opportunities and new challenges to validity of social science research The growing interest in the academic and applied research communities to understanding the consequences of a radical transformation of the conventional methods of social and behavioural sciences as a result of the global digital revolution is the most obvious trend of the second decade of the XXI century. The digital revolution has turned a considerable part of the global population into the "Virtual Villagers", access to whom, on the one hand, simplified (according to ITU preliminary statistics, Internet access rate reached 43.4 per 100 inhabitants for the world and 82.2 for those living in developed countries), and on the other, partly because of the above-mentioned simplification, became subject to new risks of reducing the quality of the social science data − from "survey weariness" to "professionalization" of participation in such reactive procedures as surveys or experiments. These ambivalent developments has led to the intensification of methodological experimentation and increasing amount of meta-analytical studies devoted to comparison of external and internal validity of the data collected offline and online, as well as possible approaches to improve the quality of online questionnaires and study designs, taking into account the specific threats to the validity and reliability of measurement as well as the validity of statistical inference threats emerging in online research. Some of the latter has been recognized rather incurable, like self-selection bias in web surveys (Betelheim, 2010), while others still remain underestimated or even portrayed as benefits, like using crowdsourcing websites for participants’ recruitment (Gosling et al. 2004; Gosling, Mason, 2015, but see also Chandler, Mueller, and Paolacchi, 2014).

While supposedly nonreactive measures, like public online records and other digital footprints (including mobile sensing technologies or metadata), are claiming to superiority in validity of statistical inference, reliability and even immunity to social desirability effects, the empirical evidence behind these claims remains meagre. The review of a number of “old” and “new” threats to validity for translational, phenomenological and innovative online research (Skitka, Sargis, 2006) will be presented (with the special attention paid to transformations in both data sources and to our current understanding of the very idea of

119

selection bias). Basing on accumulated findings from different sources, the general conclusion can be substantiated that not only web surveys, but also web-based experimentation as well as the nonreactive Internet data (including dataset archives, log files, etc.) can provide the biased estimates leading to invalid conclusions due to different forms of self-selection and over-coverage (Pury, 2011) errors which can hardly be alleviated due to the progress in Internet access per se.

Keywords: web surveys, online experiments, Web-based nonreactive data collection, self- selection, over-coverage, data quality

120

Dietrich, M. and Mey, G. (Magdeburg-Stendal University of Applied Sciences, Germany) Towards a Visual Grounded Theory Methodology: A framework for image analysis in culture studies During the last decade, sociologists and cultural scientists have often argued for a “pictorial turn” (Mitchell 1992) in the predominantly logocentric social and cultural sciences, and especially in the field of qualitative methodology. At the same time, a productive discourse about images and how they can be analyzed more systematically has unfolded internationally and in Germany (Margolis & Pauwels 2011; Leser & Kauppert 2014). German scholars, for example, offer sophisticated approaches to image analysis based on concepts of Barthes (1957; 1964; 1980), Panofsky (1932; 1955) and Imdahl (1996) while other approaches are developed and intensely debated. Perhaps surprisingly, one of the most popular methods among social scientists, the Grounded Theory Methodology (GTM), has only rarely been extended to allow for the study of images (Konecki 2011). In contrast to the GTM credo “All is data” (Barney Glaser 2007), a GTM focused on images, a Visual GTM, irrespective of whether they function as a tool for data collection or as already produced images becoming an object of a study, has not been established until today.

Texts and images are different type of media. In our paper, we outline a Visual Grounded Theory (VGTM) that focuses on the analysis of manifest images and its systematic integration with text analysis. We point out how (a) images circulating in popular media can be analyzed with a procedure sensitive to their mediality and (b) texts linked to images can be integrated in the analysis in the VGTM approach. We demonstrate the VGTM framework drawing on examples taken from the analysis of covers and texts of German punk magazines (sample: 1981–2013) in a large-scale research project on presentations of youth, juvenility, and generation(ality) in youth cultures funded by the Federal Ministry for Education and Research.

Keywords: Visual Grounded Theory Methodology, popular media analysis, juvenility

121

Dluhopolskyi, Oleksandr Ternopil National Economic University, Ukraine Comparative studies of social capital Development in Ukraine Post-Revolution of Dignity 2014 The modern ideology of reforms, made systematically both in industrial and in developing countries, is based on the achievements of the new political economy and institutional theory, using macro and microeconomics theory. Recently, importance for reformation has been given to the institutional modernization that was not taken into account in the ideologies of the reformers. The main problem in developing countries (especially in Ukraine after ‘Revolution of Dignity’- 2014) in the context of formation of institutional environment study is insufficient. From 2008-15, a large-scale study of social capital was conducted by Legatum Institute. Statistical data and survey results are summarized in a single index of social capital, which is called Legatum Prosperity Index. The worst condition of Social capital index in 2012-14 was observed in India (132 position 2014 and 138 in 2012), while the best was in Norway (1st position in 2012-14). When comparing the rate of social capital change, a significant improvement is observed in Portugal (+21), Turkey (+19), Ukraine (+18), Argentina (+17) and Pakistan (+15). But since most of these countries belong to the developing economies, the level of social capital there is low compared to the developed economies. But never the less in such countries like Belarus and Russia social capitals level during 2012-14 increases just by 4-5 points. Ukraine ranks 63 globally in the 2014 Prosperity Index between 142 countries, having risen one place since last year. Ukraine’s best performance is in the Social Capital sub-index, where it ranks 40 in 2014. Ukraine’s lowest rank is in Governance sub-index, where it ranked 12 in 2014. The deterioration of social capital primarily relates to the developed markets. This occurred due to global financial crisis, which “undermined” public confidence not only in governmental and financial institutions, but also in beneficial motivation and mutual support. In the developing countries, the levels of social capital increased, which was due to its low overall index, and citizens’ desire to find new ways of solving social and economic problems, “bypassing” the state. In many developing countries similar situation exists in the areas of health care, education, public administration, defense and security, since the state does not perform its functions. We emphasize that the impact of institutions on the dynamics of economic development is twofold,

122

both positive and negative. Everything depends on the initial institutional conditions, which determine the success of reforms. If the initial set of institutions happened to be relatively ineffective, the economic system will reflect those ineffective conditions unless the new situation will arise. In the developing countries (including Ukraine), excessive centralization of political system and fixed vertical power structure often narrow the space for social initiatives, reduce the possibilities of social capital for improvement of living standard and economic growth. “Modernization from below” is blocked by the absence of public trust in government and that, individual or collective efforts to protect the violated rights can bring positive results.

Keywords: Prosperity Index, Social capital Index, Social development

123

Dorau, R. (Bundesinstitut für Berufsbildung) Comparability of job histories between datasets The occupational integration of individuals is often analysed and classified as being integrated, precarious or disaffiliated (e.g. the zones of integration by Robert Castel). The definitions of these classifications vary, sometimes even the subjective handling of the occupational situation is part of the definition. However, for cross-sectional data referring to the occupational status there are some crucial, more or less generally accepted aspects for the assignment of a particular occupational status to one of the three types of occupational integration.

Precarious occupation is most often defined as being paid with less than 2/3 of the median income or by working in a temporary employment. In contrast, an occupation is defined as integrated if an employment contract is permanent and the payment is above 2/3 of the median income. While the former two definitions solely refer to the occupational status at a particular date, the definition of affiliation additionally takes into account the duration of unemployment and short-term employment. Only if these statuses persist in the long run the occupational integration is classified as being disaffiliated.

However, assigning job entry histories to a zone of occupational integration is more complex if one does not merely want to analyse the end of the observation period. For the alignment of different employment histories the method of optimal matching analysis is commonly used. In this case the clustered sequences often cannot easily be assigned to a zone of occupational integration and the comparison between datasets – does not make sense.

To evaluate histories in longitudinal studies we have developed two indices, one for integration and one for disaffiliation. Both indices include parameters for the total time of statuses, assigned to a particular zone of occupational integration in the observation period, the duration of continuous statuses and the number of employment interruptions. These parameters are weighted by the temporal incidence of these statuses and events. The index for integration is constructed as the ratio of the integrated parameters compared to the precarious and disaffiliated ones, while the index for disaffiliation is defined as the ratio of the decoupled

124

parameters compared to the parameters of the other two zones of occupational integration. If there is overall sufficient information regarding occupational integration (unlike histories characterized by family work, further training etc.) for the integration index as well as for the disaffiliation index the parameters for integration respectively disaffiliation should be dominant for defining a history as integrated or decoupled. Otherwise, if none of the parameters is dominant, histories are assigned as precarious.

As an example we analyse a three-year period of young people entering the labour market after completing vocational training, using the data of the German Sample of Integrated Labour Market Biographies (SIAB) 2008 and compare it to a supplementary survey of the BIBB/BAuA Labour Force Survey 2012.

125

Dörfler, T., Brandt, S. and Klärner, A. (University of Bayreuth, University of Hamburg, and University of Rostock, Germany)

Towards a Relational Analysis of Social Space – Methodological Implications for the Research on Space and Place The imagination of space as a surface on which we are placed, the turning of space into time, the sharp separation of local place from the space out there; these are all ways of taming the challenge that the inherent spatiality of the world presents. Most often, they are unthought.« (D. Massey, On Space, 2005: 7)

Contemporary spatial analysis experiences a wide reaching and quite unpredicted florescence, which is related to what is discussed as the ‘spatial turn’ in social sciences (Döring/Thielmann 2008). This shift in conceptualizing space also brought achievements in doing research on spatial aspects of the social (Thierbach/Raschke/Hering/Baur 2014). Instead of considering (social) space as ‚static‘ or ‚innate‘, spatial analysis nowadays focusses on the multiple aspects of how space is ‚made‘, ‚constructed‘ or ‚produced‘ by interactions, discourses and other social processes (an overview in Dörfler 2010: 15 ff., Dear/Flusty 2002; Belina 2011, a new approach in Baur/Herring/Raschke/Thierbach 2014).

In spite of that, not many sociological or geographical approaches have incorporated an explicit empirical framework for the analysis of (social) space (as exceptions – in part – may count Bourdieu 1999 and Löw 2001). Most of the works on globalization, transnationalities, home (- lessness), migration or other aspects place-related identities lack a sufficient analysis of (social) space as a constitutive part of the sens pratique of the involved subjects.

In contrast we chose the daily life sphere (Lebenswelt) to offer clues for that and consider habitation as a spatial practice (Hahn, 1997) based on lived space (Bollnow 1961). In its unavoidable givenness it influences biographies and mentalities and shapes habitus and doxa. While this is often neglected in contemporary theory, early phenomenology already conceived »living-in« as a corporeal basis for every human being (Merleau-Ponty 2012

[1945], Bollnow 2011 [1963], comp. Cresswell 2016). We will therefore focus on the ‘material’ aspects of social relations in a habitat to explain why every social practice needs a socio- spatial

126

environment like atmospheres to be performed (Dörfler/Manns 2012). To understand how place and space are relevant and contested aspects of every social interaction, we need to approach them in a relational perspective (Donati 2010, Bourdieu 1999), founded on corporeal („leibliche“) experiences of the other (Wacquant 2016, Dörfler/Rothfuß 2016).

Given this background our paper provides a framework for analyzing the relevance of place and space under conditions of a ‚fluid‘ or ‚flexible’ modernity. We exemplify this by asking how do (post-)modern individuals refer to place and space in a post-traditional manner? We will give an example of how place-related identities can be analyzed by reconstructing place and space- related positionings based on the concepts of ‘spacing’ and ‘synthesis’ (Löw 2008). They are processes relevant for constituting spatial aspects of life-world-experiences of subjects. Lifeworlds as well their spatial aspects bear on interpretations of relevances (‘synthesis’) to specific place-related experiences (‘spacing’). It is therefore central for every empirical analysis to reconstruct these experiences as typified and shared knowledge of the social (Schütz), depending on class-specific incorporations (Bourdieu): the relational incorporation of the other’s socio-spatialities to position oneself as a relevant actor in a social field.

References: Baur, Nina ; Herring, Linda ; Raschke, Anna Laura ; Thierbach, Cornelia: Theory and methods in spatial analysis: towards integrating qualitative, quantitative and cartographic approaches in the social sciences and humanities. In: Historical Social Research 39 (2014), 2, pp. 7-50. DOI: http://dx.doi.org/10.12759/hsr.39.2014.2.7-50 Belina, Bernd (2011): Raum. Münster: Westfälisches Dampfboot Bollnow, Otto Friedrich (1961): Lived-space. Philosophy Today 5 (1), 31-39. Bollnow, Otto Friedrich (2011/1963). Human space. London: Hyphen Press. Bourdieu, Pierre (1999): Site Effects. In: Bourdieu, Pierre et al.: The Weight of the World: Social Suffering in Contemporary Society. Stanford: Stanford University Press, pp. 123- 29. Cresswell, Tim (2016): Place and Space. In: Janz, Bruce (ed.): Contributions to Hermeneutics. Springer. Forthcoming Dear, Michael/Flusty, Steven (2002): Spaces of Postmodernity. Oxford: Blackwell. Donati, Pierpaolo (2011): Relational Sociology: A New Paradigm for the Social Sciences (Ontological Explorations). Milton Park: Routledge. Dörfler, Thomas (2010): Gentrification in Prenzlauer Berg? Milieuwandel eines Berliner Sozialraums seit 1989. Bielefeld: transcript. Dörfler, Thomas/Manns, Carsten (2012): Suburbane Atmosphären als Milieuräume. Ein Beitrag zur sozialwissenschaftlichen Raumforschung. In: Ausdruck und Gebrauch. Architektur Wohnen Umwelt, Heft 11_2012, S. 128-155.

127

Dörfler, Thomas/Rothfuß, Eberhard (2016): The Social Reconstruction of Place – A Post- Ontological Perspective. In: Janz, Bruce (ed.): Contributions to Hermeneutics. Springer. Forthcoming Hahn, Achim (1997): Wohnen als Erfahrung. Reflexionen und empirisch-soziologische Untersuchungen zur Pragmatik des Wohnens. Münster: Lit Verlag. Löw, Martina (2008): The Constitution of Space. The Structuration of Spaces Through the Simultaneity of Effect and Perception. In: European Journal of Social Theory. Volume 11, No. 1, 2008, S. 25-49. Löw, Martina (2001): Raumsoziologie. Frankfurt a.M.: Suhrkamp. Merleau-Ponty, Maurice (2012 [1945]): Phenomenology of Perception. London: Routledge Thierbach, Cornelia/Raschke, Anna Laura/Hering, Linda/Baur, Nina (Hrsg.) 2014: Spatial Analysis in the Social Sciences and Humanities. Towards Integrating Qualitative, Quantitative and Cartographic Approaches. HSR – Historical Social Research / Historische Sozialforschung 39 (2), Special Issue. Wacquant, Loic (2016): A concise genealogy and anatomy of habitus. In: The Sociological Review, Vol. 64, 64–72 (2016) DOI: 10.1111/1467-954X.12356

Keywords: Spatial Analysis, Social Space, Relational Sociology

128

Dunning, M. Social processes and the development of ‘jihadist terrorists’ There are both long-term and short-term processes that contribute to the development of so- called ‘jihadist terrorism’. That is, there is an interdependent relationship between what Elias referred to as ‘sociogenetic’ processes at the societal level and ‘psychogenetic’ processes at the individual level. Among the many difficulties when researching such processes is how to develop an understanding of how these ‘long-term’ and ‘short-term’ processes interweave with each other. In relation to ‘jihadist terrorism’ we can ask ‘how have long-term social processes contributed to the development of individuals who have undertaken acts that have been labelled as “terrorism”, and how does one go about researching such processes?’ In his seminal work On the Processes of Civilisation (2012) Elias showed how changes on the sociogenetic level contributed to changes on the psychogenetic level and vice versa. Accordingly, he was able to show how state formation processes contributed to and were interdependent with a range of psychogenetic changes, such as growing thresholds of shame and embarrassment in relation to bodily functions, controls on violence, increasing rationalisation and an increase in mutual identification and foresight. At the same time these ‘micro-level’ processes contributed to changes at more ‘macro’ levels of integration. He was able to show this through an examination of time series data from manners books and other literary and pictorial sources. In this paper I discuss how a similar research process can be applied to developing knowledge of ‘homegrown’ ‘jihadist terrorists’. That is, through the examination of time series data from documentary sources we can see how complex inter- and intra-state social processes are interdependent with the development of individuals who undertake ‘terrorism’ as part of a ‘global jihad’ against their perceived enemies. In this particular case I examine these processes in relation to the so-called ringleader of the 7/7 London bombers, Mohammed Sidique Khan, including the difficulties posed by such an endeavour.

129

Durand, C., Charest, A.-S., Valois, I. and Ibarra L. P. (Université de Montréal) The use of longitudinal multilevel models to understand change in trust in institutions over time and between countries The paper will present how multilevel longitudinal analysis is used to perform a comparative and historical analysis of change in trust over time in different countries. We combine together all the public data sets that include questions pertaining to trust in institutions. This merged file includes the Eurobarometer, the Latino Barometro, the Asia Barometer, the Arab Barometer, the Africa Barometer and LAPOP (Latin American Public Opinion Project). All these surveys have asked similar questions about trust in a number of institutions like the police, the army, the government, religion, media, and the like. The idea is to figure out the likely change in trust over time taking into account factors that occur at three level. First, at the country level, events that have occurred in different countries or regions of the world may have influenced trust in some institutions. Second, at the individual level, socio-demographic characteristics like age, sex and socio-economic status are likely associated with trust. Finally, at the measurement level, the same individuals are asked their evaluation of trust in different institutions. Trust may vary according to the institution that is the object of trust. We therefore end up with a multilevel model at (at least) three levels, i.e. the country/year level (which may be divided in two levels), the respondent level and the measurement level. The paper will present how this type of analysis allows for answering questions that cannot be answered easily using other models, like, for example, whether the relationship between age and trust in police varies over time and between countries.

130

Durrant, G., Sturgis, P., Maslovskaya, O. and Brunton-Smith, I. (University of Southampton, University of Surrey, England) Interviewer effects on response latencies in a face-to-face interview survey Analysis of paradata can greatly help to improve survey processes. One form of paradata is the length of time it takes a respondent to answer a survey or particular survey questions. Interview length is often used as a key indicator of both response quality and fieldwork costs and a better understanding of interview length is therefore important. One key influence on interviewer length in interviewer administered surveys is the interviewer.

It is well known that interviewers are key in determining the quality and cost of data collection in household interview surveys. They influence the likelihood that sample members will respond and, therefore, the bias and precision of estimates. They also affect the answers that respondents give in ways that can make them more, or less accurate. These effects have been robustly demonstrated in the existing literature. In this paper we consider a different and less well studied outcome; the extent to which interviewers affect the time respondents take to answer questions. We apply a cross-classified mixed-effects location scale model to response latencies in the UK Household Longitudinal Study. The model extends the standard two-way cross-classified random-intercept model by specifying the residual variance to be a function of covariates and additional interviewer and area random effects. This extension enables us to study interviewers’ effects on not just the ‘location’ (mean) of response latencies, but also on their ‘scale’ (variability). Furthermore, by linking the response latency data to paradata on interviewers from an independent survey of interviewers, we are also able to model these interviewer effects as a function of their demographic, work experience, attitudinal, and personality characteristics.

131

Eberle, J., Heining, J., and Müller, D. (Institute for Employment Research (IAB), Germany) A modern Job Submission Application to access IAB’s confidential administrative and survey research data. The Institute for Employment Research (IAB) is the research institute of the German Federal Employment Agency. Via the Research Data Centre (FDZ) at the IAB, administrative and survey data on individuals and establishments are provided in a standardized way to researchers. Administrative data originate from employment notifications to social security providers and information on unemployment benefit receipt and registered unemployment collected by the German Federal Employment Agency and municipalities. IAB also carries out several comprehensive surveys that are provided to external researchers via FDZ. While some of these data sets are anonymized Scientific Use Files and can be transferred to research institutions, access to so-called weakly-anonymous research data sets can only be granted in a controlled environment. These are highly sensitive data sets with very detailed information on persons or establishments (although no original or clear-name identifiers are entailed) and show a high risk of disclosure. There are currently two ways of access for the weakly-anonymous research data sets of IAB: The first is on-site access in safe rooms at IAB or access centres at partner institutions, the second is remote data processing via job submission. When using job submission, researchers cannot see the actual data source. They submit program code that runs on the data and receive output files that were reviewed for disclosure risk. In cooperation with the Institute for the Study of Labour (IZA), IAB now provides the sophisticated Job Submission Application (JoSuA) that allows users to interact with IAB via a custom-built web interface. In order to better support the research process, a new distinction is made between interim and publication outputs. Interim output on the one hand is not downloadable and can only be viewed as image files within a secured environment. Results are provided immediately, only a filter script runs on the output to reduce disclosure issues. Publication output on the other hand can be downloaded by the researchers and therefore needs to be checked more closely for any

132

disclosure risk. This is done manually by IAB experts. The innovative distinction between interim and publication output increases the efficiency of remote data processing and at the same time reduces disclosure risk. The new Job Submission Application at IAB therefore provides the best possible usability while ensuring a high data security standard. Researchers have access to the highly detailed data sets and can even switch easily between on-site access and job submission, depending on the needs of the current research phase. This presentation will describe the Job Submission Application used at IAB and provide some feedback from researchers.

133

Edwards, D. (University of Leicester, England) Understanding Everyday Participation: Making the case for "short and dirty" multi-sited ethnographies Understanding Everyday Participation: Articulating Cultural Values (UEP) is a large, multidisciplinary and joint University-led project using qualitative and quantitative methods to explore the forms and practices, detail, dynamics and significance of peoples ‘everyday participation’. Working as an ethnographer on the project, I am responsible for conducting four ethnographies in four different locations in England within three years and for managing household qualitative interviews also taking place in those locations. Conducting ethnography to research an all-encompassing term such as ‘everyday participation’ and in these diverse locations was a daunting undertaking, especially given the tight timelines and budgets of such large projects. In the months prior to the ethnographic work, initial scoping research is carried out in the location to find the focus for the ethnography. Once a focus has been found, the intensive fieldwork is scheduled a two month slot. At conferences I have found myself having to justify, mainly to anthropologists, this 'short-term field visit', which has been labelled as a "short and dirty" approach; rendering my work somehow less 'valid' amongst certain academic circles. There are varying accounts across disciplines of what constitutes a ‘real’ ethnography; all grounded in particular ethnographic time frames spent ‘in the field’. I will begin this paper with an introduction to UEP and discuss where ethnography is positioned in such a large project. Using examples from my 'mini-ethnographic' fieldwork in Manchester, Gateshead and Dartmoor I will continue to share experiences of the challenges I have faced on such a fast-paced project as this. For example, building trust and relationships with ethnographic research participants takes time and this important work can be difficult in a two month fieldwork timeline. But I will also contemplate the great benefits of having a 'focus' for the ethnography (Knoblauch 2005). For example, I have found that I have had to push the boundaries of my own knowledge and skills to find new ways of 'doing ethnography'; I have had to tweak the method/ology to fit the diverse needs of participants in each site making each ethnography unique. I will conclude this paper by suggesting that our "short and dirty" method, in its combination with interviews, mapping,

134

historical analysis and quantitative analysis, has been innovative in its exploration of everyday participation.

135

Enzler, Heidi Bruderer (ETH Zurich, Switzerland) Air Travel for Private Purposes. An Analysis of Survey Data Combined with Spatial Data Air travel for private purposes has increased sharply over the past decades and causes a non- negligible amount of greenhouse gas emissions. Therefore, it is important to know its determinants. The present study examines socio-demographic, spatial and attitudinal predictors of air travel for private purposes. The analyses are based on a large representative general population sample (the Swiss Environmental Survey 2007) and a subsequent computation of the respondents' environmental impact by air travel, as well as spatial data from various sources.

To account for spatial characteristics, measures of population density and potential airport access were created. In order to do so, the respondents' residential addresses as well as sixteen airports were geocoded. This allowed assessing local population density as well as straight-line distances, road travel distances and road travel times between the home addresses and sixteen relevant airports. Based on these measures of distance and airport passenger volumes, several versions of a measure of potential airport access were created for every respondent.

These data were then combined with the survey data and analyzed by means of lognormal hurdle models controlling for environmental attitudes, partisanship, health and socio- demographics. The results clearly indicate that air travel is related to airport access in the sense that better access is correlated to a higher probability of air travel and also to higher emissions. This result is robust to alterations of the accessibility measure and it even remains in place when a very simple measure, namely the straight-line distance and passenger volume of the closest airport, is used instead. As many previous studies have linked air travel to urbanity or population density, additional analyses accounting for local density were carried out. These indicate that airport access is positively correlated to air travel even when controlling for density. However, the question remains whether there is another variable

136

confounding the link between air travel on one side and density and airport access on the other.

137

Fischer-Neumann, M. (University of Bamberg/BAGSS, Germany) Religious Affiliation and Regime Support among Turkish Immigrants in Germany Over the last decades there has been a steady increase in the discussion about the role of immigrants’ religious affiliation in their integration process, as European countries have become religiously diverse due to large migration flows to Western countries since the 1960s. Phenomena as the political Islam presses recent attention to an important, yet under researched dimension of integration of Germany’s largest immigrant group: Turkish immigrants’ support for the political system of the German host society, which focuses on immigrants’ psychological sense of political inclusion. Despite contentious public and scientific debates which commonly suggest a negative relationship between a Muslim identification and democracy support, there is a lack of empirical and theoretical knowledge on the relationship. Hence, the current paper addresses the impact of first- and second-generation Turks’ religious and national identities on their support for the political regime in Germany by drawing on social identities theories. This broad theoretical framework suggest an interplay between religious and national identification as well as conditional effects on perceived discrimination. Longitudinal (random effects) regressions on data from two waves of the German Socio-Economic Panel (GSOEP) 2005/2010 show an independent negative effect of Muslim affiliation on Turks’ democracy satisfaction, which is moderated by perceived discrimination as well as low levels of national identification. Moreover, there is also indication that the effects are more severe for second- than first- generation Turks. In sum, the paper extents the theoretical and empirical understanding of the relationship between Turkish immigrants’ cultural identities and their cognitive political inclusion in the German host society.

138

Fearon, K. (De Montfort University, England) Reproductive decision making in women with Turner Syndrome: an exploratory study Turner Syndrome (TS) is a chromosome disorder in which 90% of girls are born without ovaries, or their ovaries fail before puberty. Developments in reproductive technology mean that women with TS can have a child via egg donation, either from a family member or an anonymous donor. Mothers of women with TS may be faced with decisions about whether and how to preserve their daughter’s reproductive potential, for example via maternal egg freezing, while some girls can freeze their own eggs or ovaries. Yet pregnancies in women with TS are risky: there is up to a 40% rate of miscarriage (Abir, 2011) and a 2% risk of death (Karnis, 2012). The ethical and psychosocial challenges of these choices, both for women with TS and for their families, are currently under- researched.

This study, currently in its early stages, uses photo-elicitation interviews to explore these issues in depth with two groups: women with TS and mothers of girls with TS. Photo-elicitation was initially chosen as an adaption: even allowing for the sensitive subject matter, women with TS tend to be self-effacing and reticent (Bondy, 2007) and the use of images taken by the participant can engender confidence and build rapport.

Photo elicitation was also chosen because of the way it can evoke perspectives that are not available through verbal questioning alone: in placing the visual over the verbal, it provides another way for participants to describe personal emotions and memories. As such it may provide an additional dimension to an in-depth interview, which could function to triangulate the verbal discussion (Harper, 2002).

Although Turner Syndrome is associated with a specific cognitive profile, published research rarely describes any adaptions to the way the research was conducted that addressed the potential requirements of participants. This research was designed to accommodate those needs.

However, the desire to be flexible and accommodating raises an ethical challenge: giving participants choices over some aspects of the way the interview is managed avoids imposing

139

uncomfortable or inappropriate conditions on them. Women with TS are likely to have executive function impairment, and so may have difficulty making choices; this approach may increase their stress. Paradoxically, a more paternalistic approach may be more appropriate.

In this paper, I will outline the reasoning behind the choice of research methods, and the adjustments I made to the methods to accommodate potential cognitive impairments. I will then discuss the consequences for recruitment, participant selection, communication with interviewees, conduct of the research interview, and follow up.

Keywords: Turner Syndrome, Visual methods, decision-making

140

Fuchs-Sawert, L. (Freie Universität Berlin, Germany) Maps in Discourse Analysis: The Construction of Space in Socioecological Conflicts in Peru The last decade’s boom of resource exploitation in Peru has led to an increase in socioecological conflicts that have become hotspots for the negotiation of national identity. The conceptualisation of national territory is of particular importance here, because conflicts related to natural resource exploitation link social dynamics with material space. The way this space is represented is relevant for the dynamics and outcomes. Consequently, maps are predestined as objects of analysis.

According to Anderson’s concept of “Imagined Communities”, we imagine our communities not only dependent on the words we hear and read, but especially on the pictures we see; maps become the “logo of the nation”. In the Peruvian case, the nation is imagined as being split in three very different and utterly separate parts: The costa, the sierra and the selva (coast, mountains and forests). This came about through a process of transferring ethnical difference, as it followed from colonial rule, into geographic location. Each part of the country was given a role within the nation in terms of population and national identity. This construction came into being after Peru’s independence in 1821, and is still in place today. While the major concern of the political elite in the post- independence period was to keep the indigenous population under their dominance, current governments concentrate on resource exploitation. The imagined separation of the Peruvian territory, and the racialisation interwoven with it, play an important role here: Since resources are exploited primarily in the sierra and selva, resistance from these areas is dismissed as irrational and hostile to progress or made invisible. Thus, the depiction of the national territory becomes an instrument of domination.

The examination of different maps clearly shows these tendencies: The official map of Peru depicts the national territory as separated in the three zones mentioned, continuously, from the young republic until today. While the centre of political and economic power is located at the costa, the sierra and selva areas are depicted as full of natural resources ready to be exploited. Due to their association with the natural sciences, maps seem to offer an objective

141

and neutral depiction of material reality so that the power relations woven into them become invisible. Hence, they can be powerful means of producing (perceived) spatial reality.

Like other elements in discourses, maps are produced in a social context and contain power relations. Therefore, they can be understood as visual texts that can be deconstructed. I used a guideline for the analysis of maps, asking questions on who depicts what for whom and how/why. More specifically these included: Which perspective is the map drawn from? Are there any actors depicted? How is nature depicted? Which categories are used? Which elements are shown/left out? Which colours were used? Which scale is employed? Are there expressions of mobility or change?

142

Fernandez, K. (University of Graz, Institute for Business Education) Sampling challenges in ethnography of homeless youth groups The starting point for the article is a study I undertook with the aim to describe the patterns of street-careers. The research was undertaken in a group of homeless young people in Austria. The chosen research methodology was a Grounded Theory Ethnography, which included a six- month participatory observation and 45 interviews.

During the fieldwork several problems concerning the sampling strategy occurred. The ethnography was planned as a follow up study to an interview research I undertook several years ago. Building up on the results of this previous study, the sampling strategy was designed as theoretical sampling with already defined contrasting groups in the beginning. Since the aim was to describe street careers I wanted to undertake a comprehensive ethnography in the group on the one hand and on the other hand interview young people who were either not yet or not more part of the group. With this strategy, I hoped to observe youth who enter the group and leave the group in the time of the ethnography but also interview young people who are just starting to discover the group and others who have left the group already long time ago. Problems with this sampling strategy occurred on different levels. First of all, it was difficult to get access to the group. Even when I successfully entered the group, I did not achieve stable contact to all members. The group was parted into different groups of age. I obtained very good and stable access to the young people that were about my age at that time (young adults from 18 to 28). I could not achieve a stable relationship to the really young members from 14 to 18. Besides this general access problem, it turned out to be a big difficulty to conduct interviews. I wanted to interview some group members, who turned out to be very interesting for me in terms of specific categories in the theoretical sampling. Although they agreed to give an interview, they very often did not come to the appointments we made. It is a characteristic for this group, that they do not keep up to appointments and they do not answer phone calls for days. Additionally they live a very unsteady life, which often forces them to change their sleeping places or even leave the town for weeks and months. Some also went to prison or to psychiatric clinics and five group members died. This led to an extended period of

143

fieldwork, because I tried to find other interviewees who suit the sample design. Besides these efforts, I was constantly reshaping the sampling design itself. Similar problems occurred when interviewing the people who left the group. Especially those who achieved a stable life in socially accepted conditions did not want to remember their life in homelessness and often refused to take part in the study. In the oral presentation I will illustrate how I handled these challenges in the study and I will discuss what general conclusions can be made.

Keywords: Homeless Young People, Grounded Theory Ethnography, Difficult Access

Fitzgerald, R. (City University, London) ‘Attempting to reduce comparison error across the survey life cycle. Illustrations from the European Social Survey (ESS) and future solutions from the Synergies for Europe’s Research Infrastructures in the Social Sciences initiative’. Total Survey Error already raises a series of hurdles that the researcher must overcome before they can achieve equivalence of measurement amongst a single population. The additional challenges of a cross-national survey augment the existing error sources as well as introduce new ones that make the task considerably more challenging.

SERISS (The Synergies for Europe’s Research Infrastructures in the Social Sciences) is a new initiative that aims to address the most troublesome methodological challenges that cross- national surveys face and help to set new state of the art standards for cross-sectional and longitudinal survey methodology – from survey design to data access and storage.

Focused around main key themes – the key challenges facing cross-national data collection, breaking down barriers between research infrastructures, and embracing the future of social sciences – SERISS will address issues relating to survey design and data collection, data management and curation from a collaborative, cross-national perspective.

144

The European Social Survey is a leading cross-national general social survey that seeks to reduce comparison error by addressing TSE at every stage of the survey and data life cycle.

The paper will draw on experiences from the European Social Survey and SERISS to illustrate how large scale infrastructures in Europe are trying to minimise comparison error with limited resources. The importance of following an input harmonised approach will be stressed and the continuing challenges of balancing centralisation with local measurement concerns will be highlighted.

145

Furakhutdinov S. Sheludkov A. and Bystrov E. (Council of Municipalities of the Tyumen region)

Maps and spatial analysis for visualization and interpretation of cluster analysis results Cluster analysis is quite extensively used in sociological research. One of the crucial issues for the researcher is an optimal number of clusters, so he/she can, on one hand, not fall into excessive details describing, on the other hand, the number must be big enough to reveal the whole range of possible groups of elements in order to reach main research goals. In studies, where research objects have spatial localization, the use of mapping technics and spatial analysis can be of great help for determining an optimal number of clusters. This idea is based on the experience of research work dedicated to rural municipalities cluster analysis. The main purpose of the research was to describe the processes and phenomena, associated with economic and social polarization of rural areas in post-soviet time. The case was Tyumen region. This polarization process continues for more than 20 years. It is considered to be the result of socio-economic changes linked to the USSR collapse. Soviet administrative planning system was based on the state ownership of all capital goods. In the absence of private ownership and market relations any economic decisions, including the establishment of enterprises, were taken by government officials. In many cases they did it without taking into account such important factors as cultural and national peculiarities, the landscape features of the territory or historical forms of economic activity. For several decades the soviet planned economy was able to fundamentally change economic specialization and settlement system of many regions. However after its end new market forces were unleashed, which affected many of the processes taking place in Russian society. If the big cities adapted to the new system quite quickly, the rural areas still can feel the effects of changes. In order to describe these processes, to define their vectors and possible consequences our team has gathered quantitative as well as qualitative information about all the 293 municipalities of Tyumen region. We organized 10 field expeditions and visited for the period of year more than one hundred villages, including remote settlements accessible only in winter. On the basis of field data a hypothesis of the existence of 8 types of municipalities was put

146

forward. In order to check it we chose 7 key quantitative markers and conducted with SPSS Statistics a hierarchical cluster analysis using the Ward method and Euclidean distance measurement. It is believed that an optimal cluster model consists of the number of clusters, equal to the discrepancy between the number of observations and the number of agglomeration steps, marking the stage when the distance between two clusters starts to increase in leaps and bounds. If you do not stop clustering at this stage, it will be risk to unite into one cluster too different elements. To our astonishment a program showed that in our model an optimal cluster number was not 8, but 49. Deep interpretation of all the 49 clusters would take a lot of time, and such a huge number of clusters were really hard to perceive: we had to find any subtle differences between them. So we needed somehow to update our method. For this we used cartography and spatial analysis. First of all we noticed that socio-economic development of municipalities as well as whole phenomenon of rural territories’ polarization was determined by their geographical position. Natural and climate conditions, soil fertility, degree of remoteness from towns and major roads, proximity to oil and gas pipeline stations, the presence of large medical institutions on the territory, agricultural enterprises, administrative borders and national enclaves are the "magnets" with different force of gravity or repulsion. We used open source QGIS to put these “magnets” on the map to define their strength as factors distinguishing clusters. Subsequent analysis allowed us to reveal general trends and latent mechanisms of rural space polarization, discard minor factors, enlarge and rebuild our cluster model. Eventually we combined cluster model and center-periphery model, which consists now of 10 relatively stable groups of rural municipalities. Using mapping technics and spatial analysis is not only a good way to visualize data in order to ease its perception. This method is valuable as a research tool to identify new social phenomena. If the object of the research is localized in space, visual information allows complementing analysis with factors, difficult to reveal and describe by both quantitative and qualitative means.

147

Gaia, A. (Institute for Social and Economic Research (ISER), University of Essex) The influence on substantive results of the sampling strategies adopted in qualitative and quantitative paradigms: Evidence on men who pay for sex Sex workers’ clients are an elusive, hidden and rare population; given these characteristics, sampling men who pay for sex presents peculiar challenges. Both in the quantitative and qualitative literature, social researchers willing to collect first hand data on sex workers’ clients have often relied on purposive samples.

The research paradigm and the sampling strategy adopted have influenced the substantive results on the demand for prostitution. I focus specifically on two contrasting findings identified by the literature: the “every man” perspective which implies that sex workers’ clients are not statistically different in terms of socio-demographic characteristics and psychological traits from “non-clients”; and the “peculiar man” perspective, which implies that customers have distinctive characteristics and peculiar traits (e.g. psychological, social, or physical inadequacies) compared to “non clients”.

Qualitative studies which used purposive samples suggest that paying for sex is transversal to different social groups and do not present peculiar psychological traits; this result is based on the evidence of the existence of men who pay for sex in several different social groups. Conversely, quantitative studies relying on probabilistic samples have identified some social groups that are more likely to pay for sex than others.

The sampling strategy and the research paradigm adopted influence the substantive conclusions of the literature; while it is clear that the phenomenon is transversal to different social groups, there is also clear evidence, at least for Britain, that there are particular segments of society that are more likely to be sex workers’ clients than others.

148

I will discuss the difference in these substantive findings with relation to the sampling strategy adopted, highlighting the challenges in sampling men who pay for sex and the results obtained in both the quantitative and qualitative literature.

Keywords: sampling, hard to reach populations, sex workers' clients

149

Garner, M. (University of Roehampton, England) Do as I do, not just as I say: can students learn from their teachers’ research experience? The pedagogical approach that consists of a combination of teacher input, the study of appropriately chosen readings and focused class discussions might be styled the default approach to teaching research methods (RM). Although there have been few attempts at rigorously evaluating it, it is probably a reasonably effective means for students to learn about research, and few alternatives are proposed in the literature on research pedagogy. Equally importantly, however, students need to learn how to do research: what is involved in being responsible for the design and conduct of a project, including the many conceptual, practical and logistical problems that inevitably arise. A typical RM curriculum tackles this pedagogical need by front-ending learning about methodology and then requiring the students to do some research of their own, during which they often encounter unexpected problems for which they have been inadequately prepared. Are there ways in which, while learning about the methodology, they can also develop some sense of the inevitable mismatch between the ideal and the actual of doing research hands-on? Some writers (e.g., Roth, 2009; Seale, 1999) advocate an apprenticeship model for teaching RM, and there is some evidence that this can be an effective as a means of striking the balance. Apprenticeship, however, makes considerable demands on the teacher’s time and mental energy, It is impracticable in most classes, which consist of classes of more (often a great many more) than a few students, for everyone to be engaged in actually conducting research throughout the course. A more practicable alternative can be for the teacher to share with the students his/her own experience in particular research projects. This no doubt occurs frequently in many teachers’ practice, when specific experiences are used to illustrate specific points as they arise. I have done this in my own classes from time to time, but only ad hoc and informally. In the current academic year, I decided to try this as a systematic approach, with classes at regular intervals in which I recounted the history of a study, from the decision about the topic, through the various phases of data-gathering and analysis to the outcomes. I have made a point of including in the narrative some of those aspects that are crucial to understanding what doing research involves, yet virtually never

150

appear in published reports of studies (including my own). These are such things as theoretical and logistical puzzles, problems, mistakes, changes, and compromises—which can be far more demanding than the conceptual exigencies in reaching a successful conclusion. The primary objective was to give the students, albeit vicariously, some appreciation of what it means to put into practice the methodology they had learned about. The secondary objective was to develop more realistic expectations for their own research and resilience in confronting difficulties as they arise. All students will be interviewed at the end of the course, and their responses to this innovation reported in the paper. Any further evidence from their research dissertations (not yet submitted at the time of writing the abstract), will also be reported.

Reference: Roth, Wolff-Michael (2009) Apprenticeship: induction to research through praxis of method, in Garner, M., Wagner, C., & Kawulich, B. (Eds). (2009). Teaching Research Methods in the Social Sciences. London: Ashgate. Seale, Clive (1999) Quality in qualitative research Qualitative Inquiry, 5 / 4, pp. 465-478

151

Gartner, J. (Johannes Kepler University Linz, Institute for Innovation Management) Motivational Effects of Gamified Surveys compared to Standard Online Surveys Gamification, a discipline in which game-design elements are used in a non-game context, has been applied for data collection in scientific surveys. A fully gamified questionnaire based on the MDA (mechanics, dynamics, aesthetics) framework (Hunicke et al., 2004) and the three layers of form design combined with the conceptual foundations and design process according to Harms et al. (2014) has been developed.

A double cycle quasi-experimental study has been deployed, where participants have been allowed to choose between a gamified- and a traditional online-survey in the first round. In the second round participants had to fill in the other survey. Both approaches were compared regarding their motivational effects such as participation, drop outs, invested time and missing values as well as differences in answers on an ethically sensitive topic about students’ cheating behavior. The dimensions covered in both surveys are based on the theory of planned behavior (Ajzen 1985, 1991) applied to ethical decisions of students. The results show some significantly (partly long term) motivational effects on the voluntary gamified group in addition to significant de-motivational effects on the “forced” groups as well as some minor differences in the given answers.

Keywords: gamified survey, theory of planned behavior, ethical decisions of students

152

Gebel, T., Hense, A. and Schork, F. (Bielefeld University; Sociological Research Institute Goettingen (SOFI); Institute for Employment Research (IAB), Germany) How to deal with convergent, complementary, and divergent findings in mixed methods research? Layoffs and subsequent reemployments by the same employer (“recalls”) are seen as a discontinuous employment relationship that has hardly been investigated. We implemented a sequential mixed methods design to examine the determinants and consequences of recalls. The design combined qualitative and quantitative approaches as well as secondary analyses and field research. Firstly, we analyzed survey data. Secondly, we conducted narrative interviews with recalled employees and semi-structured expert interviews with employers. Third, we implemented a quantitative survey among employees. On the one hand, the design can be described as a complementary design because quantitative and qualitative analyses dealt with the same phenomenon and research questions but used different methods to get different insights. On the other hand, it can be seen as a development design as results of our secondary analyses were used for the sampling of our qualitative interviews, and findings of these interviews were used for the construction of a quantitative questionnaire.

First, secondary analyses of the German Socio-Economic Panel Study (SOEP) and data from the German Institute for Employment Research (SIAB, LIAB) delivered results on individual and operational determinants of recalls and their consequences. We used these mass and panel data for causal explanation based on probabilistic inference and for information about the distribution of recalls in certain groups, industries, and regions. We identified causal links between variables using panel regression analyses, but were not able to examine the processes that explain why variables are linked together. Furthermore, quantitative research is restricted to the variables in the dataset and our knowledge/assumptions about important factors. Thus, we secondly conducted expert interviews within companies that make use of recalls and narrative interviews with recalled employees to gain further insights into their rationale, appraisals, and practices. The SOEP was used to obtain access to recalled employees within different contrast groups, and IAB’s datasets were used for the sampling of the expert

153

interviews. Purposive sampling based on the results of the secondary analyses was combined with theoretical sampling based on the qualitative analyses. The qualitative findings were used to develop the questionnaire for our own quantitative survey. The third component of our research was a standardized telephone survey of employees. This was used to analyze the statistical effect of different determinants and outcomes of recalls which had been discovered during the qualitative research.

The results partly converge or complement one another. The presentation will give three examples how the joined analysis enhances the interpretation of recalls: The transaction cost approach could be corroborated, the assumptions about the decision process had to be modified, and unexpected effects of recalls on social recognition could be found. Furthermore, the presentation will discuss three methodological aspects: Causal analysis based on probabilistic inference can benefit from qualitative research to identify causal mechanisms. The problem of unobserved heterogeneity will be moderated using mixed methods research. Purposive sampling based on survey data should be followed by theoretical sampling based on the qualitative findings to fully exploit the potential of qualitative analyses.

Keywords: complementary mixed methods design, development design, meta inferences

154

Giehl, C. and Mayerl, J. (Technische Universität Kaiserslautern, Germany) Explaining systematic measurement errors using dual-process models of response behavior and the attitude towards surveys Dual-process models of response behavior distinguish between two groups of respondents: those giving answers based on simple decision heuristics and automatic-spontaneous cognitive processes, which is typically associated with short response latencies, and those giving answers based on deliberative thoughts, which is associated with long response latencies. Empirical studies show that both groups are susceptible to different types of response effects such as acquiescence effects for quick responders or contrast effects of question order for slow responders. Furthermore, there is evidence that the attitude towards surveys in general as well as the accessibility of the attitude towards the study case influences the occurrence of response effects.

Therefore, we propose that there is a general association between specific types of response effects, the response elaboration and the respondents’ attitudes. Based on a dual-process model of response behavior, we suggest that respondents with a higher level of motivation and the opportunity to deliberate before answering during a survey are more likely to answer based on well-considered thoughts and therefore tend towards slower response latencies. Furthermore, slow response latencies seems to be a necessary condition for response effects which demand higher levels of elaboration like contrast effects of question order, especially for responders with a negative attitude towards surveys. On the other hand, respondents with lower levels of motivation and/or no opportunity to deliberate should tend to finish surveys as quickly as possible giving answers based on automatic-spontaneous cognitive processes. We assume that those respondents who give automatic-spontaneous answers without chronic accessibility towards an attitude of the research case are those who will most likely be affected by response effects which demand lower levels of elaboration like the acquiescence effect. Thus, we propose that a respondent’s attitude towards surveys and the degree of chronic accessibility of the research case are both predictors for specific types of response effects.

155

According to dual process models, we expect different response effects within the following groups of respondents with different cognitive processes: a) respondents with high levels of motivation/opportunity and a positive attitude towards surveys, b) respondents with high levels of motivation/opportunity and a negative attitude towards surveys c) respondents with low levels of motivation/opportunity and with chronic accessibility of the study case, and c) respondents with low levels of motivation/opportunity and without chronic accessibility of the study case. To examine this assumption, we investigate the link between the general attitude towards surveys, the attitude towards the study case, the level of answer elaboration and the occurrence of response effects (in particular the acquiescence effect, the assimilation and contrast effect of question order and effects involving socially desirable response behavior) to explain method effects according to dual-process models. For this examination, we will use the data of 1.) a paper and pencil evaluation project conducted 2014 to 2016; 2.) a web survey among students in 2016 and 3.) a German longitudinal mixed mode study, the GESIS panel. Early findings, tested with multiple group SEM, using data from our web survey and the evaluation project, will be presented.

Keywords: Response bias, Response latency, Attitudes

156

Gilbert, E., Stella, C., Fisher, K., Calderwood, L., Cleary, A., Huskinson, T. and Gershuny J. (Centre for Longitudinal Studies) A mixed-mode approach to measuring young people’s time-use in the UK Millennium Cohort Study Time diary data provide a comprehensive and sequential account of daily life and are used for a wide range of analytic purposes. Recent years have witnessed a steady growth of large-scale time diary data collection in cross-sectional as well as longitudinal surveys, driven by the increased research interest in population activity patterns and their relationship with long-term outcomes. The majority of social surveys collect paper-administered diaries, which have been shown to produce the most accurate and reliable daily activity estimates, but present challenges relating to respondent burden and administration costs. The use of new technologies for data collection could address these weaknesses by providing less burdensome diary instruments, improving data quality, and simplifying post-fieldwork data coding costs.

The Millennium Cohort Study (MCS) is the first large-scale longitudinal survey that uses a mixed- mode approach for the collection of time-use data among teenagers. A web diary and a smartphone app were specifically designed for the sixth sweep of the survey, when cohort members were aged 14. Cohort members were asked to choose their mode of preference, while a paper alternative was offered to those who did not have a smartphone/ Internet access. Participants were asked to complete a weekday and a weekend pre-coded time diary, and to wear an accelerometer during the two designated days. This presentation will focus on issues surrounding research design and implementation, and present the MCS online diary and smartphone app. We will also report findings from the MCS pilot survey and the Dress Rehearsal that were conducted in 2014: findings on take-up of mode, and data quality across different modes, will also be presented, and methodological challenges will be discussed.

157

Girkontaitė, A. (Vilnius University) Agency-structure dialectics in discourse analysis: networks of discourses Discourse analysis, based on either linguistics or cultural studies, poses two problems: (1) a lack of connections between micro and macro levels; (although various researchers have addressed this problem recently) (2) the emphasis on structures dismisses agency. In this paper I address the second one, because in the research of conflicting discourses it is important to raise the question what social groups stands for which discursive positions. This problem requires elaborated methodologies and methods to account for it. I discuss how some of discourse analysts suggest to deal with this problem (linguistics, Foucault, CDA, Hajer and SKAD), elaborate possibilities of implementing Bourdieu's theory on logic of practice and symbolic power of language and suggest my research program accounting for the question of agency in the research of orders of discourses.

In discourse analysis some theories pay considerable attention to the question of agency, some ignores it, some suggest clear research strategies to capture it, some are quite obscure in methodical level. We could summarize it as follows: (1) no concern with an agent / a speaker, because the main focus are structures of language (linguistics; with the exception of sociolinguistics); (2) agents are seen as more or less passive carriers of a discursive orders almost without the possibility of agency (Foucault (although in “Subject and Power” Foucault suggests some form of subjectivity), Gee); (3) agency and (discursive) structure are seen as dialectical, influencing each other (Keller); (4) agent is seen as the one with the power or the one who should be empowered, although no research strategies or suggestions for implementing this is presented (the most of CDA: Fairclough, van Dijk, Wodak); (5) the centre of analysis are practices of actors, not the discourse in particular (Hajer, Leipold and Winkel).

Assuming, that the discourses are impossible without agents, but considering the importance to account for structural limitations and also to see an agent as historical and embedded within his or her structure, the dialectical approach is taken in this paper. Keller suggests importing Giddens' and Bourdieu's approach to discourse analysis.

158

Following Bourdieu's critical thought and relational approach and CDA's idea of empowering, I suggest a new methodological tool. Assumptions are taken about the importance of agency, social structures, critical evaluation of status quo, but still focusing on language as a power tool and discourse as a structure, reproduced (or challenged) by social agents. Implementing Hajer's ideas on discourse coalitions, Fillipakou's suggestions of network research, Fairclough's stress on genre differences, Keller's concern with qualitative approach, but not limiting to it, the suggested research program consists of the exploratory qualitative research, coding according to main categories, mapping of the main concepts, story-lines, genres, agents and interpreting this network from a qualitative standpoint. This would help to see main discursive positions and their associations with each other.

Theoretical reflections on question of agency in discourse analysis and examples of empirical cases on higher education discourses in Lithuania, based on suggested research program, will be presented.

Keywords: discourse analysis, agency, structure

159

Gollini, I. and Murphy, B. (Birkbeck, University of London; University College Dublin) Latent variable models for multiple network views Network data arise in a wide range of applications including social and biological sciences. In many cases different relations are available. Nodal attributes and links are often in strong relation.

We introduce a new framework of latent variable network models which combine the information given by heterogenous relational network structures in order to analyse, identify latent traits or groups, visualise network data and predict missing links and nodes.

The likelihood functions of the models proposed cannot be evaluated analytically. To overcome this problem, we adopt a variational approach to estimation which turns out to improve considerably the computational efficiency without a significant loss of accuracy with respect to other existing methods.

The effectiveness of this methodology is demonstrated on the analysis of a wide variety of networks.

The analysis is carried out with the lvm4net package for R (available on CRAN).

160

Goodwin, J. and O’Connor, H. (University of Leicester, England) Jephcott's Unpublished Notebooks as Biographical Method In this paper we explore the methodological implications of using unpublished notebooks in sociological research. We focus on the life and career of Pearl Jephcott as a researcher whose contribution to sociology, and the sociologies of youth and community in particular, is suitable for reassessment. Her books - Girls Growing Up (1942), Rising Twenty (1948) Some Young People (1954) Married Women Working (1962), A Troubled Area: Notes on Notting Hill (1964), Time of One’s Own (1967) and Homes in High Flats (1971) – were both formative for many themes within sociology of the 1940s, 1950s and 1960s, as well have having contemporary relevance despite being largely forgotten. Focusing on two books in particular - Time of One’s Own (1967) and Homes in High Flats (1971) and data collected over a ten-year period from various archives and Universities, we discuss (i) Jephcott’s sociological practice based on ‘reality congruent’ theory developed in order that she may cast a lens on the realities of working class life. She wrote richly detailed studies that offered an unsentimental reflection of ‘lived realities’, of the ordinary, of the mundane, of the quotidian; and (ii) Jephcott’s methodological innovation and pluralism in the use of text, image and non-standard data sources that mark her out as being sociologically ‘ahead of her time’.

161

Górnicka, B. (University College Dublin) Nudism, homines aperti and the methodology behind my study of emotions Ever since the role of studying emotions in sociology and its methodology has been on the rise during the past decade, it has become more important to draw attention to the role of researcher’s emotions during fieldwork. In this paper I would like to present two outlooks: one being a theoretical outline of conceptualisation and advantages of Norbert Elias’s theory of involvement and detachment in relation to the studies of emotions; and second one, which offers an account on how this theory works in practice, in other words how useful it has been during my fieldwork to use my own experiences from the time I attended an Irish naturist club. I will argue here how valuable a strategic use of researcher’s own emotions can be for understanding the complexities behind emotions such as shame and embarrassment in relation to nakedness.

162

Graham, S. and Ravn, T. (Centre for Family Research, University of Cambridge, England; Department of Political Science and Government, Aarhus University) Ideals versus realities: exploring methodological influences on eliciting narratives of solo mothers Drawing upon two case studies on solo motherhood, one in Denmark and one in the UK, that used complementary but distinct novel in-depth qualitative methodology, this paper explores how differences in study findings can relate to methodological differences in the research process rather than just varying cultural ideologies and narratives.

Using biographical narrative interviews as a key strategy of inquiry, the Danish case study explores how solo mothers made sense of their personal experiences in relation to identity construction, relatedness and medically assisted reproduction (MAR) when drawing upon, negotiating and transforming specific socio-cultural narratives. The UK study used multiple in- depth interviews with women thinking about and embarking upon fertility treatment with donor sperm in order to follow them through the process of becoming a solo mother. It too used the narratives of the solo mothers as a lens through which to explore, not only the local moral worlds of these women, but also the wider socio-cultural narratives of solo motherhood in the UK. Both studies’ methodological approaches allowed for an exploration of a temporal aspect to the narrative of solo motherhood, highlighting the nuances and connectedness of decision- making and meaning-making, as well as how the construction and attachment of meaning may change over time. In this paper we show how through the process of becoming a solo mother, including the choice of, and meaning attached to, the reproductive technologies and donor gametes used to achieve motherhood, these women constantly renegotiate their ideals versus realities of family building and relatedness. At the same time we will show how our own methodological choices, the limits and constraints to data collection, and our own role as a research instrument may have affected our data analysis and consequently the study findings. In doing so we will examine the strengths and weaknesses of the two methodological approaches used in eliciting the temporal aspect of solo motherhood, and discuss how differences in results may be tracked to the differences in methods utilised. Elucidating the ideals versus the realities

163

of these novel methodological approaches we ask to what extent these methods can be useful in exploring the sensitive and nuanced personal narratives that are part of the meaning making surrounding reproductive technologies and new family forms?

Keywords: solo mother narratives, reproductive technologies, methodological comparison

164

Graeff, P. and Fuchs, S. (University of Kiel, Germany) Bridging the historical link between corruption and social trust: reproducing a contemporary effect with historical data Recent studies in macro-sociological research have shown that the absence of corruption in a country is positively high correlated with the presence of social trust (Serritzlew et al. forthcoming; Graeff & Svendsen 2013; Bjørnskov 2010). If citizens act with a low level of corruption, it will be easier to trust others, e.g. strangers. In turn, if people interact with others openly, it might prevent clandestine, illegal actions.

In this study, we are going to ask whether this solid result derived with contemporary data also exist in historical contexts (Kousser 1980). Methodologically, this poses a challenge as historical data are scarce (e.g. only a small number of cases exist) and direct measures of corruption and social trust are not available. We show that there are proxies for social trust and corruption for which contemporary and historical data exist. A certain methodological challenge is the application of official figures for corruption, which are only partly related to the prevalence of corrupt practices. Since survey data for corruption and social trust are not available for historical years, one has, however, to pick up official figures and/or other statistics such as data from clubs (such as choirs) which implicate certain methodological difficulties per se.

In order to reproduce the link between corruption and social trust in history, we apply a mixed method design by which historical data for corruption and trust indicators are considered for their historical fit. By this, we are able to replicate the link between corruption and social trust with contemporary data for historic years of 1885-1890 in specific German regions. This finding implicate that both social trust and corruption might have historical roots which shape the social conditions in a society in the long run (Becker et al. 2011).

References: Becker, Sascha O., Boeckh, Katrin, Hainz, Christa & Woessmann, Ludger (2011). The Empire Is Dead, Long Live the Empire! Long-Run Persistence of Trust and Corruption in the Bureaucracy. IZA Discussion Paper No. 5584, Bonn.

165

Bjørnskov, C. (2010). “How does social trust lead to better governance? An attempt to separate electoral and bureaucratic mechanisms.” Public Choice 144, 323-346. Graeff, P. and Svendsen, G. T. (2013), “Trust and corruption: The influence of positive and negative social capital on the economic development in the European Union”, Quality and Quantity. In Press. Kousser, J. Morgan (1980). “Quantitative Social Scientific History.” In: Michael Kammen, ed., The Past Before Us: Contemporary Historical Writing in the U.S. Itheca : Cornell University Press, 433- 56. Serritzlew, S., K.M. Sønderskov and G.T. Svendsen (2014), “Do Corruption and Social Trust Affect Economic Growth? A Review”. Journal of Comparative Policy Analysis 2, 121-139. Key Words: Process-Oriented Methodology – Historical Analysis – Spatial Analysis – Corruption– Trust

166

Granda, P. (Inter-university Consortium for Political and Social Research (ICPSR), University of Michigan, US) Creating Structured Metadata: Quality Profiles and Ex-Post Data Harmonization With ongoing survey research now more than fifty years old, more social science researchers have increasingly focused on merging data from existing surveys to create new resources for the analysis of important social and economic conditions over time or across nations and cultures. The process of transforming existing data, not originally intended to be comparative, into a single new resource is completed through an approach called “ex-post harmonization”. Variables of interest from all of the original surveys are carefully studied; similar or identical questions are sought; and strategies are prepared to create comparable indicators. Large amounts of metadata, often unstructured, must be reviewed and used in the harmonization process.

This presentation will describe the process of generating structured metadata from ten cross- sectional surveys that focused on reproductive health, contraceptive use, pregnancy, and other issues vital to understanding and tracking the health of the U.S. population over five decades starting in 1955. It will describe the emphasis placed on creating variable-level metadata to perform ex-post harmonization of these surveys but also argue that this is only one side of the greater need to standardize all aspects of the data and documentation life cycle in order to adequately address issues relating to data quality.

Examples from the harmonized data set, the Integrated Fertility Survey Series (IFSS), will illustrate the creation of new types of structured metadata using the DDI Lifecycle model to adequately document the harmonization process, how quality judgements can be prepared and rendered in DDI, and advocate the need to prepare a quality ‘profile’ to alert users to the potential strengths and weaknesses of the harmonized data. Finally, it will address the topic of how the entire transformation of source materials into target materials, the key to produce harmonization resources, is related to the Total Survey Error (TSE) paradigm.

Keywords: Metadata; Harmonisation; Total Survey Error

167

Green, J. and Binsardi, B. (Glyndwr University) Frontier Rescinding in Classic Grounded Theory Classic grounded theory produces objections, threatens traditions and voices fears in the academic arena precisely because the outcomes of using the stand-alone methodology, which is a further academic concern, frequently cross recognized bodies of knowledge. Properties and categories from multiple disciplines may emerge as generalized others influence and direct the empirical data.

The nature of these issues has the potential to become problematic for researchers seeking positive reviews and, therefore, investigation is warranted as this is a clearly identified methodological problem. The findings provide a means of shedding new light on ways in which this embedded viewpoint can be addressed. There is value for academics, in particular doctoral supervisors and examiners, and also reviewers of conference papers and journal articles in undertaking this investigation.

In order to provide a framework for the paper, the bodies of knowledge will be limited to business and management, a subject which struggles to keep abreast of the changes occurring in practice. Increasingly there is an over-lapping of traditional practices as short-termism, reduced life cycles and technological leaps have increasing prominence and impact, specifically the illustrative problem is strategic communication.

The methodological problem is further exacerbated as grounded theory is conceptual, rather than descriptive which leads the researcher into diverse areas in order to provide concern resolving proposals for the identified problem, subsequently supported by relevant literature.

This paper draws on the concept of a basic social process that has a minimum of two distinct, or separate, emergent stages and occurs over time. There is a close operational fit between the basic social process and the patterns of behaviour evident within the social problem that is being researched.

168

Imagery and movement are integral aspects of a basic social process; however require clarity and may be retrospective; the movement is contingent upon contemporary events and settings. A good basic social process has high impact meaning and should emerge within grounded theory practice to discover what is actually happening. Problems are complex with foundations in interactional contexts, negotiation and development, and shape human behaviour; they do not always fit neatly under a precise category heading. A clear stance regarding epistemology and the researcher position as the active analyst undertaking practical ways of discovering concern resolving behaviours and the social construction of knowledge also require clarification.

Closure of the identified gap is to utilise the notion of the venatic paradigm which has a close fit with classic grounded theory techniques. The foundations of the venatic emphasise qualitative and inferential thought. Historically this approach can be traced back to Aristotle and has three main strands; metaphor where the meaning of words are applied in a different sense; antithesis or a figure of speech and finally energeia, dwelling to produce effective expression. The outcome will depict strong inter-dependencies, or saturation, which is sought after in the generation of a grounded theory.

The result of considering two approaches is to produce a synthesis of ideas drawn from classic grounded theory and the application of venatic practices as a support mechanism to strengthen the view that working across bodies of knowledge should be encouraged and assists in the practice of conceptualization, a further requirement of classic grounded theory.

The findings will demonstrate that, as classic ground theory research is inductive, it is carried out in a sequence which differs from the majority of other methodologies in that data is collected and undergoes initial analysis in advance of searching the literature. This approach enables the generation of new theory as opposed to the verification of previous works and therefore the researcher is unable to anticipate the direction in which the study will take. Clarification of how classic grounded theory is structured forms the second section of the paper to illustrate this point.

The structure of the paper emphasises the methodological problem, with supporting contemporary examples and experiences to place the work into context followed by a review of

169

the literature which explains the nature and make-up of a basic social process, within a classic grounded theory setting. To strengthen the argument a similar exercise will be carried out for the venatic paradigm. The conclusions, using theory generation, add to the knowledge related to conceptualization and the effective application of classic grounded theory.

Key words: basic social process, bodies of knowledge, classic grounded theory, conceptualization, figuration, venatic paradigm

170

Gupta, A. (National University of Singapore) Time Versus Depth: challenges to conducting an ethnographic study into “the field” of education Ethnography—a methodology for understanding a culture, or a part of a culture within familiar or unfamiliar context—has emerged as an increasingly popular way of conducting a sociological inquiry. It has been used quite popularly by educational researchers and sociologists of education, for it provides the researchers with sufficient discipline for carrying out the planned elements of research, along with a scope of providing sufficient space to finding out, exploring and understanding the new elements in the field. The ethnographic methodology, despite being used in a multitude of subjects across disciplines, is characterized by a thick description of the culture—which requires a “sensory fieldwork” (O’reilly, 2012). The methods and the processes of carrying out an ethnographic project demand efforts and resources for it entails intense participation and observing, along with routinely reflecting on ethnographers’ experiences in the field. Concurrent debate in ethnographic fieldwork is about producing an ethnographic account in a very limited time period and within scarce resources. Although all the ethnographers at some or other point have faced this challenge during their research, the very nature of allowing different interpretation to the methodology, and acceptability to conducting innovative methods in order to collect the sought-after data, provide an opportunity to gather requisite information through creative means in a shorter period of time.

In this paper, I share my experience of conducting an ethnographic project. Tentatively titled as “Quality imperative amidst the discourse on public-private binaries in school education in India”. This research aimed at understanding the system of school education in an urban context. The fieldwork was conducted from December 2014 to December 2015 in Dehradun. In this fieldwork, schools were one of the most important sites in the field. I conducted guided interviews, participant observation, and informal conversation with the teachers at three type of—public, private and aided—schools. This paper gives a detailed account, and relevant examples, on the question of organizing my fieldwork in the limited time span such that I fulfilled the methodological requirements, while efficiently dealing with arising issues and challenges.

171

The paper argues that in order to achieve depth in the limited time period, it is crucial to organize a fieldwork through careful Planning, Preparation, Reflexivity, and Positionality. In general, the selection of methods varies according to the context and research questions. Educational projects, especially if those are situated in institutional settings, may be crafted in a shorter span of time with mindful understanding and utilization of the above four elements. Planning and preparation include aspects such as making timely arrangement for gaining access, scheduling oneself for necessary and foreseeable meetings, develop guidelines—such that it allows space for unforeseen themes and ideas from the field—for conducting methods such as interviews, group discussion, observation and so on. These aspects entail an effective use of communication with the key players or gatekeepers who provide an access to entering into the field. Reflexivity and positionality are concurrent to the fieldwork, and this requires ethnographers’ reflection and thoughtful actions simultaneous to the fieldwork.

172

Gupta, P. (Suvitam Foundation, India) Key Issues for better Academic Governance and Performance in Indian Higher Education Sector There is big demand for higher education in India. Both Public and Private Institutions in higher education are growing rapidly. Governance impacts performance of these academic Institutions. This paper will throw light on key issues in governance of such Institutions in higher education space in India. Often the measurements, rewards and whole ecosystem is working in such a way that is counter-productive to efficiency and performance. This research will be based on a pilot study, survey and focused group discussions. Key issues and factors will be identified and possible solutions for better governance will be identified. This research will be highly useful for the academic community as well as for the students, researchers and the Indian higher education industry.

Keywords: Academic governance, Higher education, Reward

173

Guth, J. and Nelsen, B. (Furman University, Department of Political Science, USA) The Changing Role of Religion in Public Support for the European Union. In recent years several scholars have considered the role of religion in the politics of the European Union. For decades, the EU’s institutional growth and territorial expansion were strongly supported by Catholic politicians, religious leaders and publics—and were regarded much more skeptically by their Protestant counterparts (Nelsen and Guth 2015; Foret 2015; Leustean 2014)). By the 1990s the continuing secularization of European societies, as well as the greater religious diversity created by the EU’s expansion and immigration, began to change that pattern. More recently, economic and financial crises have also shifted these historic religious configurations, as several Catholic nations—usually a bedrock of popular support for the EU— suffered the most severe consequences of those crises. We use data from the 2009 and 2014 European Parliament Election Studies to examine the ways in which the religious underpinnings of the integration project have shifted under the pressures of recession and fiscal austerity. We also examine many of the limitations of these studies for analysis of the role of religion in the contemporary EU.

References: Foret, François (2015) Religion and Politics in the European Union: The Secular Canopy. Cambridge University Press. Leustean, Lucian (2014) The Ecumenical Movement and the Making of the European Community. Oxford University Press. Nelsen, Brent F. and James L. Guth (2015) Religion and the Struggle for European Union: Confessional Culture and the Limits of Integration. Washington, DC: Georgetown University Press.

Key words: Religion, European Union, parliamentary elections.

174

Haan, M., Toepoel, V. and Lugtig, P. (Utrecht University) Device Use in Online Panel Surveys Online surveys are a popular method for data collection because of the time and cost savings, the ever-improving techno-logical possibilities, and the increase in Internet use in Western countries. Traditionally, respondents participated in these surveys by using a computer or laptop, but nowadays, there are more devices respondents can choose from when filling out an online survey such as tablets and mobile phones.

This paper investigates whether panel members of the German Social Science Infrastructure Services (GESIS) panel have specific device preferences. Several questions on device use were implemented in a wave of the GESIS panel in December 2014. These items included topics such as: device ownership, device preference for survey participation, and possibilities to use the device in combination with the Internet. Panel members could state there device preference for the following online modes: Computer/PC, Laptop, Tablet, and Smartphone. Additionally, they could also select a Paper and Pencil ques-tionnaire if that was their preferred participation mode (GESIS also offers an offline PAPI survey).

First, we present which devices are currently used by the panel members. Second, we are interested in finding out the device preference of the panel members and which panel member characteristics may predict a specific device preference. Third, we use subsequent waves of the GESIS panel to study whether the panel members participate using their preferred device, another device, or that they switch devices between waves.

The findings of this study have important implications for survey research. By using data of multiple waves, it is possible to study if device preferences in general are stable over time, and more specific which panel members participate in their preferred mode and which members may switch devices between waves. Furthermore, information about device prefer-ences may be used for targeting panel members in future waves.

175

Haas, G.-C. (Institute for Employment Research, Germany) How does asking for linkage consent in the beginning of the questionnaire affect data quality? Through combining survey data with administrative data researchers additional research is possible. Usually researchers need the consent of each respondent which shall be linked to do so. Naturally we want to obtain the highest consent rates to have the most possible power. It is best practice asking the linkage consent question at the beginning of the questionnaire (Sakshaug et al. 2013). However, if asking this question at the beginning the linkage consent question may have an impact on respondents’ response behaviour because they may have the feeling of being observed. We hypothesized that this feeling leads to two possible outcomes. First, respondents effort to answer questions increases because they fear consequences if survey data differs from administrative data (better-respondents hypotheses). Second, respondents may become more satisficing in answering questions because they believe all important information about her/him is covered through the administrative data and survey responses can be corrected afterwards. For our analyses we use a German Survey that asked one half of respondents at the beginning and the other half at the end. Since both groups suffer from selection bias, we use the entropy balance technique to simulate a randomized distribution over both groups and calculating average treatment effects on measurement error and item nonresponse. Also we compare the effects for a CATI and web mode. Our results show that no effects in either of these two modes occur, when asking the administrative data linkage question at the beginning of the interview.

176

Hájek, M. (Department of Sociology, Faculty of Social Sciences, Charles University, Czech Republic) Analysing the controversial dimension of public discourse – combining corpus linguistic with L. Boltanski’s approach to the study of public controversies Michel Foucault’s emphasis on “thinking the outside” has been beneficial for sociology in that it weakened the traditional tendency to search for the essence or fundamental qualities of social phenomena and directed attention to the study of the mechanisms and processes in which these phenomena are socially (discursively, materially) constituted. However, in the sociological discourse analysis this resulted in the systematic preference for researching power relations and hidden mechanisms, and, at the same time, in the reduction of semantic dimension of discourse mostly to the thematic analysis. In order to overcome this analytical asymmetry, I use the sociology of knowledge approach to discourse (SKAD) as a suitable framework which is capable to encompass both structural and semantic aspects of discourse. In my contribution I will present the analysis that methodologically combines a formal analysis of discourse with a nontrivial analysis of its semantic (and also pragmatic) dimension. In doing so I will draw on Luc Boltanski’s concept of tests, namely on discursively articulated truth, reality, and existential tests (Boltanski 2011). In face-to-face communication or public controversies social actors invoke the tests in order to confirm or criticize the current state of affairs, depending on whether these comply or are in conflict with shared beliefs or institutionally acknowledged facts. Concurrently, through the application of tests actors make sense of what is happening in the situation and what is justified or not. By observing instances of tests and keywords occurring in the discourse it is therefore possible to capture the discursive semantics (what is or isn’t considered as true or real) as well as the dispersion of statements which eventually forms the object of discourse.

The proposed approach will be demonstrated via the analysis of the controversial public debate in Czech printed and online media that has emerged after the Czech National Bank launched the foreign exchange intervention in November 2013 (thus weakening the exchange rate of the Czech currency Koruna). The declared objective of the intervention was to change the “mindset” of actors in the Czech economy in order to stimulate spending and “to push general inflationary expectations towards the 2% inflation target”. However, the intervention caused an intense

177

public backlash that made apparent how different were the knowledge and expectations of economic experts from the knowledge and expectations of lay people.

References: Boltanski, L. (2011). On critique: A sociology of emancipation. Polity. Keywords: public controversies, discourse analysis, Boltanski’s tests

178

Halford, S. and Savage, M. (University of Southampton & London School of Economics) Symphonic Social Science & the Challenge of Big Data This paper aims to make a critical intervention into the current impasse between the social sciences and the proponents of ‘big data’ in the computational sciences. On one hand, (many) in the social sciences have argued that these new forms of data are of low quality – partial, often of unknown provenance and error strewn, lacking capacity for causal analysis – and that big data analytics fail to incorporate a hermeneutic and critical aspect. As such, it is argued, we must defend the superiority of conventional social science methods of data generation, collection and analysis, at least in analysing the social world. On the other hand, the proponents of ‘big data’ have argued that these new forms of digital data – characterised by hitherto unimaginable volume, variety and velocity – will displace all previous methods of data collection, theories and even disciplines as ‘with enough data the numbers speak for themselves’ (Anderson 2008).

Our starting point is that both positions are deeply flawed. ‘Big data’ is happening and we should make the very best of it. To position the social sciences against ‘big data’ risks leaving social scientists without a foothold in the emerging data assemblages that play an increasingly important part in the production of knowledge and information in the 21st century. Meanwhile it is increasingly clear that taking numbers at face value and by-passing substantive and methodological expertise in analysing the social world can lead to some very unsafe and, indeed, embarrassing conclusions and we see calls from the ‘big data’ field for engagement across the disciplines. As the naivety of early ‘big data’ rhetoric wanes, social scientists have a remarkable opportunity, perhaps even a pressing responsibility to take up the challenge and engage constructively with ‘big data’.

We are under no illusions about the tensions that are involved here. However, rather than rehearse these in this paper, our aim is to practice what we preach and develop a constructive, prospective line of thinking. In a lateral move, we suggest that we might take inspiration from three of the recent (and in fact all-time) best-sellers in social science: Puttnam’s (2000) Bowling Alone, Wilkinson and Pickett’s (2009) The Spirit Level and Pikkety’s (2014) Capital in the 21st

179

Century. Taken together we suggest that these works might offer a new model for social scientists’ engagement with large scale data sets and inform a new means of engagement with ‘big data’.

Specifically, we will argue that these works constitute a ‘symphonic aesthetic’ that focusses on drawing together descriptive assemblages of data drawn from multiple and asymmetric data sources to make claims about deep causal relationships, rather than on the processes of creating new and/or bespoke data sets and modelling the relationships between the (inevitably limited) variables within. In each case, the links between these data sources are worked through the repeated visual analytics rather than the more familiar assemblage of text and numbers.

Whilst this approach challenges recently dominant approaches to data collection, analysis and representation in (much of) the social sciences, it also challenges the technocratic approach to ‘big data’ in the computational sciences. Specifically, the symphonic aesthetic insists on repeated theoretical and conceptual interrogation of data rather than depending on computational tools (data mining, pattern recognition, etc.) to produce technically mediated visualisations of the data.

In the final section of the paper we explore how this approach might be extended more fully into ‘big data’ analytics, with particular consideration of emergent methods of data integration and linkage.

180

Hameister, N., Klaus, D. and Engstler, H. (German Centre of Gerontology) Age and Data Quality? Lessons learned from the German Ageing Survey (DEAS) In this presentation, we will put forward findings for the interrelation between age and data quality, using data from the German Ageing Survey. The German Ageing Survey (DEAS) is a nationwide representative cross-sectional and longitudinal study of the German population aged 40 and older. The first survey was conducted in 1996. Participants are questioned in detail on their objective and subjective living situation, covering a wide range of topics. Methodologically, information is gathered via two modes: a computer-assisted personal interview and a self- administered paper-and-pencil questionnaire. In addition, tests of cognitive capability and physical functioning are applied.

In our presentation, we will concentrate on two main issues: First, the sampling and methodological design allow us to provide results on the interrelation of age and several methodological topics, such as data quality and mode effects. We will present findings on whether and how age correlates with attrition, item non-response, response styles (like acquiescence, midpoint or extreme answering tendencies) and the replication of the factoral structure of selected psychological instruments and scales. Due to the sampling design, the case number of elderly respondents is sufficient and we can provide fine-grained comparisons through several phases of ageing. Furthermore, we aim to explain the mechanism behind the presented age effects: For some issues we can test whether the age effects are rather driven by cohort differences or life-course dynamics.

Second, we analyse the longitudinal accuracy and plausibility of retrospective biographical information: in each panel wave, respondents are asked for the number and demographics of all of their children. This repeated recording allows to test for the reliability of measuring biographic events and to analyse whether intra-personal variance rises with age.

Keywords: German Ageing Survey (DEAS), longitudinal reliability, data quality

181

Hanna, E. (Centre for Men’s Health, Leeds Beckett University) Researching male infertility online: Bringing researcher emotions into the debate The growth of online research methods for a variety of health related topics, offers an important and interesting way in which naturalistic discussions around sensitive and emotive topics can be accessed. Such methods may have particular relevance for aspects relating to men’s fertility, which can be seen to be stigmatised, sensitive, and topics which may be pragmatically challenging for the researcher to otherwise access. Whilst the parameters around how to ethically research online are now becoming more consolidated within the social sciences, as with other aspects of qualitative research, the emotional implications for the researcher is an area which has received less attention in relation to online research, and particularly in relation to reproductive research online.

This paper then draws together the perspective of the researcher in online research in relation to researching male infertility. Drawing on a qualitative research project around emotional sharing and help seeking by men on a UK men only online discussion forum, this paper then explores the intersection of the researcher with those experiences men documented online. During the course of our research, the emotional impact of the use of online methods was identified and carefully considered. Distress we argue is often highly visible within the context of delayed or challenging conception journeys and the ‘online’ researcher is exposed to this distress, often absorbing the emotive narratives during the course of online analysis in a manner not dissimilar to reading a harrowing testimony; powerful narratives are often slow to leave our memories.

This paper then explores three key themes around the emotional implications for researchers about online forum research around reproduction.

These themes are:

• The emotional implications of the question of ‘what happened next?’ particularly in relation to the often unresolved story of the male fertility journey

182

• The challenge of empathising but not being able to convey such empathy within a ‘one way’ research relationship

• Challenging the notion that online research involves detachment from the participants and their experiences and that researchers can be emotionally impacted by researching online

Drawing together these themes this paper argues that although online research can sometimes be portrayed as an ‘easy’ option in terms of accessing naturalistic accounts, the experience of conducting such research around fertility is not always emotionally ‘easy’. Researchers we argue should adopt methods to protect their own wellbeing, utilising debriefing and supervisory approaches, in order to ‘work through’ the feelings they may have about the emotional impact of their online research. Emotions in research are not inherently a problematic aspect, emotions are part of human life and thus an important means for understanding the lives of social actors, thus we do not seek to argue for detachment, or emotional restraint, rather argue that researchers should adopt a careful consideration of why online narratives may evoke powerful feelings, as these emotions may be a useful tool for further appreciation of the challenging reproductive experiences of those whose lives we seek to understand.

Keywords: Male infertility, Online/digital research, Researcher emotions

183

Haken, M. (Technical University Berlin, Germany) Videography of audience emotions in religious events Since the method of video analysis was extended to the method(ology) of videography by Tuma, Schnettler and Knoblauch (2013) it emerged that it is quite a fruitful instrument for nearly every realm of social research, because it allows analyzing communicative actions in their natural context. Nevertheless, there are some challenges in capturing sequentiality of social actions and interactions for that method(ology). The contribution aims to expose methodical and methodological challenges concerning the videography of emotions of audiences in religious events. First and foremost I want to reflect briefly the challenges we are confronted with, if we want to investigate collective emotions videographically and beyond that in the field of religion, whose exploration entails itself a lot of problems. I assume that specific problems concerning the study of collective emotions and the problems in the field of religion each other apparently reinforce. Thus we are confronted with a "double" challenge in the exploration of emotions of audiences in religious events. The advantage of Videography, that gestures, facial expressions, as well as glances are captured and preserved for later analysis, raises in events with several thousand visitors the question of choosing the adequate perspective. The second aspect I want to focus in that contribution is the camera action. The influence of the researcher’s personal attitude towards - in this case - religion on the camera action cannot be prevented despite methodological agnosticism. But how do we deal with it as good as possible? The videography requires experiencing the recording in the field - we can say synesthetic. Taking it seriously, we are confronted with the problem that if 2000 people adoringly worship Jesus or 70000 in the Olympic Stadion of Berlin cheer Benedikt XVI., because of the blessing of infants, emotions probably carry one of easier if the researcher is receptive to the content. How to take on distance to the subject (collective emotions), if the method requires to be part with all your senses? Based on datas collected from a project run out in 2013 and an ongoing project about “audience emotions in sport and religion”, I want to discuss these challenges for the method(ology) of videography.

Keywords: collective emotions, camera action, methodological agnosticism

184

Hansell, A. (Imperial College London) Spatial analysis of administrative datasets at small area level The UK Small Area Health Statistics Unit (SAHSU) was set up in 1987 following discovery of a cluster of cases of leukaemia and lymphoma near a nuclear power plant. It has a remit to be able to rapidly investigate administrative data in response to concerns about environmental contamination. SAHSU maintains a database of over 300 million patient records at postcode and address level on births, deaths, cancer registrations and hospital visits using Office for National Statistics, Public Health England and NHS data.

Examples of SAHSU studies are investigation of birth outcomes near all 22 municipal waste incinerators in the UK, an investigation of cardiovascular hospital admissions and mortality in relation to aircraft noise contours near Heathrow airport and an interactive online atlas showing smoothed risks of 14 major health conditions and exposures to five environmental agents at small area level searchable by postcode.

There are a number of key challenges in small area spatial analysis: (1) Use of routine administrative data. Careful interpretation is needed. Quality issues may become particularly apparent at small area geography, resulting in ‘mapping the holes’ rather than the hotspots. (2) Population denominator data. This is often from the national Census. Detailed breakdown by age-sex-ethnicity in small areas is not released from the UK Census. SAHSU work with unpublished Census data to construct synthetic population estimates for these denominator data. Plans to move to a limited national Census supplemented by survey data in future years may further impact on validity of small area population figures. (3) Geographical issues. Geographies may differ across administrative datasets. Different harmonization methods may be used including population weighting. The geographical scale used may artefactually affect whether associations are detected or not – the Modifiable Area Unit Problem (MAUP). Geocoding from address is not 100% successful and may result in errors in counts that may be particularly important at small area level. Postcodes change (and change

185

location) over time and the area covered is larger in rural than urban areas. (4) Adjustment for deprivation. Choice of deprivation measure may affect results. In environmental analyses deprivation may be closely associated with the environmental pollution. Adjustment for confounding effects of deprivation may then reduce or remove observed associations with the exposure. (5) Statistical methods. Small area analyses usually involve small numbers, therefore unstable rates. SAHSU make extensive use of Bayesian methods especially in mapping to borrow strength from surrounding areas. A number of methods are used to display statistical uncertainty in maps. (6) Maintaining confidentiality for sensitive data. Good information governance procedures are important and care taken to maintain confidentiality of sensitive data where small numbers in small areas are involved, including in mapping outputs.

To help rapidly investigate questions about possible disease clusters including in relation to industrial installations, SAHSU have developed a Rapid Inquiry Facility (RIF). This integrates health, social (deprivation, ethnicity) and exposure data in a geographical framework and allows mapping and risk analysis at chosen small area level. It incorporates decisions made in response to the key challenges listed above. The RIF is currently undergoing a substantial revision using open source software. Design features include an intuitive interface and a high degree of choice of study parameters. Visualisation is a particularly important component as this may influence perception of a problem and subsequent actions

186

Hering, L. (TU Berlin) How to trace the Intrinsic Logic of Cities using historical material The idea of the Intrinsic Logic of Cities tells us that cities cultivate their own specific constellations of knowledge and expression. This quite new concept in the field of urban sociology was established in 2005 by Helmuth Berking and Martina Löw. Based on the assumption that the local economy forms at least one part of the urban Instrinsic Logic I construct the space of local economy as a figuration that is developed processual and depending on the hidden structures enscripted in the place and its peoples’ (daily) practices over time. Comparable with an imprint one can track the conditions that pre-structure the figuration from a daily (as well as a) historical perspective. The aim of the presentation is to show how I develop a strategy to research the Intrinsic Logic of the City from a historic perspective (as a long-term process) by combining a macro analysis on the economic development of the city (city profile) with a micro analysis on the daily economic practices of the individuals. When researching past social processes one has to consider several questions such as how to : - define the research object when it is changing over time (e.g. city boarders), - delimit the time period that will be researched (whole time phases or specific periods of upheaval), - get access to the archival material and then make a selection of useful data, - connect findings on the macro level with those on the micro level. I will discuss these problems by presenting first findings from my PhD project on the Intrinsic Logic of Birmingham bringing forward the argument that more or less stable patterns can be identified which can be interpreted as evidence that the Intrinsic Logic of Birmingham exist and that it is a complex long-term process.

187

Hlebec, V. (University of Ljubljana, Faculty of Social Sciences, Slovenia) Secondaru Quatitative Data in Explanatory Mixed Methods Research Design This presentation focuses on methodological dilemmas that occur while using secondary quantitative data (e.g. survey data such as European Social Survey) in studies with mixed methods research designs. The research design we focus on is the research design with sequential use of quantitative and qualitative methods – explanatory participant-selection research design (Creswell and Plano, 2007). We systematically evaluated the role of secondary quantitative data in methodological dimensions of two-stage mixed methods research design. More specifically: the research questions we addressed in the study were:

- How the costs for conducting the research are modified by usage of secondary data?

- How data access is handled in this research design?

- How quality of secondary data sets affect the research design?

- In what ways the time between quantitative and qualitative phases affects data quality and results?

- What challenges are posed to sampling in this two phased research design and how researchers are solving them (e.g. selection of participants for the second, qualitative, phase; time lag between phases)?

- What issues may arise concerning the quality of indicators (variables) that are provided in secondary data sets and how this affects research questions and quality of research findings?

We conducted systematic literature search in several steps: First, by keywords (mixed methods, mixed methods designs, explanatory design, participant selection model, secondary quantitative data, quantitative data, qualitative data, survey, interviews, focus groups...). Second, with various search engines (Google Scholar (scholar.google.com), Sage Journals online (online.sagepub.com), SpringerLink (www.springerlink.com), ICPSR – Inter-University Consortium for Political and Social Research (icpsr.umich.edu)). Third, among references of already found articles and articles on meta-analysis of mixed methods. Altogether 149 articles were obtained using a number of

188

enquires by various combinations of the keywords. Of those, 69 were eliminated because the different mixed methods design was used, 53 were eliminated because of insufficient information about the study. For our metaanalysis we used 27 articles (6 were found with Google Scholar, 15 were found on Sage Journals, 2 were found on SpringerLink, 2 were found on ICPSR). While reviewing these articles we managed to find another three articles that met all our requirements. Together with our article (Hlebec et al., 2010) we are dealing with 31 published articles.

With meta-analysis of studies, we critically evaluate methodological dilemmas related to use of secondary quantitative data and most common solutions of these dilemmas. Restrictions of secondary quantitative data may exist to a greater extend, but this does not reduce the useful value of secondary quantitative data. Regardless, we recommend that researchers evaluate restrictions of the use of secondary data in explanatory mixed methods research designs and thus ensure and retain the methodological validity of chosen solutions.

Keywords: explanatory design, secondary data, meta analysis

189

Hlebec, V., Kogovšek, T., Manfreda, K. J. and Lobe B. (University of Ljubljana, Faculty of Social Sciences) Usability of Offline and Online Focus Groups for Pretesting Sensitive Questionnaires Pretesting of survey questionnaires takes place in several steps ranging from early stage qualitative pretesting in cognitive laboratories with qualitative techniques such as expert appraisal, eye-tracking, cognitive interviews or focus groups to later stage quantitative pilot studies or split ballot experiments in the field. The purpose of our presentation is to explore usability of offline and online focus groups for pretesting wordings in sensitive questionnaires via qualitative meta-analysis of 42 offline and online focus groups conducted in November and December 2014. The following topics will be systematically and synthetically compared:

- design issues (recruitments of participants, structure, size and composition of the groups),

- moderating issues (ability to prepare, relax and motivate participants, probing responses, nonverbal interaction, following up responses, restoring off-topic interaction, managing overly active exchanges, minimizing distractions, avoiding departures),

- quantity and quality of output (number of unique ideas, relevant/off-topic comments, amount of disclosure, role of nonverbal communication, group dynamics),

- quality issues (balance of time, participant effort , moderator effort, and outputs), and

- sensitivity issues with regards to all stages (recruitment, participation, moderation, output).

Keywords: Focus groups, pretesting survey questionnaires, sensitive topics

190

Höfig, C. (Research Associate at the Bundeswehr Center for Military History and Social Sciences (ZMSBw); University of Potsdam) Different target groups, same error? - Investigating the robustness of mode differences regarding item nonresponse The presentation examines and discusses mode effects on item nonresponse in surveys about political attitudes towards security and defense policy, which are assumed to be sensi-tive topics. Therefore, surveys conducted with less interviewer contact should produce less item nonresponse.

In order to examine mode differences, a large-scale method experiment was conducted in two concurrent surveys (CATI and CAPI) with probability samples of the German population. Both surveys included the same 44 items with varying content, stimuli, response labels and number of response options (split-ballot: 4-, 5- and 7-point scale). Hence, the robustness of the effect can be assessed.

Preliminary empirical results show robust mode differences in item nonresponse rates for different item types. In addition, findings indicate that mode effects vary depending on respondents characteristics. Findings and implications for survey design will be discussed.

Keywords: survey mode, item nonresponse, respondents

191

Höglinger, M. and Diekmann A. (ETH Zurich, Clausiusstrasse 50, 8092 Zurich, Switzerland) Do special techniques for surveying sensitive topics provide valid measurement? A validation design that detects false positives The valid measurement of sensitive issues such as norm-violations or stigmatizing traits through self-reports in surveys is often problematic. Validation studies show, that a considerable share of respondents falsely deny sensitive behavior when asked about such in surveys. In the best case, sensitive behavior is then just underestimated using such data, in the worst case, conclusions about correlates and causes of the sensitive behavior in question are plain wrong. Despite this serious flaw, research in deviance, epidemiology, and many other areas relies heavily on self- report data. Finding ways to achieve a valid measurement of sensitive items is, therefore, very important.

Special sensitive question techniques that use some randomization procedure to introduce systematic noise into the response process such as the Randomized Response Technique (RRT, Warner 1965) and among its variants the recent Crosswise Model RRT (CM, Yu, Tian and Tang 2008) should generate more honest answers by granting respondents full response privacy. Full response privacy means, that there is no possibility to infer from a single respondent’s response her actual answer to the sensitive question. In turn, respondents are supposed to answer more honestly and the validity of the self-reports should increases. However, the RRT sometimes does not work as expected and evaluating whether particular implementations actually improve data validity is essential before their application in substantive surveys.

Different types of validation studies have examined whether particular techniques actually improve data validity, with varying results. Yet, most of these studies did not consider the possibility of false positives, i.e. that respondents are misclassified as having a sensitive trait even though they actually do not. Assuming that respondents only falsely deny but never falsely admit possessing a sensitive trait or behavior, higher prevalence estimates have typically been interpreted as more valid estimates. If false positives occur, however, conclusions drawn under this assumption might be misleading.

192

We present an easy-to-apply comparative validation design that is able to detect systematic false positives without the need for an individual-level validation criterion – which is often unavailable. This is achieved by introducing one or more zero-prevalence items among the sensitive items. If a sensitive question technique systematically leads to false positives, the estimates of the zero- prevalence items will be non-zero. Results from its application in a survey on “Organ donation and health” (N=1,686) showed that the most widely used crosswise-model RRT implementation produced false positives to a non-ignorable extent. This serious defect was not revealed by several previous validation studies that did not consider false positives – apparently a blind spot in past sensitive question research.

193

Hu, Y., (University of Essex) Netnography, Forum Discussion and Focus Group: A Comparative Experiment on Three Modes of Qualitative Data Collection Digital data are framed as an attempt to move beyond the ‘coming crisis of empirical sociology’. While researchers have mainly focused on the ‘quantity’ of digital data, less attention has been paid to the ‘quality’ of such data, and fewer have examined online and face-to-face qualitative data collection under a comparative lens.

In this research, comparative experiments are conducted on three modes of qualitative data collection, namely netnography (researcher-identity undercover), forum discussion (researcher- identity explicit) and focus group. Drawing on the UK-based Chinese student body, we experiment discussions on the meaning and value of of international education as perceived by the participants in each of the three modes. Netnography and forum discussion across seven online forums, as well as five focus groups with 30 participants are conducted. Issues pertaining to selectivity into the three distinct modes are assessed. The data are analysed in terms of substantive content as well as interactive dynamics underlying the process of discourse generation.

Despite similarities in the substantive content generated from the distinct modes, we find considerable differences in participants’ perspective of engagement, level of reflexivity, expression of subjectivity, emotionality, and between-participant and participant-researcher relations – between face-to-face and online modes, with and without the presence of an explicit researcher.

As participants’ subjectivity, positionality, reflexivity and emotionality are crucial to the interpretation of qualitative data beyond mere substantive content, we underline the importance of considering the interactive dynamics underlying the process of qualitative data generation in the development of digital research methods and ‘big data’ analytics.

194

Hülle, S., and Sakshaug J. (Bielefeld University, Germany and Manchester University, England) Record Linkage Consent: Assessing its Determinants and Magnitude of Bias Linking administrative data records with survey data increases the research potential. But the precondition for data linkage is usually respondents´ consent which is typically gained only for a subset of respondents. Moreover, research has provided evidence of differences between consenters and non-consenters that can cause record linkage bias. This article investigates record linkage consent (RLC) under consideration of mode-effects and assesses the determinants of RLC and the magnitude of RLC-bias. Empirical results are based on the first wave of a large population survey of people surveyed all over Germany (n=4,731). The database is especially suitable for methodological analyses since the survey consists of two independent random samples that differ in modes: A sample with face- to-face (CAPI) and a sample with self-administered (CAWI/PAPI) interviews. Respondents were asked to link their survey data with administrative data of the German Employment Research Agency (IAB) regarding their individual employment history and characteristics of their employer. Significant mode effects exist: The RLC-rate for face-to-face is rather high with 92%, but only 54% of the respondents in the self-administered modes gave their consent. This raises the question of the determinants of RLC and which (mode specific) mechanisms exist. A general theoretical framework that considers privacy risks and explains why people are (un)likely for RLC is lacking. The state of research suffers not only from this theoretical deficiency, also empirical deficiencies exist: Most evident are the inconsistent results of prior studies with effects of characteristics related to demographics, behaviors, and attitudes, while other studies do not find such effects. Another issue is the common use of indirect measures of privacy concerns by looking at objective indicators such as nonresponse of sensitive items. To overcome these theoretical and empirical flaws we derive and test hypotheses regarding concepts of trust and justice and consider respondents attitudes, i.e. subjective indicators. Trust is regarded as a precondition for RLC, since no rational individual will consent to share their own personal data if a misuse is expected. Therefore, people need at least a minimum level of trust for revealing personal information and to avoid such negative consequences. One way in

195

which people can develop (dis)trust is by their experience of situations with (in)justice. These experiences are likely to influence the general level of people´s (dis)trust which in turn affects RLC. The assessment of RLC-bias is usually not possible because data for non-consenters are not available. Contrary to that, the article´s analyses are additionally based on data of the entire gross-sample with the consent indicator and linked administrative variables. This unique dataset allows us to estimate the size of potential RLC-biases for administrative variables (e.g., benefit receipt) and the relative size of RLC-bias in relation to other biases (e.g., nonresponse). Besides the analyses of RCL-bias, this study provides theoretical and empirical contributions regarding the determinants of RLC. We present empirical support for trust- and justice-related hypotheses derived from theory. Our findings suggest that trust and justice play a decisive role in explaining why people (do not) consent to record linkage. Keywords: record linkage consent, administrative records, privacy

196

Hunter, Michael D. (University of Oklahoma Health Sciences Center) Revisiting Spinach and Ice Cream: A dynamical systems analysis of why social science is so challenging Genetic epidemiologists repeatedly find that once adulthood is reached shared environmental experiences explain virtually zero variability in any behavioral phenotype. Yet sociologists, social psychologists, developmentalists, and the like frequently discuss the importance of shared environmental influences. In part to explain this discrepancy, Turkheimer & Gottesman (1996; Turkheimer, 2004) proposed a system of simple rules as a rudimentary model for the interactions between genes, environments, and phenotypes. We recast this model as a nonlinear differential equation. We replicate prior simulations that showed invariant recovery of genetic influences, but only the ability to find common environmental effects in the presence of multiple cloned individuals. Moreover we provide an analysis of the proposed model as a dynamical system by investigating its fixed points, fractal structure, attractors, and Lyapunov exponents. These properties are then also compared to those of a simplified linear model in which gene activation is not influenced by the environment. Finally, simulated example data are fit to these models using polynomial approximation and ordinary least squares regression via state space reconstruction (Giona, et al., 1991; Casdagli, et al., 1991; Abarbanel, 1996).

Keywords: Behavior Genetics; Dynamical Systems; State Space Reconstruction

197

Hunter, M., Ou, L., Reddy, S., and Chow, S.-M. (University of Oklahoma Health Sciences Center; Pennsylvania State University; Pennsylvania State University; Pennsylvania State University) Comparing Numerical and Analytic Solutions to Stochastic Differential Equations One possible approach of fitting differential equation models to empirically observed, discrete- time processes is to use a continuous-discrete state-space model to represent the underlying change processes as evolving in continuous time as governed by a system of stochastic differential equations subject to unmeasured process noises, but the observed processes are measured at discrete times and are also subject to measurement noise. Under this model specification, a solution to two differential equations is required: (1) the solution to the differential equation that determines the predicted means of the latent processes at time t based on estimates from the previous time point; (2) the solution to the differential equation that determines the predicted variance-covariance matrix of the latent processes based on estimates from the previous time point.. In turn, these estimates and other related by-products are used in computing the likelihood function under prediction-error decomposition. In this presentation, we provide analytic and Monte Carlo simulation comparisons of three methods for deriving these solutions when the underlying continuous-time change processes of interest are in the form of a system of linear SDEs: (1) numerical methods for simulating the solutions of SDEs (available in the dynr package in R), (2) the analytic method based on the Kalman-Bucy filter (Kalman & Bucy, 1961, Eqs. 8 & 29; available in the OpenMx package in R), and (3) the analytic method based on the exact discrete method (available in the ctsem package in R). Models based on single- and multi-subject data generated using univariate and multivariate SDEs will be used to compare the solutions provided.

Keywords: Stochastic Differential Equations; Kalman-Bucy Filter; Exact Discrete Model

198

Ils, A. and Buchler, S. (Goethe-University Frankfurt am Main, Germany) Changing Representations of Motherhood and Fatherhood in Germany’s Media The social construction of parenthood and childhood, the role of mothers, fathers and children in families, and the time and resources spent by mothers and fathers on childrearing shifted profoundly in the majority of Western nations over the last decades. One of the most striking developments is a shift to so-called ‘intensive parenting’, which assumes that child development is almost exclusively shaped by parent’s influence, amplifying the social importance of the parent role. While a substantial amount of research has examined ‘new’ definitions of motherhood and fatherhood, there is very little research on how depictions of fatherhood and motherhood have changed in mainstream media. Yet, research on the representation of parents in mainstream media is essential, as media not only reflects but creates a form of reality that helps individuals to gain a certain understanding of a phenomenon. The portrayal of parents in mainstream media, especially in newspaper media as part of the public discourse, accessible to everyone at any time, is therefore a resource for (future) parents to construct their ideas about parenting and defines ideals of what ‘good’ parenting entails. Research examining parenthood in newspaper articles, however, is scarce and a systematic study of the discourses surrounding parenthood in newspaper media is absent. We aim to fill this gap by exploring German newspaper articles containing the terms ‘parenthood’, ‘motherhood’ and ‘fatherhood’ and examine change over time. As current research on the construction of parenthood in newspapers focuses on ‘deviant’ forms of parenthood, we draw on literature on new trends in parenthood, including both ‘intensive motherhood’ and ‘intensive parenthood’. Given these recent trends we hypothesize: (1) an increase in the occurrence of parenthood terms over time and (2) a change in discourse towards more ‘intensive’ forms of parenthood, including a greater focus on the mother-child dyad and ‘engaged fathers’. Our analysis is based on newspaper articles from 1994 to 2015 from the German high circulation and high quality newspaper Süddeutsche Zeitung. Applying a combination of relevance and multistage sampling we select

199

articles from the LexisNexis-database. Using the explorative method of text mining, we combine frequency and co-occurrence analysis and topic models to gain a comprehensive understanding of the discourse on parenthood. Initial findings suggest that the frequency of articles containing the terms ‘parenthood’, ‘motherhood’ and ‘fatherhood increased since 2007, which coincides with significant policy changes in parental leave in Germany, and have remained high since. Overall, our analysis is based on a quantitative analysis of qualitative data, enabling an innovative examination of changing parenthood ideals.

Keywords: Parenthood, media, text mining

200

Inna, B. (LINK Institute for Social and Market Research, Zurich, Switzerland) How much time should we allow to participate? How fieldwork duration affects data quality. There is empirical evidence that the length of the field period affects sample composition (e.g., Kreuter et al. 2010) which potentially introduces bias in survey statistics. The general assumption is that longer field periods could minimize this bias and as such increase data quality. While general rules-of-thumb exist, it is less clear what the minimum standards for the recommended length of a field period among data collection agencies should be. Some organizations try to minimize data collection periods and try to deliver data to the client as fast as possible in order to meet specific needs. Others aim at maintaining a longer duration of the fieldwork in order to obtain better data quality.

Practical decisions on the length of the field period are therefore often made on an ad-hoc basis and non-systematic personal experiences potentially compromising data quality. We provide statistical arguments to guide market researchers in the question: How long should a minimum study duration be in order to obtain high quality data? We also investigate which other factors, including reminders and device type used to complete a study, influence this decision. This knowledge helps to provide an optimal allocation of resources with respect to the output quality (as short as possible but long as needed) and helps to effectively reduce data collection costs.

The overall objective is to investigate how the length of the fieldwork period affects data quality. While other studies rely on indirect assessments of bias, the ultimate strength of our approach is the use of gold standard information as a criterion. More specifically we use individual-level information from the sampling frame to quantify the amount of the potential bias. First, we compare demographic distributions for individuals invited to participate in a survey to those who participated early in the study and those who participated later in the field period. We also analyze which of these variables is more strongly biased due to the differential length of the fieldwork as well as the whether the effects of duration of field period vary in different situations, such as by device type (desktop/smartphone/tablet), and the effect of reminders.

201

Finally, we investigate the study-specific variables and how substantive results and management decisions made on this basis would differ according to the differential sample composition throughout the fieldwork.

Our study shows significant differences in the distribution of the socio-economic variables associated with the duration of the field period. Thus, bias in the data is very likely when a minimum duration is not given. With the help of our customer study examples, we demonstrate how the differences in substantive results would affect management decisions if based on the data by early respondents alone.

1. Our results demonstrate the need for an appropriate minimum length of the fieldwork period and its effects on survey data quality. We discuss the implications of how much quality is improved with every additional day of fieldwork and provide guidance to determine an optimal minimum duration in the bias-cost tradeoff.

202

Israel, G. and Newberry, M. (University of Florida, University of Georgia, US) Effects of Stem and Response Order on Response Patterns in Satisfaction Ratings Although there are many design factors that affect rating scales responses, relatively few studies have examined response order effects. Order effects, when found, are often explained as resulting from satisficing behavior. One alternative explanation is that respondent heuristics influence the communication process between survey designers and respondents. We think that a typical heuristic among US residents assumes that positive response categories will begin at the left for horizontal scales and at the top for vertical scales. When the question design is consistent with the heuristic, the response distribution will elicit more positive ratings than when the question design is inconsistent. We expect that the order of options in the question stem can similarly affect response distributions, although perhaps less pronounced. That is, the stem order might also provide a verbal cue to respondents when the response order does not correspond to the heuristic. This study reports on two experiments embedded in the annual survey of FCES clients. In experiment 1, four items on aspects of service quality are presented in a grid with the response options listed horizontally. Experimental factors are the order of the response categories (‘Very satisfied’ first versus ‘Very dissatisfied’ first) and the order of the options in the question stem (‘satisfied’ first versus ‘dissatisfied’ first, resulting in a 2-by-2 design. Similarly, experiment 2 involves a single item asking about overall satisfaction with the response options arranged in a row below the question stem. A web/mail mixed-mode design yielded 1,618 responses (51.4% RR2). We found consistent evidence that response order affected answers in both experiments, with those having the ‘Very dissatisfied’ response option first being 2-3 times more likely to select the very dissatisfied response option than those having the ‘Very satisfied’ response option first. The order of satisfied or dissatisfied in the question stem also affected responses for the items in the grid but not for the single item on overall satisfaction. No interactions between the two factors were significant and response mode did not affect the results. The statistical evidence was also supported by respondent comments on mail questionnaires, where “Oops” or similar comments were marked on questionnaires with ‘Very dissatisfied’ listed first. In general, there was clear evidence of response order effects consistent

203

with theories of satisficing and respondent heuristics but inconclusive evidence of stem order moderating these effects.

Keywords: Order effects; Response Heuristic; Satisfaction Ratings

204

Järv, O., and Masso, A. (University of Helsinki, Finland, and University of Tartu/ETH, Zurich) Validating the implementation of mobile phone data: an example of spatial segregation research To better understand human social actions and how our societies function, enhanced knowledge of people in space and time is needed. The emerging ‘mobility turn’ within the social sciences is challenging the ways in which much of the social science research ignores the multifaceted spatialities that daily social life presuppose. Thus, new ways of looking at socio-spatial divisions in societies are embraced and new data sources and innovative methods are adopted.

Given the ubiquity of mobile phone usage in people’s daily lives, mobile phone based data has been proposed as one novel complementary data source to explore human spatial mobility, and widely used to reveal societal structures, social processes and phenomena. However, the implementation of given data imposes many conceptual and methodological challenges, whereas one important issue is the accuracy and validity of results. This has been surprisingly less addressed by scholars, to date. For example, passively collected anonymous data about mobile phone usage is recently introduced in examining ethnic differences in human spatial mobility and for assessing a multifaceted phenomenon of socio-spatial segregation. Without validating results against the conventional data collection methods, however, raises the question of relevance and applicability of such data to study social phenomena, and in particular socio-spatial inequalities.

Against this backdrop, we attempt to contribute to the question of validity and applicability of mobile phone data in social sciences research. Hence, we conduct a comparative study between mobile phone data and a traditional questionnaire survey in case of urban segregation research both at an individual (agent-based approach) and an aggregate (place- based approach) level. We sought to find answer to three research questions: 1) how and to

205

what extent ethnic differences in individual spatial mobility coincide in case of two data sources; 2) can questionnaire survey reveal additional underlying personal variables describing differences in individual spatial mobility and how it influences the effect of ethnicity; 3) how and to what extent spatial segregation at aggregate level coincides in case of two data sources.

The quantitative study is conducted in Estonia, whereas focusing on differences between Estonian- speaking majority and Russian-speaking minority. We apply a definition of ‘ethnicity’ by referring to groups of people identifying each other based on common ancestral, social, cultural or national experiences. Here, the division between ‘Estonians’ and ‘Russians’ fits this definition in that both language groups are distinguishable, not just in terms of their language, but also in terms of historical background and current socio- economic position. We examine two different datasets: 1)1.3 million mobile phone call activities from 550 mobile phone users to map their individual spatial mobility during a one year period; and 2) an in-depth questionnaire survey (e.g. regarding spatial mobility) of 550 respondents. At individual level, we examine the extent of monthly and annual individual activity spaces and apply general linear models. At aggregate level, we examine conventional segregation indices to assess segregation levels in case of both datasets. This comparative study is an ongoing project and empirical analysis are currently being conducted.

Keywords: Spatial Analysis; Mobile phone data; segregation

206

Johnson, J. (Centre for Longitudinal Studies, UCL) Discovery, provenance, harmonisation and re-use of longitudinal studies The UK is home to the world’s some of the longest-running longitudinal studies most notably the national birth cohorts starting in 1946, 1958, 1970 and 2000 and many other regional studies such as ALSPAC, the Southampton and Hertfordshire cohorts and the successor to the BHPS, Understanding Society. Cohort & Longitudinal Studies Enhancement Resources (CLOSER) aims to maximize their use, value, and impact both at home and abroad though the development of a rich metadata portal which has the broad aims of easing discovery, providing provenance of both questions and variables thereby encouraging comparison, harmonisation and reuse of questions.

At the heart of the project is the assembly of a metadata repository encompassing the full range of information about the studies, from study description to data collection methodology, questionnaire capture and data. Layered on top of this are a controlled vocabulary and other contextual information.

The project has very specific challenges. The studies are based in six organisations geographically dispersed across England. Leadership of the studies crosses scientific domains (some primarily medical, some social science focused) and the longevity of the studies (starting as early as 1946) means that the information available is inconsistently catalogued and held in various formats.

Utilising the DDI-Lifecycle metadata standard, the project is creating a research discovery portal that will be interoperable with similar efforts on other similar studies to enable discovery and comparison of questions and data from these studies.

The project has primarily used off-the shelf software, combined with some lightweight tools has enabled mass ingest of metadata into a repository in a consistent and manageable process.

The paper will focus on the challenges of creating such a rich metadata repository, the lessons learnt and its implications for reuse of questions in the development of future studies, in discovery for research and how such infrastructures can assist harmonisation and comparison within and across these and similar longitudinal multi-purpose studies.

207

Keywords: Metadata; Discovery; Longitudinal

208

Jurado-Guerrero, T., Domínguez, M. and González, M. (Departamento de Sociología II, Universidad Nacional de Educación a Distancia, Madrid, Spain; Observatoire Sociologique du Changement, Institut d’Etudes Politiques de Paris (Sciences Po), France; Universitat Pompeu Fabra, Departament de Ciències Polítiques i Socials, Barcelona, Spain)

Analyzing couple’s life courses: (un)doing gender in dual-earner couples in Spain This paper describes three different examples about how to analyse in-depth over-time between-couple and within-couple differences, as applied by the Spanish team of the international research project transPARENT. Ten researchers collaborated in interviewing 68 dual-earner couples in 2011 and 2013 in four different Spanish towns, in coding the interviews, systematizing the socio-demographic information and in analysing the transcriptions according to different topics.

First, we present an in-depth analysis of five heteronormative couples that focuses on the couple-specific experiences and looks at “her” and “his” discourses concerning their work-life balance and the division of domestic and care work between the spouses. All five women in these couples had expressed egalitarian attitudes, but one adapts her discourse to the traditionalization of her relationship; another woman accepts some specialization despite shared egalitarian attitudes; the third resists traditionalization until becoming unemployed and the fourth and fifth construct relatively egalitarian parenting practices with their employed/unemployed partners. The in-depth description of few couples allows best to unfold the potential of longitudinal analysis in order to show how people sometimes adapt their discourses to their practices, but also are able to sustain discourses that do not really fit their daily routines.

The second study describes the transition of 57 men to first-time fatherhood and focuses on the individual, couple and context factors in order to explain the classification of fathers into occasional carers, helpers and involved carers. Ideals, plans, domestic care practices and relative resources at wave one, help understanding how their trajectories into fatherhood differ. In this case, the theoretical perspective of gender as a social structure (Risman and Davis 2013) and previous classifications of fathering practices constitute the lens with which between-couple

209

differences are studied by focusing mainly on men. Starting from wave two, we look at men’s trajectories and their relation with partners’ evolution into different forms of motherhood practices.

The third study divides into two analyses: a first one after wave one and a second after wave two. The couple-data of 68 couples of the first wave were analysed to identify 31 couples sharing routine domestic work relatively equally. A Qualitative Comparative Analysis was used to explain, how different configurations of couple factors lead to equal-sharing of domestic work. Then, the evolution of 27 of these couples was studied to understand why 17 were able to maintain a 60/40 division of domestic work, while 10 couples traditionalized. We found that the identified practices at wave one revealed a predictive character for when the first child arrives. A necessary condition for undoing gender in the division of domestic work was to have a non- traditional domestic arrangement before birth of the first child, and some specific characteristics of the division of work before childbirth were related to what happened until wave two. This study focused on between-couple differences at two points of time.

Key words: gender equality, division of labour, families and work

210

Kandt, J. (University College London) Socio-cultural space over time: the geography of population dynamics in Great Britain

This paper explores how population and consumer registers can be employed to include an important, under-examined empirical dimension into spatial investigations of socio-cultural phenomena: the geo-temporal context of places. The paper is motivated by the observation that spatial investigations of socio-cultural phenomena (for instance, neighbourhood effects research) often rely on a static and generalised notion of place, without accounting for the specificity of places in geographical and temporal terms. Yet, specificity and temporal stability of places affect socio-cultural phenomena, such as social networks, bonds, practices and identities, which each in turn shape the workings of neighbourhoods, cities or regions. In this paper, I seek to conceptualise and empirically capture one layer of a geo-temporal perspective on socio- cultural ‘place specificity’. First, I show how, in the UK, broader, probable regions of socio- cultural specificity can be identified and employed in the study of lifestyle geographies and their specific, regional expressions. Socio-cultural variations of lifestyle milieus can be observed when milieus are viewed within six bespoke cultural regions that broadly correspond to southern England, northern England, urban England, Wales, Scotland and Northern Ireland. The results highlight the importance of regionalisation in socio-cultural rather than administrative terms in conceiving of geographical, social phenomena. Second, I add a temporal perspective in which I measure the degree to which neighbourhood ensembles in Great Britain have likely changed in socio-cultural terms since 1881 and more recently before and after the financial crisis. This step reveals a complex picture of population turnover in Great Britain, which results in parts from geographically varying naming conventions, ethnic, cultural and linguistic diversity and, most importantly, domestic and international migration. Population dynamics in those terms appear to be most intense in southern England over the long-run, whereas the patterns observed in Scotland, Wales and northern England indicate a significantly lower level of change. This imbalance of population dynamics is indicative of economic imbalances that have emerged over

211

the same period with arguably profound implications for conditions of social cohesion, community stability and place identity. The paper will conclude with a discussion on how a geo- temporal perspective of place specificity may be empirically captured, and how supporting methods of spatial analysis may aid conceptual work on the role of place and space in shaping further socio-cultural phenomena.

212

Kanu, W. (Department of Sociology, Imo State University Owerri, Nigeria) Religiosity, Gender and Political Participation in Rural Areas of Imo State, Nigeria That there is a connection between religion and politics in Nigeria is no longer contentious. However, what is unclear is the nature and direction of this relationship in relation to the political behaviours of individuals Nigeria. As uncertain as this relationship is in Nigeria, the impact of religiosity on political participation has so far received little attention; consequently researches in this direction are very few. Previous studies on impact of religiosity on political participation were mostly done outside Nigeria. Moreover, the studies showed discrepancies in their findings. Furthermore, very few of the existing studies in Nigeria were conducted in rural areas where majority of the people live and none was done in Imo State of Nigeria. Understanding this impact is pertinent, especially in a democratic and religious diverse society like Nigeria. This study examines the impact of religiosity on political participation in rural areas of Imo State. The study also examines how the impact varies between men and women. It is hypothesized that: religiosity is inversely related to political participation and; that the impact varies significantly by gender. The instruments for the collection of primary data are questionnaire, in-depth interview and Focus Group Discussion guides. The sampling techniques adopted to select the local governments and communities for the study are simple random sampling technique, while the sampling techniques employed to reach the respondents are “snowball” and “accidental” methods. Religious leaders, community leaders and some selected indigenes were involved in in-depth interviews in each community sampled. Two sessions of FGD, one for males and another for females were conducted in each community. The indicators of religiosity are denominational affiliation, frequency of church attendance and participation in church activities, commitment to religious rites. Political participation is divided into conventional and un-conventional forms. The indicators of conventional political participation are voting, attendance of political meetings, campaigns, competing for political positions, membership of a political party, involvement in petitions and protests while the indicators of un- conventional political participation are involvement in political thuggery, rigging, looting/vandalization of polling box, rioting and misconduct. Persons from 18 years and above

213

formed the target population and the sample size is 346 respondents. Analytical techniques include both descriptive and inferential statistical tools. Recommendations are made based on the findings.

214

Kawulich, B. and D'Alba, A. (University of West Georgia, United States) Teaching Qualitative Research Methods Online with Second Life Since the 1950s, when Heilig created the Sensorama, a simulator providing the illusion of reality, additional multi-user virtual environments (MUVEs) have been developed to simulate real world experiences. These virtual worlds have evolved with realistic graphics and large numbers of users with whom to engage in virtual representations of real life situations. This paper investigated the usefulness of one of these MUVEs, Second Life (SL), as a tool for teaching qualitative research methods to a doctoral level Qualitative Research class. Students were assigned to groups of three students and were charged with finding a SL community in which to conduct ethnographic research. Students were required to participate in a SL culture of their choice where they conducted observations and interviews of patrons as a simulation of conducting ethnographic research in an unfamiliar culture. Data were collected through an end of course survey, observation field notes, interview transcriptions, journals, discussion board posts, and video conferencing. Results indicate that SL provided an innovative means for teaching important strategies to conducting qualitative research, such as securing key informants, gatekeeping, observation skills and taking field notes, interviewing skills (like securing participants), and the importance of dressing appropriately and behaving according to societal norms, among other strategies. Students experienced the affective aspects of conducting research as well, including how to handle rejection and frustration, the need for flexibility, learning to manoeuvre in a strange world, and how to handle awkward situations, sometimes with hilarious results. While the majority of students found SL to be an innovative way to learn research skills, a few experienced frustration that hindered their learning. Students created a list of tips for new users of SL to ease their transition into the virtual world. Implications for future research are shared.

215

Keller, R. (Institute of Social Sciences, Faculty of Philosophy and Social Sciences, Augsburg University, Germany) SKAD and the methodologies of sociological discourse research In her influential book “Situational Analysis”, American sociologist Adele Clarke (UCSF) argued that in today’s societies, every social phenomenon including social actors are awash in a sea of discourses. Nevertheless, and despite the groundbreaking works of Michel Foucault, sociology has to date been hesitant in developing methodologically sound ways for approaching discursive phenomena that have been adequately reflected upon. This is due in part to the dominance of a rather loosely defined concept of discourse in cultural studies, and of other, more widely used linguistic tools and devices in discourse research across the humanities.

In this context, the session’s introduction intends to discuss methodologies of discourse research that are adapted to sociological purposes and are able to address the complex discursivity of today’s orders of discourse and knowledge production, accounting for actors as well as discourse structuration in methodologically substantiated ways by using sociological methods and tools for inquiry. In addition to a general discussion of such methods, the session will offer and in-depth focus on the methodology and methods of the sociology of knowledge approach to discourse (SKAD) which combines elements of the sociological interpretive paradigm with Foucauldian concepts and strategies of qualitative research for doing discourse research today. In this way, the session aims to strengthen genuinely sociological methodologies of discourse research.

Keywords: discourse analysis, methodology, sociology of knowledge, SKAD

216

Kelly, G. A. (North Dakota State University) An Examination of Underage and Adult Binge Drinking in Rural North Dakota Society as a collective, provides evidence of cultural norms and expectations. These expectations are both motivated from within an individual and also motivated from family, friends and others around us who provide both positive and negative impacts. Most individuals strive to find positive rather than the negative. But what happens when our negative influences blurs social norms? This research examines cultural competency with underage and adult binge drinking through the Strategic Prevention Framework – State Incentive Grant (SPF-SIG). The SPF-SIG had two primary goals to address major health concerns for Ransom County: 1) Underage Drinking and 2) Adult Binge Drinking. A total of 1,500 randomly selected households in Ransom County, North Dakota were chosen for the survey. An introductory postcard was mailed to respondents followed by a survey packet (i.e., survey and a postage-paid self- addressed return envelope). Approximately three weeks after the survey was mailed a reminder postcard was sent to non-responders. A total of 236 completed surveys were returned for analysis. The response rate is 18 percent and the confidence level is 95 percent with an error rate of +/- 6 percent. These findings provide evidence that alcohol use and misuse is a problem in Ransom County among adults and teenagers. Adult and Underage Drinking is a problem in Ransom County, North Dakota. There is evidence scientifically through external (i.e., state evaluators and researchers) and internal channels (i.e., community members and professional stakeholders) to support these problems.

217

Khavenson, T. (National Research University Higher School of Economics) How well School Age Children Report Family SES. Systematic Review School age children more and more often turn out to be the single respondents, i.e. the single source of information in social science surveys. However researchers need to gather not only children’s opinion or attitudes, they also need some factual information, including data about parents or family socioeconomic status (SES). Gathering information only from children without comparing it with other informants implies that students’ responses are not prone to errors or at least that their measurement error is equal to the adults’ one. Is that true? Are school age children competent enough to answer questions about parents’ education, occupation, articles in home and other family SES related questions?

Recently, partly due to the increase of the large scale educational studies, the number of investigations of the children’ response quality have been carried out. Even though they are sporadic and provide controversial conclusions, therefore we believe that these results need to be summarized and assessed in general. The purpose of this study is to analyze the quality of different approaches to measuring family SES within surveys of school age children. We can formulate three interrelated research questions:

Which SES measures appear to have higher consistency in the responses of children and parents?

To what extent different SES measures are prone to be missed?

Are there any respondents’ characteristics that are related with the response quality?

We employ a systematic review (Petticrew & Roberts, 2006) of the studies devoted to investigation of the quality of children responses on family SES variables. The papers selection criteria are as following: 1. The research is about the quality of the school age children answers on the family SES questions; 2. The paper should have methodological focus, i.e. it should be solely devoted to these response quality issues. 3. We also excluded studies with too specific

218

samples. 4. We mostly focused on the papers published since 1990, as in 1989 similar review had been done (Looker, 1989). 5. Papers published in English language were selected.

We used the following approaches to look for relevant papers: 1. Snowball method – checking the references in the papers more or less relevant to the topic. 2. Search in the citation databases (SCOPUS, WoS) and periodical databases (JSTOR, EBSCO, Science Direct and others) for the papers that cite the articles identified through the snowball search. The special emphasis was made on (Looker, 1989). Actually we tried to spot all the papers that meet our criteria.

16 papers were included in our final list. 14 of them allow answering research question 1 about consistency. 11 papers cover the missing data issue (research question 2). 8 papers contain part about students’ characteristics and response quality (research question 3). Correlation between children and parents responses on education varies from moderate to quite strong. The average and the median of correlation coefficients are 0.6 for both parents. Correlation on question about parents’ occupation varies in the same range, however the average and median are slightly higher – 0.7. Median percent of agreement is about 60% and median Kappa coefficient is 0.5 for the education and 70% and 0.7 for occupation, respectively. Results on missingness differ from paper to paper. For mother’s occupation the minimum number of missing cases is 0.6% and maximum is 46%. Average percentage is 12%, median is 9%. Number of missing in the variable on father’s occupation lies in a range from 1% to 34% with an average at 7% and median at 5%. So at least in a half of surveys children left without answers less than 5% for father and 9% for mother. In regard to educational variables the situation is a little bit better according to the analyzed studies. On average children miss 7% (median 5%) in mother’s education and 8% (median 3%) father’s education. Range is 0.7-15% and 0.2-39% for mother and father, respectively. Medians here are even less than for occupation. Yet a “Do not know” option usually is also available in “parents’ education” question and a substantial share of student tick it. As far as the students’ characteristics are concerned, age, achievements, family structure and family SES turned out to be positively related to the parents-children consistency. Age and family structure are the only ones to be related with the tendency to skip questions. Almost in all studies the results for “parents’ occupation” questions is a little bit better than for “parents’

219

education” questions. In general school age children tend to misreport one category up or down in all SES proxies. So merging categories may increase consistency.

220

Kim, J., Shin, I. and Yang, D. (Sungkyunkwan University) Identifying respondents who are very difficult to contact using interviewer’s reports Social environments across different countries influence the contactability of respondents. We examine the characteristics of respondents who are very difficult to contact in Korea. We used the 2003-2013 cumulative Korean General Social Survey (KGSS) (N=15,328), which is a nationally representative survey of adults conducted via face-to-face interviews carried out by Sungkyunkwan University. At the end of the survey, interviewers responded to the question, how difficult is it to contact the respondents? They reported 11% for not difficult at all, 31% for not difficult, 42% for difficult, and 16% for very difficult. Using logistic regression, we found that those who are not married, are living in urban areas, and are employed are associated with being difficult-to-contact respondents. But gender, education, and family income, and living alone or not are not associated with being very difficult-to-contact respondents. Also, male interviewers are more likely than female interviewers to report very difficult-to-contact respondents. Interviewers' evaluations of very difficult-to-contact respondents may be used to tailor the survey design. Keywords: contactability, Interviewer evaluation, paradata

221

Knappertsbusch, F. (Justus Liebig University, Giessen, Germany) Qualitative Methods and Causality in Mixed Methods and Multimethod Research – A Methodological Comparison The current literature on method integration in the social sciences is marked by a curious division between disciplines: While the methodology of multi method research in comparative political science has a strong focus on methods of causal analysis and set-theoretic qualitative methods, mixed methods research in sociology and educational science often pays much less attention to causality and is driven by an interpretivist understanding of qualitative methods.

This paper seeks to further a methodological dialog between the two strands of integrative research methods, focusing on notions of qualitative research and the different concepts of causal analysis they include. The underlying premise is that their respective foci on method integration are complementary and could be combined to mutual benefit: Mixed methods approaches could gain a stronger and more systematic concept of (qualitative) causal analysis, while multi method research could accomplish a deeper understanding of the hermeneutic aspects of social research methods.

My presentation aims at mapping common ground as well as areas of contrast between both lines of research via a systematic literature review. I will then offer points of departure for an integration of the two notions of integrative social research, including the following:

(1) An understanding of causal analysis which incorporates efficient causes (i.e. cause-effect explanations) as well as final causes (i.e. teleological or functional explanations) as mutually dependent aspects of explanation in the social sciences,

(2) a process based understanding of causality in the social sciences, which might serve as a common ground for both multi and mixed methods research,

(3) a notion of explanation and interpretation in the social sciences (Weber: “Erklären” and “Verstehen”) which highlights the logical interdependency of both concepts.

Keywords: Mixed Methods, Multi Method, Qualitative Methods

222

Knight, V. (De Montfort University, England) Concept-formation, Complexity and Social Domains: Investigating Emotion(s) in a Prison Setting This paper reports on an ethnographic study1 focusing on the impact of in-cell television on prison life in a male adult prison in the UK. Layder’s social domains model (1997; 2006) and his adaptive analysis (1998; 2014), were used to give shape, meaning and organization to data from interviews with prisoners and staff and also television-use diaries. The research highlighted how television is adopted for its care-giving qualities (Knight 2015; 2016). This paper focuses primarily on what prisoners do and what prison does, with emotions. The paper focuses on examining and developing conceptual links between self, emotions and control in the prison setting.

223

Krause, R.W., Huisman, M., Steglich, C.E.G. and Snijders, T.A.B. (Department of Sociology, University of Groningen; Nuffield College and Department of Statistics, University of Oxford) Multiple Imputation for Network Analyses Missing data on network ties is a fundamental problem for network analyses. The biases induced by missing edge data, even when missing completely at random (MCAR), are widely acknowledged and problematic for network analyses (Kossinets, 2006; Huisman & Steglich, 2008; Huisman, 2009). Although model-based techniques for handling missing network data are quite promising, they are not available for all analyses (Koskinen, Robins & Pattison, 2010; Koskinen, Robins Wang, & Pattison, 2013). Multiple imputation for network data is able to overcome this problem. This study expands on recent work on imputation of missing data in networks (Wang et al. 2016). In the current simulation study, different models for treating missing data are compared on different simulated networks varying in size, density and clustering. The compared models range from simple models (available case analyses, reconstruction (Stork & Richards, 1992), imputation of no-ties) to multiple imputation using exponential random graph models. The effects of different missing data mechanisms (MCAR, MAR, MNAR), different percentages of missingness (10-40%), and different types of missing data (actor non-response, boundary misspecification) are investigated . The performance of the imputation techniques is examined by comparing parameter estimates of ERGMs estimated on both the complete and imputed network data.

References Huisman, M., 2009. Imputation of missing network data: some simple procedures. Journal of Social Structure. 10 (1), 1–29. Huisman, M., Steglich, C., 2008. Treatment of non-response in longitudinal network studies. Social Networks. 30 (4), 297–308. Koskinen, J.H., Robins, G.L., Pattison, P.E., 2010. Analysing exponential random graph(p-star) models with missing data using Bayesian data augmentation. Statistical Methodology. 7, 366– 384. Koskinen, J.H., Robins, G.L., Wang, P. and Pattison, P.E., 2013. Bayesian analysis for partially observed network data, missing ties, attributes and actors. Social Networks. 35(4), 514–527. Kossinets, G., 2006. Effects of missing data in social networks. Social Networks. 28(3),247–268. Stork, D. and Richards, W.D. (1992). Nonrespondents in communication network studies. Group & Organization Management, 17, 193-209. Wang, C., Butts, C.T., Hipp, J.R., Jose, R. and Lakon, C.M., 2016. Multiple imputation for missing edge data: A predictive evaluation method with application to Add Health. Social Networks. 45(), 89-98. Key words: Missing Data, Network Analyses, Multiple Imputation, ERGM, Simulation

224

Krebs, D. (University of Giessen) Agree-Disagree versus Item Specific Scales – The Impact of Different Response Formats on Response Behavior Likert scales with agree-disagree (A/D) response format are the most preferred and widely used method for measuring attitudes in the social sciences. Nearly from the beginning of the career of the A/D response format there have been questions with respect to data quality and response styles associated with method. In recent times (see f.e. Saris et al., 2010 as well as the recent study published by Liu et al., 2015) to avoid these drawbacks of A/D-scales an alternative method has been suggested: the item-specific (IS) response format providing for each item a content dimension specific for that item. Compared to the studies mentioned above, this paper describes an additional approach to design the IS response format. While in the A/D response format respondents are requested to map their attitude into a category on an agree-disagree dimension, in the IS format respondents can describe their attitude on content specific dimension directly related to the item content. Often this type of response scale is therefore called “content-specific (CS).

The comparison between the A/D and the IS response format is based on an experimental study using a split-ballot design with repeated measurement. In both waves of this panel study questionnaires were randomized according to response format A/D and IS. Thus there are four groups allowing comparisons between as well as within response formats. Items belong to different content areas (different aspects of job motivation, achievement motivation, anomia and authoritarianism). Attention was given to a comparable length of items (or questions) as well as response options. Respondents were university students.

First results suggest that depending on item (or question) content there are only small differences between the two response formats: attitudes (or responses to attitude items) suspected to be “created on the spot” and not reflecting a deliberate attitude are susceptible to response format effects. Mostly these are items describing the status of society or of moral concepts. In contrast, items presumed to reflect a self-conscious description of the respondent her/himself yield comparable responses in A/D and IS response format.

225

Reference: Saris, W.E., Revilla, M., Krosnick, J.A. and Schaeffer, E.M. (2010): Comparing questions with agree/disagree response options to questions with item-specific response options. Survey Research Methods, 4: 61-79. Liu, M., Lee, S. and Conrad, F.G. (2015): Comparing extreme response styles between agree- disagree and item-specific scales. Public Opinion Quarterly, 79, 4: 952-975.

Key words: item-specific response format, response behavior, scaling.

226

Kropp, P. (Institute for Employment Research, Germany) Comparing regional delimitations by Newman’s modularity approach Newman's modularity approach allows comparing the clustering of a network with a random clustering of a network that has similar properties as the network investigated. Using this approach, we evaluate different clustering methods with regard to their efficiency. Furthermore, if clustering procedures several solutions – as it occurs typically for hierarchical clustering methods – the modularity approach can be used to identify the optimal solution. Evaluation of different clustering methods and identification of the optimal solution for the best methods will be applied on the example of the commuting network in Germany. This network consists of commuting ties between regions, which vary in their strength according to the number of commuters between them. Finding clusters in this valued network is important to identify regional labour markets.

227

Krusell, S. (Tallinn University) Ethnic Cleavages in the labour market –Which concept to use for defining a target group? In most countries, the data and the underlying theoretical approaches tend to emphasise the greater success of natives or ethnic majority in the labour market (Chiswick 1978, Priore 1979, Esser 2004, Heath and Cheung 2007, etc). Depending on the author and country data the concept of origin or ethnicity are being used but rarely it is discussed which concept would be better to measure ethnic cleavages between Estonians and Russian speaking population in the labour market. In Estonia the ethnic or immigrant differences in labour market have had a attention in several studies. Basis for comparison have been in several studies the distinction between Estonians-Non-Estonians, Russians-Estonians, Russian speakers/Estonian speakers or natives/immigrants. The objective of the current paper is to analyse which concept is justified to use in Estonian context through the labour market figures. The Labour Force Survey (LFS) was used for the empirical analysis and the data cover the years 2013–2014. The most important finding from the analysis are that in Estonian context the ethnic concept would be more useful for measuring ethnic cleavages. This conclusion is supported by empirical results where labour market figures of Russian speaking natives are more resembling to Russian speaking immigrants than native Estonians. Although being a case study, the results of this article accents importance of discussion about used concepts for forthcoming analysis also for other countries when disadvantages and different positions in labour market are being analysed.

228

Kuiper, R. (Department of Methodology & Statistics, Utrecht University)

Combining Statistical Evidence from Multiple Studies Introduction Researchers may have several data sets that can be used to address a research question with respect to the relation between two variables. The main variables of interest may be operationalized in different ways, be measured on different scales, and the statistical model used to relate both variables may differ between studies. This presentation shows how Bayesian updating can be used to summarize the evidence in the data sets for hypotheses on the relation between the two variables.

Disadvantage Meta-analysis

To combine multiple studies, one can employ (Bayesian) meta-analysis. Meta-analysis can be based on i) a parameter estimate and its standard error or ii) a corresponding effect size for each study or iii) a p-value. The first two can only be done when parameters or effect sizes are comparable. Hence, one cannot use parameters from another type of model or effect sizes that cannot be transformed into one type (e.g., odds ratios). In addition, the design of the model should be the same in all studies, that is, the predictors in the model should be the same for all studies. Namely, the parameter estimate or effect size is a conditional one and, therefore, it might change when adding or discarding predictors. A drawback of the third method is that p- values do not only reflect effect size but also the number of observations.

Proposed Method

To combine multiple studies of different types and designs, but regarding one theoretical concept, we introduce a Bayesian updating method. In this method, as opposed to meta- analysis, the hypotheses do not address a specific parameter. Instead, it covers an underlying effect such that the parameters are indicative for the same underlying effect.

The method can be employed to evaluate the following hypotheses:

H0: no effect

229

H>: positive effect

H<: negative effect

Notably, our method does not combine estimates but the evidence for a positive (H>), negative

(H<), and null effect (H0) of the predictor of interest (which measures one theoretical concept, say, past) on the dependent variable (which measures one theoretical concept, say, trust).

The input for our method is the estimate of the parameter of interest and its standard error, which adequately summarize all the necessary information in the data with respect to the effect. Thus, this method does not require comparable parameter estimates, and it can be used for different types of models as well as studies with different sets of predictors.

It should be stressed that the researcher has to make sure that the parameter estimates of the studies (to be combined) do reflect comparable relationships between the (two) key variables.

This method quantifies the evidence for the three hypotheses of interest based on all the studies (instead of rendering a parameter estimate based on all studies).

Presentation

The method will be explained and illustrated by combining the evidence from the studies on how trust depends on previous experience. Moreover, the sensitivity of the method is addressed and ways to deal with this sensitivity.

Keywords: Bayes factor; Bayesian updating, posterior model probabilities

230

Kunz, T., and Fuchs, M. (Darmstadt University of Technology) Respondent burden and satisficing in grid questions: A comparison of early versus late positioning in a Web survey instrument Questionnaire length has been identified as a key factor inducing satisficing behavior. Accordingly, common wisdom assumes that respondent motivation declines with each additional question respondents have to work through. An increasing number of survey questions that have already been answered and the associated respondent burden may lower the respondents’ motivation needed to thoroughly process the questions and thus increase the risk of satisficing behavior. When it comes to grid questions as a frequently used question format to measure attitudes and opinions, the number of items presented in a grid may also affect the extent of effort devoted to the cognitive processing of the items. Especially items presented later in a grid are assumed to be affected by satisficing behaviors and hence are likely to exhibit lower data quality.

This paper investigates how different indicators of data quality in a grid question—item nonresponse, response times, and response differentiation—vary depending on the relative position of the grid question within the Web questionnaire as well as depending on the relative position of the items within a grid. In a between-subjects design, a grid question comprising ten items and a five-point response scale was randomly provided in one of four distinct positions, whereas the remaining survey questions were held in a fixed order. Furthermore, the items of the grid were presented in a completely randomized order using JavaScript programming in order that each respondent answered the items in random order. Thus, we were able to assess the impact of the number of questions answered before the grid as well as the impact of the relative position of an item within the grid on data quality.

At the macro level of the questionnaire, findings revealed that the grid question asked later in the questionnaire compared to the earlier positions was associated with higher item nonresponse rates, shorter response times, and lower response differentiation. At the micro level of the position of each item within the grid, however, no consistent evidence was found for

231

the assumption that items presented towards the end of the grid would be more prone to speeding. Furthermore, the risk of items being unanswered and the degree of differentiation seemed to be unaffected by the relative position of the items in the grid.

These findings indicated that grid questions appearing later in the questionnaire seem to be associated with higher respondent burden lowering the respondents’ motivation to accurately answer the grid. By contrast, there was no clear indication that respondents generally speed through a set of items or that the respondents’ tendency to rely on satisficing behaviors increased over the course of item completion within a grid. Obviously, the positioning of grid questions in the questionnaire has a greater impact on data quality than the particular positioning of items within the grid.

Keywords: rating scales, question order effects, data quality

232

Kumar, M. and Kumar, A. (Banaras Hindu University, Varanasi, India)

Changing socio-economic condition and livelihood of Aged society of tribal community in eastern Uttar Pradesh, India The tribes are children of nature and change is inevitable in nature, it may be rapid or slow, good or bad and complete or partial. Majority of tribals are living below the poverty line, possessing meagre assets and are exclusively dependent on wages, forest produce and agriculture. The income and living condition of the elder tribes in comparison with other community elder people are much discouraging. In order to get a clear view of this issue, this study attempts to examine the changing pattern in socio-economic condition and livelihood of the elder in tribal community of eastern Uttar Pradesh.

The study is the outcome of intensive fieldwork analysis of collected data from 25 villages of three districts in eastern Uttar Pradesh, India. The composite index of development is the basis for the selection of districts. Randomly one district each from low, middle and high index categories, and from each selected districts one sub districts were selected at random basis , out of which one village was selected randomly and at the end complete enumeration of the elderly tribal population in a particular village was carried out. However, if after complete enumeration the required sample size has not been obtained in a particular village, then we select another village randomly from the sub-district and follow the same process. But if the minimum sample size after complete enumeration is still not obtained, then the same procedure will be followed to complete the minimum sample size which is proportionate of the district population and total sample size.

233

Main findings of the study can be succinctly summarised as; only 15.7 % of the total households had Pucca house in 2005 which is increased to 27.8 % in 2015 and only 3 % have toilet facility in

2005 which goes up to 10.6 % in 2015. Annual income of 43.7 % household in 2005 is less than

INR 5,000 as compared to 24.2 % in 2015 and Out of the total sample only 8.6 % had access to loan in 2005 which has increased to 18.2 % in 2015. Prior 2005 only 11.6% were having saving bank account, during 2005-2015 number of households having saving increased to 25.5 % and major source of household income in 2005 was forest gathering and daily wage labour but in

2015 it include minor share of government and private sector job. The study noted that as a result of developmental interventions like IAY (Indria Awas Yozna), MGNREGA, SGSY( Sanjay

Gandhi Swarojgar yojana), OLD AGE PENSION and other awareness programs; the Lifestyle of elderly tribal have not changed and it’s not much helped them to join the mainstream.

Keywords: Socio-Economic Survey, Tribal, Livelihood

234

Kumar, A. (Banaras Hindu University, India) Demographic and Physical Well Being of Elderly Tribal People in North India In India, the current demographic scenario indicates that mortality has declined significantly and continues to show sign of further decline. This reduction in mortality, is contributing to an increase in size of population, suggesting that ageing in India is likely to have a number of demographic, social, economic, psychological, environmental, and epidemiological problems for the elderly population. Life becomes increasingly stressful during ageing process and elderly people become a liability to the families and communities involved. In rural community, family is the great single source of support and the centre of activity for most elderly people. The elderly need certain amenities such as health care, nutrition and a sense of belonging. But the type and amount of treatment they receive mainly depend on the culture of the family. Thus elderly cannot be viewed as a homogeneous group and be stratified into several categories based on the level of age, activity, morbidity and general health condition to facilitate an intensive study of incidence of disease, physical wellbeing and other related complications. One of the approaches towards studying the well-being of the elderly has been to study their "competence" in daily living. Such "functional competence" based approaches assess the ability to manage the tasks of daily living by the elderly, without dependence on others. Tribes are among the poorest and most marginalized sections of Indian society. Prevalence of poverty, low rate of literacy and a general lack of basic amenities constitute a major problem for the elderly population. This study aimed to sketch the demographic profile of tribal elderly people, and its linkage with their physical wellbeing and health hazards. This study is based on a sample of elderly tribal population, defined here as those who are 60 and above, taken from the densely populated &

235

tribal dominated region of northern India. Applying a multistage stratified random sampling procedure a sample of about 410 elderly people has been taken. The data at both macro level (at the household level) and at micro level (at individual level, elderly person concerned) have been taken. Basic instrument of data collection was a personal interview method. The detailed information have been taken on the demographic & socio-economic characteristics, health and other old age problems, duration and type of morbidity involved with the elderly, familial relations, etc. Data have been collected on household structure, household facilities to have an idea about the occupational, social and economic status of the household.

Results revealed that about 54% of the elderly lived together with their son/ daughter and a small proportion live with other relatives or alone. The most common health problems reported being asthma/ Respiratory problems (27.3%) and arthritis/ joint aches (36.1%). Near about half of the respondents preferred allopathic treatments for their health problems and about 10% of respondents use home remedy or indigenous medicines for their ailments as the first preference for treatment.

Keywords: Mortality, tribes, health-survey

236

Kutschar, P. (Institute of Nursing Science and Practice, Paracelsus Medical University Salzburg, Austria) Data Quality of Survey Research among Older Populations – Item-Nonresponse and Interviewer Effects in a Survey examining Pain of 655 Nursing Home Residents Potential threats of data quality are of special concern when it comes to survey research in nursing home residents (NHR), where dementia is deemed to be the most common disease affecting cognitive impairment. Recently, there has been a (re)growing engagement to include elderly with cognitive impairment in survey research [1]. Although the respondents’ tasks in the question-answer-process [2] have been broadly discussed, little is known about the specifics of cognitively impaired elderly. Many experts assume that surveys can be applied in elderly with up to moderate cognitive impairment [3]. However, the quality of data may be limited. To frame ‘survey quality’, Weisbergs’ Total Survey Error approach is used [4]. This talk focusses on ‘Response Accuracy Issues’, which comprise error sources due to incomplete measures, and to respondent/interviewer characteristics.

Data come from the health services research project ‘Action Alliance Pain-free City Muenster, Germany’. Questionnaires were used to assess the pain situation in 13 nursing homes. According to their cognitive function, NHR were stratified into NHR with no/mild (Mini-Mental State

Examination MMSE18-30) and NHR with moderate impairment (MMSE10-17). Data was collected by research assistants (RA). Item-nonresponse rate (INR) is predominantly used to assess data quality. It is hypothesized that INR varies in relation to survey duration, cognitive function, age, and further characteristics of NHR. Also, it is assumed that response distributions differ between interviewer-respondent-dyads (e.g. RAmale/NHRmale, RAfemale/NHRmale).

Results portend a significant INR-difference between NHR with no/mild (4.1%) and NHR with moderate (11.9%) cognitive impairment. The likelihood of INR>5% for NHR with moderate cognitive impairment is 3.8-times of that for NHR with no/mild cognitive impairment. While neither survey duration nor age exhibit remarkable effects, being in pain (OR=0.50) and being male (OR=0.67) seem to lessen the chances of INR. Furthermore, NHR’ response distributions significantly vary by interviewer gender. Female NHR interviewed by male RA report fewer pain- associated situations, less pain and less interference in daily activities than those interviewed by

237

female RA. In male NHR, response distributions show less pain, less interference and less frequent anxiety/depression states if questioned by male RA.

Standardized surveys seem adequate for NHR with no/mild cognitive impairment but INR-results cautiously indicate that data quality is threatened in NHR with moderate impairments, potentially limiting generalizability and validity of responses. Moreover, interviewer gender effects seem to be another considerable source of error. Findings affirm the necessity for further sophisticated research on survey methodology in elderly with restrictions in cognitive function.

Reference: [1] Krestar, M.L., et al., Including Individuals with Memory Impairment in the Research Process: The Importance of Scales and Response Categories Used in Surveys. Journal of Empirical Research on Human Research Ethics, 2012. 7(2): p. 70-79. [2] Schwarz, N., Cognitive Aspects of Survey Methodology. Applied Cognitive Psychology, 2007. 21: p. 77-87. [3] Clark, P.A., et al., Consistency of information from persons with dementia: An analysis of differences by question type. Dementia, 2008. 7(3): p. 341-358. [4] Weisberg, H.F., The Total Survey Error Approach. A Guide to the new Science of Survey Research, 2005. Chicago, London: The University of Chicago Press.

Keywords: Data Quality, Elderly, Response Accuracy Issues (TSE)

238

Lamprianou, I. Errors and Biases in mixed-mode administration Reaching informants through gatekeepers (e.g. accessing children through parents’ consent) has been discussed thoroughly in qualitative research (see, for example, Wanat, 2008). Ge&ng access to participants is important for the validity of qualitative research, but it can be equally important for quantitative, survey studies (Buchanan and Bryman, 2007). Lindsay (2005), is one of the very rare examples of researches which deal with the complex and +me-consuming problem of gaining access through gatekeepers in survey research. Motivated by the discussion above, our study uses an elaborate experimental design to measure the impact of sampling individuals through gatekeepers, on the findings of a survey, when different modes of administration are used. The context of the study: The research described here was part of a wider project funded by the University of Cyprus. This project used samples of students, parents, administrative staff and aca-demics to investigate familial involvement in the academic and social lives of students. The substantive aim of the project was to investigate the degree to which parents are involved in the lives of their children, during their University studies. Also, to investi-gate the perspective of the various stakeholders regarding the effect of familial in-volvement. Methods: For the purposes of this particular experiment, we had three samples: 1. We drew a random sample of University classes, and we asked the students to for-ward a questionnaire to their parents. We sampled 767 students and 432 of them (around 56%) forwarded the sealed envelope to their parents (note: it was not possible to know if they had used the right address). Out of those 432 parents who had – most likely – received the letter, 125 (29%) actually completed and returned the question-naire. This is the “gatekeeper sample”. 2. We asked the University to sample 400 random students and post to their parents the questionnaire; 75 parents completed and returned the questionnaire (around 19%). This is the

239

“indirect random sample”. We boosted this sample with additional 28 parent questionnaire from a near-by University using the same method. 3. We also conducted a telephone survey of the general population: out of 5184 telephone calls, 31% persons accepted to participate to the study. Out of those 1595 persons, only 429 (around 27%) had children which had completed University studies. This is the “direct random sample”. Results: We used descriptive and inferential statistics to compare the findings of the three samples. We also used Multi-group Structural Equation Models to study the structural invariance of the results. We found that the findings from the three samples have sig-nificant similarities but also some important differences. Discussion: At the end of the presentation, we will discuss the messages regarding the potential bias when sampling through gatekeepers in survey research. References Yarrow, M. R. (1963). Problems of methods in parent-child research. Child Develop-ment, 34, 215-226. Lindsay, J. (2005). Ge&ng the numbers: the unacknowledged work in recrui+ng for sur-vey research. Field Methods, 17 (1), 119-128. Wanat, C. L. (2008). Ge&ng past the gatekeepers : differences between access and coopera+on in public school research. Field Methods, 20 (2), 191-208. Cresswell, J. W., & Miller, D. L. (2010). Determiningvalidity in qualita+ve inquiry. Theory into prac+ce, 39 (3), 124-130. 5. Buchanan, D. A., & Bryman, A. (2007). Contextualizing methods 6. The Multi-Trait Multi-Error Approach to Estimating Measurement Error

Lamprianou, T. A. (Centre for Educational Research and Evaluation, Ministry of Education and Culture, Cyprus) The length of Likert response scale (number of options) and the quality of measurement: a double experimental design The optimal length of the response scale (e.g. Likert scales from 1-5 or Likert scales from 1-10) has been debated heavily in the last decades. For example, Kieruj and Moors (2010) investigated

240

the length of the response scale and found that it could be related to specific response style behaviors. Preston and Colman (2000) investigated the optimal length of response categories and found significant differences regarding reliability. Lozano et al (2008), for example, suggested that longer scales could lead to more reliable measures, however, other researchers do not agree with them (e.g. Jones and Loe, 2013).

Our study used two independent samples of students (N1=235 and N2=301) that responded to two, somewhat similar, questionnaires. The questionnaires asked them to rate the prestige of different occupations.

The first study was an elaborate experimental design where groups of students were administered a questionnaire in a section-by-section style. In each section the same questions were administered to students, but with response Likert scales of different lengths (Section A: 0- 8; Section B: 0-3; Section C:0-1; Section D: mixed 0-3 and 0-8).

The second study, asked the same students to rate each of the occupations on various dimensions, such as occupational prestige, perceived income, education needed to exercise the occupation etc.

The study uses descriptive as well as advanced psychometric methods to investigate issues of reliability and validity of measurement. We found that longer scales indeed lead to more precise measures (smaller standard errors), thus they are more appropriate to rank order either occupations or respondents (for example, reliability reduces from 0.88 to 0.78 and 0.68 for 9 options, 4 options and 2 options). When respondents were asked to reply using shorter response scales, we came out that they had found it very difficult, and some had preferred to skip questions.

However, the second study suggests that there were non-negligible validity and reliability issues which emerged when we used the same length of response scale, but the wording of the question was slightly changed. Thus, our study suggests that researchers should consider more the validity of the questions, rather than the length of the response scale.

241

Landrock, U. (TU Kaiserslautern, Department of Sociology) How Real and Faked Interviews Differ in Explaining Political Participation In face-to-face interviews the interviewer plays an important role. On the one hand he can improve data quality for example by asking probing questions. On the other hand there is the risk that interviewers falsify interviews partly or entirely. This paper presents results regarding the research question, which effects faked interviews have on substantive findings in social science research; we use the example of explaining political participation. To analyze the effects of falsifications we use experimental data, because data sets with confirmed real and confirmed falsified interviews are rarely found. We obtained two datasets. For the first dataset our interviewers conducted about 700 real face-to-face interviews. For the second dataset the same interviewers fabricated survey data in the lab: The falsifying interviewers were briefed about the sex, age, place of residence, and other socio-demographic characteristics of the 700 real survey participants. Then the interviewers were instructed to fill in the questionnaire like a person with the given characteristics would probably do. We calculate multivariate regression analyses and compare the results for the real and the faked interviews. The dependent variable measures political participation. The questionnaire covers 12 activities of political participation, adopted from the German General Social Survey (ALLBUS) 2008. The explaining variables are organized in three blocks. The first block covers the theoretically expected determining factors for political participation as described by Opp and Finkel (2001), the second block consists of socio-demographic and control variables provided to the falsifiers, and the third block refers to further situational relevant correlates found in research. The main results are the following: For the theoretically predicted determinants there are more significant effects in the real than in the falsified data. The falsifying interviewers probably do not know the underlying theoretical bases of a study. Therefore they do not reproduce relationships that are explained by these bases and that are empirically proved in reality. For the real socio- demographic information known to falsifying interviewers we had expected, that there were more significant effects in the falsified than in the real data. Previous research has shown that falsifiers overestimate the existing socio-demographic differences between the respondents. We cannot confirm these findings. There are further correlates found in research on political

242

participation which enter the bases of scientific knowledge. They mostly have an influence in specific contexts, depending on the concrete form of the analyzed political participation. For these further correlates we found more significant effects in the falsified than in the real data. That means that these influences are principally guessable by laymen. All in all, the falsifiers construct more consistent cases and invent more consistent models of respondents’ behavior than the real respondents show; therefore the explained variance is higher in the falsified than in the real data. These differences can be explained by our theoretical assumptions. Keywords: interviewer falsifications, falsified survey data, data quality

243

Landrock, U. and Best, H. (TU Kaiserslautern, Department of Sociology)

Selecting a Sample for Qualitative Interviews on Corruption This paper presents our procedure to find the ideal sample for qualitative interviews in our ongoing research project “RiKo – Risk management of corruption”. One aim in this project is to study the perception and evaluation of the risk of corruption of people in Germany. We use a mixed-mode design. In a first step we conduct a representative survey with 2.003 participants (CATI), in order to obtain an overall picture of attitudes and perceptions. In a second step we conduct qualitative interviews to research the subjective perceptions and interpretations in more detail and to identify patterns of interpretation. These qualitative interviews, conducted with a so far unknown number of survey respondents, are needed to reach the intended depth of understanding. However, they are also supposed to cover the full range of common interpretative patterns in the population so that the results are generalizable. To find the ideal sample size and composition we calculate a hierarchical cluster analysis as an explorative method of data mining to identify different groups of respondents which differ in central content-related characteristics. We decided to include the perceived occurrence of corruption in different institutions and organizations and the perceived competence of these institutions and organizations to prevent corruption as cluster variables. As a result we identify six clusters. Based on these detected groups we select test persons for the qualitative interviews, defining that each cluster needs to be represented to achieve an adequate diversity. As additional criterion to identify appropriate test persons we consider their place of residence as particularly important, given that our study is focusing on corruption in local administration as well as in small and medium-sized enterprises. This criterion includes the distribution over the different geographical regions of Germany on the one hand. On the other hand we take into account the density of population, that is the urban or rural character of regions. We assume that both make a difference for the appearance and perception of corruption. The Federal Statistical Office and Eurostat offer information about the degree of urbanization on the local administrative units differentiated in three categories as densely, intermediate or thinly

244

populated areas. We use this information to assign the degree of urbanization to the respondents of the representative survey. To complete the composition and find the ideal sample size for the qualitative interviews we use the affiliation to the six clusters as well as the assignment to the three types of population density, achieving 18 combinations. Accordingly we conclude that the sample size contains of 18 test persons each representing one of these combinations. In a last step we map all respondents on a map of Germany, using the program R. This allows us to select test persons from seven regions of Germany. This on the one hand ensures decent regional variation and on the other hand each region contains a large number of potential participants, so that in each region several interviews can be conducted and travel time and expenses can be limited.` Keywords: sample size, sample composition, generalizability

245

Lang, V. (Bielefeld University) Scaling Sensitive Factorial Survey Analysis Over the last decades factorial survey experiments (FSEs) have become increasingly widespread and successful for analyzing attitudes and behavioral intentions. In FSEs ratings of multidimensional treatments embedded in textual scenarios called vignettes are measured. That these ratings are interval scaled is a fundamental assumption of standard factorial survey analysis. Past research indicates that this assumption is problematic. Therefore, the following article develops a design for interval scaling tests in FSEs and a method for analyzing FSEs which is sensitive to the scaling level of ratings. An exemplary application of scaling sensitive factorial survey analysis in comparison to standard methods yields about a sixth larger effect sizes with regard to the treatments used in the FSE, and about a third larger effect sizes with respect to influences of between respondent differences. The new method also enables the evaluation of interval in contrast to non-interval scaled rating behavior.

Keywords: factorial survey experiment, interval scale, experimental design, survey data quality, structural equation modeling

246

Lang, V. and Diewald, M. (Bielefeld University, Germany) The added value of twin data analyses for causal inference in the social sciences Over the last four decades analyses of twin data have become an increasingly successful and widespread causal inference strategy in behavioral genetics. This progress can be beneficial for all social sciences, especially given the growing availability of larger twin data samples and the “environmental turn” in behavioral genetics. Since the 2000s research interests and methods in behavioral genetics have increasingly moved away from demonstrating heritability estimates of traits towards strategies controlling for genetic confounders in order to identify unbiased environmental effects. Therefore, the aim of this contribution is to popularize behavioral genetic modeling in the wider social sciences community. Starting from traditional family fixed-effect and hierarchical mixed-effect models we introduce behavioral genetic models as extensions for improved causal inference. In a first step, we here focus on applications to cross sectional data. In behavioral genetics twin data analyses are mainly based on structural equation modeling and twin data arranged as one data row per twin pair or twin family (“wide formatted data”) which is partly conceptually different from the hierarchical mixed-effect approach using data arranged as one data row per person (“long formatted data”). We clarify the similarities and differences between both modeling strategies and demonstrate that they lead to identical results. The central feature of both approaches to behavioral genetic analyses is the ACE-decomposition: Given a variable with within twin family variance behavioral genetic models decompose the overall variance of such a variable into an additive genetic (A) and a common environmental (C) component shared by both twins as well as a non-shared environmental (E) component unique to each twin. The E component is equivalent to the within family variance component in family fixed-effect or hierarchical mixed-effect models, while the A and C component sum up to the between family variance component. Hence, the AC-part of the ACE-decomposition disentangles the genetic and the environmental fraction of the between family variance component. Importantly for causal inference such ACE-decompositions are not only possible for outcome variables but also for treatments and their effects. Building on such ACE-decompositions of treatment variables we demonstrate how behavioral genetic models are

247

applied to identify causal environmental effects over and above those identified by purely phenotypic family analyses. As a working example we analyze the influence of parental social background and perceived child competencies on parenting behavior. For our analyses we use data from the first half of wave 1 of the TwinLife study. TwinLife is a twin family panel on the development of social inequalities in different life domains, specifically, educational and labor market careers, social participation, quality of life, health and deviant behaviors. For the first half of wave 1 about 2,000 twin families of four different twin birth cohorts (1990/91, 1997, 2003, 2009) covering the whole range of socio-economic backgrounds in Germany have been surveyed. Hence, the TwinLife sample combines the strengths of an extended twin family design regarding the internal validity of analyses and the advantages of a stratified probability sample with respect to the external validity of results. Keywords: causal inference, twin data, behavioral genetics

248

Langer, W. (Martin-Luther-University Halle-Wittenberg)

Longitudinal analysis of repeated cross-sectional surveys: The usefulness of the multilevel model

Panel surveys are seldom made in the social sciences. Cross-sectional surveys mainly dominate applied research. Some of them are repeated at certain timepoints especially every month, every year or biennially. On condition that they are using the same measurement instrument, interview mode and sampling design, this lecture shows how to use multilevel models to analyze these repeated survey data for a longitudinal study. The author presents a multilevel conditional multinomial logit model estimated with GLLAMM in Stata 14. He examplarily demonstrates how the effect of long term party identification on the intention to vote for a certain party changes over time using the German Politbarometer poll with 416 monthly repeated surveys from March 1977 to December 2013 at which half a million respondents took part. On one hand the variation within a single timepoint is explained by exogenous variables of the respondents (level 2) especially the social and religious cleavages. On the other hand the variation of the effect of party identification between timepoints is explained by the growth of the domestic product, the unemployment rate. The model also estimates the effects of the substitution of face-to-face interview mode by CATI in august 1988 and the German reunification in october 1990 (level 3). Afterwards the advantages and disadvantages of the proposed repeated survey analysis are discussed.

249

Law, C. (De Montfort University) Researching Reproduction with Men: Methodological Reflections Social science research into reproduction is a vibrant and growing field of scholarly activity. However, the majority of research is conducted with women and focusses on women’s lives. This reinforces the notion that reproduction is located primarily in the female domain and overemphasis female responsibility, leaving men’s behaviours relatively neglected. Several scholars have argued for an expansion of reproduction research to include a focus on men and men’s experiences (Culley et al., 2013; Lohan, 2015). However, despite a growing interest in men’s experiences in recent years, the majority of reproduction research involving men has been concerned with fatherhood itself, leaving numerous other aspects of the reproduction process overlooked including pre-conception desires for parenthood and planning (Morison, 2013; Lohan, 2015). Furthermore, methodological literature on researching reproduction with men remains sparse. This paper presents early reflections from a PhD study into childless men’s perceptions and intentions regarding reproductive timings. It considers methodological aspects in the context of the first phase of recruitment and data collection, and will explore two main issues. Firstly, recruiting men for reproduction research can present challenges. As reproduction is traditionally considered a female domain, men may be disinclined to take part in such research and may not identify it as relevant to their lives. Furthermore, studying the pre-conception stages of reproductive planning with men who are presumed fertile necessitates the recruitment of individuals who are not a unified group and who cannot be recruited via a specific organisation or network. This paper explores these challenges, as well as strategies evoked to address them, and discusses resulting dilemmas, opportunities and concerns, particularly regarding recruitment through personal networks. Secondly, the paper addresses the issue of gender in the research process. Participants and researchers are gendered individuals and the research interview a gendered encounter, which has consequences for the collection, analysis and interpretation of data. Past literature suggests that female researcher-male participant interviews may be perceived as opportunities and/or

250

threats, may provoke anxieties and defensiveness and may result in heterosexual displays of behaviour (Hearn, 2013). The paper reflects upon the first phase of data collection and on the performance of gender within interviews. It considers the implications of this for the study specifically and for reproduction research more generally, particularly in relation to feminist methodology. Extending reproduction research to better include a focus on men and men’s lives requires a reflection of the methodological challenges inherent in such research and discussion regarding how these can be addressed. This paper aims to contribute to discussions regarding the complexities regarding undertaking reproduction research with men, and to offer a detailed consideration of some of the associated challenges, opportunities and dilemmas.

Keywords: men, masculinities, reproduction

251

Layne, L. (Colorado School of Mines) ‘A Bigger Picture’: The Ethnographic Case Study as an Under-utilized Method in Studies of Reproduction In this paper I draw on my on-going, long-term, in-depth case study of one American Single Mother by Choice to describe make the case that the ethnographic case study, is an important but underused method for studies of reproduction. I show how this approach complements the more common methods (clinic ethnographies, semi-structured interviews, questionnaires, textual analyses) and offers a unique contribution to the anthropological, sociological, and historical literatures on reproductive technologies and the emergent family forms they help create. Social commentaries which seek to explain recent changes in reproductive patterns, do so based on a comparison of statistical snapshots. I argue that a life-course approach reveals a more complex picture of these emerging family forms. Although cases studies are sometimes scorned because of their lack of representativeness, and generalizability, ironically, I argue they offer in ways much like some of the techniques David Hockney uses to create “a bigger picture.” One of these techniques he developed In his quest to be able to “depict the world as it is actually experienced” is “reverse perspective,” in which he layers “space upon space,” stacks “horizon upon horizon” (2013:30). This is similar to the technique I use in my depiction of Carmen’s reproductive history and ever-changing family life. Another Hockney technique that helps illuminate the method I adopt is the one he used in his landscape paintings of 2005, i.e., “mounting multiple canvases on easels, one beside the next…creating vitas that were not just bigger and wider, but that featured multiple overlapping vanishing points, pulling the viewer ever more actively into the scene” (Weschler 2013:35). The technique for producing these ‘combines’ -- setting up canvases side by side-- is not unlike the technique Carmen and I are using, working side by side, in crafting an in-depth, longitudinal account of her experience as a single mother by choice. Though the final product, like Hockney’s combines, is recognizable as a single landscape, it profits from the wider vantage that comes from working together, from similar, but still inevitably different points of view.

252

Keywords: case study, David Hockney, reverse perspective, multiple perspectives , combines, 'bigger picture'

253

Lazarevic, P. (TU Dortmund University) Rating Your Health: An Examination of Non-Health-Related Factors and Differential Item Functioning in the Self-Rating of Health Comparative analyses frequently base on respondents’ self-rated health (SRH), assuming that SRH is comparable between different groups of respondents. However, this comparability is questionable due to manifold non-health-related factors (NHRF) such as age- or gender-specific expectations for ones’ health and frames of reference, or cultural contexts. If SRH is examined, researchers often distinguish between two components of the respondent’s answer: latent health and reporting behavior. While latent health exclusively refers to a person’s health status, reporting behavior results from a wide variety of NHRF such as general optimism, education, the interviewer, or cultural factors. The analyses of this paper base on the information of 14,400 participants of the fifth wave of the Survey of Health, Ageing and Retirement in Europe (SHARE). First, we predicted SRH via linear regression using a latent health model comprising a wide range of health factors. Second, we used a reporting behavior model covering NHRF to explain the residuals of the first regression. To identify differential response behaviors, all analyses were carried out separately by gender and three age-groups (50–64, 65–79, and 80–90 years). The latent health model shows only minor differences between men and women in explaining SRH, although almost all effects are significant regardless of gender. The amount of explained variance is high regarding all models (with a mean of 46%). The NHRF-model identified consistent influences of interviewer’s SRH, respondent’s life satisfaction, and country of residence on SRH, while other factors differed by gender and age-group. The amount of explained variance due to reporting behavior in all models can be considered meaningful (with a mean of 9%), given the amount of explained variance by health-related variables. These results illustrate the importance of taking factors such as gender- and age-specific response behavior and NHRF into account while utilizing SRH. Keywords: Epidemiology, Interviewer Effects, Reporting Behavior, Survey Methodology

254

Lefeldt, J. (Johannes Gutenberg-Universität Mainz, Germany) Stories behind Pictures. The Photograph as a Medium of Memory History can be documented using visual data such as photographs. The stories behind pictures, the individual and/or collective experiences and meanings, however, usually remain hidden for the observer. At the same time visual media can function as a sensual link to personal memories. Intending to make this connection between pictures and memory profitable for anthropological research a project on the history of Sudeten German organizations was realized at the University of Mainz and in cooperation with the upcoming Sudeten German Museum in Munich. The concept of the project arose in the course of considerations on the existing exhibition material for the Museum which mainly contains photographs and other plane objects. Using pictures in an exhibition demands contextualization and a reasonable preparation. One possible response to these requirements is the questioning of witnesses. For one, their knowledge is capable of providing further information about the geographical and temporal placement of the objects. On the other, their individual stories could enrich the presentation in the museum using audio points in addition to the visual objects. In the course of the project, leading individuals of Sudeten German organizations were interviewed on a sample of pictures from the early 1950s until today. Combining the interview methods of photo elicitation on the one hand and biographical questioning on the other made it possible to directly compare the two. The analysis of these complementing approaches indicated differences concerning the structure and complexity of the nar- rated memories. In the biographical part the respondents would generally talk about their life in a chronological and summarizing way, whereas the presented pictures initiated more specific aspects and personal experiences that often caused a more emotional and detailed form of narration. Even pictures the respondents classified as unfamiliar could lead to personal memories. Stories associated with pictures are therefore not necessarily connected with what is visually perceptible. Visual media such as photographs are able to elicit memories. At the same time, they affect the way of remembering and the expression of memory.

Keywords: Non-Moving Visual Data, Photo-Elicitation, Oral History

255

Leguina, A. and Miles, A. (University of Manchester, England) The temporalities of cultural participation practices in the UK: life course dynamics and mobility effects ‘Understanding Everyday Participation - Articulating Cultural Values’ (UEP - http://www.everydayparticipation.org) is a five-year research project exploring the dynamics of, and stakes attached to, informal and ‘unofficial’ participation practices in the UK. It puts particular emphasis on the relationship between participation and place, and between patterns of participation and life course trajectories. In this paper, we explore several questions about the intersections of biographical, generational and historical change in the process of ‘cultural class formation’, viz: How far is cultural consumption in adulthood determined by consumption during adolescence? How do living conditions during adolescence impact on cultural consumption during adulthood? And, in particular, what is the effect of social mobility on participation patterns across lifestyles? For this, we use data from the 1970 British Cohort Study (BCS70), which covers health, welfare and a variety of life aspects for a sample of approximately 10,000 individuals born in 1970. In the case of cultural consumption, BCS70 contains information on participation in a series of activities at 16 and 42 years of age. Besides providing answer to these research questions, this article offers an insight to the methodological and substantive challenges of measuring how cultural consumption as a social process varies over time for a particular cohort. Using information from several ordinal indicators of cultural consumption measured at 16 and 42 years of age, the first stage of our analysis builds scales of engagement that represent the intensity of participation in specific combinations of activities which compose lifestyles. These scales, defined as latent variables, are built through confirmatory factor analysis. Validity and consistency of the measures are studied in detail. In a second stage, the effect of a series of sociodemographic variables on lifestyle engagements is tested through a structural equation model. Here, we emphasize on the relevance and impact of different types of social mobility on patterns of participation.

Keywords: Cultural consumption, life course, quantitative methods, UK.

256

Leszczensky, L. (MZES, University of Mannheim, Germany) Can Panel Data Resolve Questions of Causal Ordering–and If So, How? Does X causes Y, or is X caused by Y? Many social scientists turn to panel data to resolve such questions regarding the direction of causality. But alas, panel data are hardly a panacea for establishing of causal order. After all, like all models, panel models rely on assumptions that may or may not be met in the data. And researchers often tend to ignore these assumptions, let alone test them. Recognizing that causes precede effects, a key approach for addressing questions of causality is to test whether lagged values of the “independent” variable explain contemporary values of the “dependent” variable. While this approach makes sense intuitively, it still comes with strong assumptions that researchers often do not make explicit, let alone test. For example, both Random-Effects (RE) and Fixed-Effects (FE) models require strict exogeneity; i.e., the idiosyncratic error term has to be independent not only of past and current but also of future values of the independent variables. In other words, the assumption of strict exogeneity rules out reverse causality, thus making it impossible to gain unbiased estimators of reciprocal causal effects. Simply lagging the independent variable therefore does not solve the problem of endogeneity caused by reverse causality. As shown by Allison (2009), with at least three time points available, the assumption of strict exogeneity can be relaxed by allowing correlations between the idiosyncratic error term and future values of the independent variables. This is achieved by lagging the independent variable and taking first differences. Such a lagged first-difference model (LFD) thus makes it possible to estimate lagged reciprocal effects while also accounting for time-invariant unobserved heterogeneity. However, Vaisey and Miles (2014) recently showed that the approach suggested by Allison (2009) may considerably fail in practice. In short, their argument is that LFD models require rather strong assumptions about the precise temporal nature of the estimated effect. By contrast, as their simulations show, if the lags between panel waves do not match the actual causal lags in the process under study, results of LFD models may be highly misleading.

257

Unfortunately, researchers usually do not know how real-world causal lags and panel lags relate to each other. Vaisey and Miles (2014) therefore warn against using LFD models to determine causal ordering–which, of course, is precisely what Allison (2009) proposed these models for. Finally, Allison (2014) recently proposed an alternative to the LFD model that works without first differencing but still allows estimating alleged reciprocal effects. This alternative approach basically implements a cross-lagged panel model in the framework of structural equation modeling (ML-SEM) so that it can be estimated using maximum likelihood. Allison further proposed a test that can be used to assess whether panel gaps match the gaps of the real-world causal process, thus making it possible to evaluate whether the assumptions made by the LFD model or ML-SEM are appropriate. Using a substantive example with real data, I show results for allegedly reciprocal causal effects obtained by different panel models. Based on a three-wave panel data set, I not only compare conventional approaches like RE- or FE-models, but also less well-known extensions such as LFD or ML-SEM. In particular, I discuss the assumptions that this models rely on as well as the plausibility of these assumptions.

Keywords: Panel Data Analysis, Reverse Causality, Fixed Effects

258

Letterman, M. C. (RTI International) Use of Geographical Information Systems (GIS) for Sampling and Fieldwork Monitoring in Developing Countries RTI International has recently used GIS-based sampling on a number of household survey projects in developing countries where traditional sample frames are lacking. This approach presents substantial cost and time savings compared to designs that require a full listing of households. Coupling passive collection of GPS data during interviews with a GIS-based sampling strategy allows researchers to remotely monitor fieldwork to ensure that interviews are occurring in the correct locations.

Two important questions must be explored to better understand the opportunities and limitations of GIS sampling, and the use of passive GPS capture for quality control purposes.

1. How does the accuracy of GPS location information vary across environments (urban areas, mountainous terrain, dense jungle, etc.), i.e. how does GPS instrument error effect interview location information? 2. How well do interviewers perform when required to use a GIS based sampling methodology, i.e. how much unintentional error and purposeful falsification can be detected using this approach?

For this study, we examine the recorded GPS coordinates of data collected in Guatemala for a household survey conducted in 2015. Interviewers were instructed to attempt interviews with all households within secondary sampling units (SSUs) that were either 50m x 50m or 100m x 100m in size. These SSUs were located within 1km x 1km primary sampling units (PSUs), which were allocated across the country proportional to population size, based on figures from the Landscan population database. Interviewers were given identical GPS enabled smartphones and mapping software which allowed them to navigate to the PSUs and SSUs. The PSUs for this survey included densely populated cities, mountain villages, and rainforest locations with dense tree cover. The diversity of environments encountered lends itself to the study of environmental effects on the use of GIS sampling and GPS for quality control purposes.

We analyze interview location information at both the environment and the interviewer level to better understand the effects of both environment and interviewers on adherence to the sampling plan. Specific research questions include:

259

1. How does the percentage of interviews recorded as occurring within the selected sampling unit (SSU/PSU) compare across the different environment types? 2. How do the percentage of interviews recorded as occurring within the selected sampling units (SSUs/PSUs) differ across interviewers? 3. Are the recorded GPS coordinates of interviews related to other data quality characteristics that could indicate interviewer error or falsification, as opposed to issues with GPS accuracy? keywords: geographical information system, GIS, GPS, quality monitoring, data collection, fieldwork, developing countries

260

Lin, J.-P. (Academia Sinica, Taiwan) Data Science as a Foundation of Integrating and Enriching Administrative Data: The Case of Constructing Taiwan Indigenous Peoples open research Data (TIPD) Based on Taiwan Household Registration Administrative Data 1 Aims: the research aims are threefold: to (1) demonstrate the methodology of data science in constructing Taiwan Indigenous Peoples open research Data (TIPD, see https://osf.io/e4rvz/) based on Taiwan Household Registration (THR) administrative data; (2) to illustrate automated or semi-automated data processing as a determinant of constructing effective open data; and (3) to demonstrate appropriate utilization of “old- school” data format such as multi-dimensional tables as an effect means to overcome legal and ethic issues. 2 Data & Methodology: the main source data utilized to construct TIPD are THR administrative data. THR administrative data that are mainly collected for vital official statistics comprise (1) individual name, personal PIN, birth date, death date, marriage status, education, age, birth place, migration and residential mobility, employment, relationship to household head, spouse name, father name, mother name, ethnicity, etc., at micro level and (2) household information like household head name, administrative area codes, address, size, etc., at aggregate level. Legal & ethic issues are top priority in the research. Source data processing tasks are conducted in a closed supervised data lab of funding agency. The research extracts valuable information embedded in micro data of THR and enriches the extracted information through the processes of cleaning, cleansing, crunching, reorganizing, reshaping the source data. The data enrichment processes produce a number of data sets that contain no individual information but retain most of source data information. The enriched data sets thus can be open to the public for open administrative data study. The open data are systematically constructed in an automated or semi- automated way through the integration of compiler programming language, software, and script languages. 3 Results: major outputs of TIPD amount to 7,300 files in number and around 32 GB in size. TIPD now consist of three categories of open research data: (1) categorical data, (2) household structure and characteristics data, and (3) population dynamics data. Categorical data include

261

two broad dimensions. The first one is contingency tables, while the other is multi-dimensional data. Household structure and characteristics data consist of three dimensions of information: (1) household head information, (2) household member composition information, and (3) household geographical information. Population dynamics data consist of three categories: (1) increased population within a given period of time; (2) decreased population within a given period of time; (3) intact population within a given period of time. To assess the quality of constructed open data, the author proposes the following methods: (1) the source codes of programming languages and date processing procedures that construct open data should be open to the public; (2) although the source administrative data can’t be open, we can create a simulated administrative data that resemble the source data in terms of contents. The simulated administrative data serve as a source that allows us to examine the potential flaws and bugs of source codes used to construct open data. Keywords: data science, administrative data, open data

262

Lineback, J. F., Ashmead, R. and Slud, E. (U.S. Census Bureau) Using Propensity Score Weighting Adjustments to Examine Survey Mode Effects In this presentation, we discuss a case study using propensity score weighting adjustments to examine mode effects in a multi-wave survey of U.S. commodity flow. Propensity score weighting adjustments (Lunceford and Davidian, 2004) allowed us to re-weight responses so that their observed covariate distributions were equivalent across modes, essentially asking what the outcomes would have looked like if all responses had been in a single mode. This is a useful tool for comparing modes without the benefit of a designed experiment. Without these adjustments, it is possible that any observed differences in the outcomes between modes were due to different covariate distributions rather than differences between response modes. In our example, we compared adjusted estimates of unit missingness, item missingness, edit counts and imputation counts. Although this was an analysis of continuous data items for business establishments, these methods easily can be extrapolated for social surveys.

Keywords: propensity score weighting adjustments, mode effects, survey measurement

263

Linklater, A. (Aberystwyth University, Wales) Political Symbols and Civilizing Processes in Long-Term Perspective Norbert Elias invented the term, symbol emancipation, to explain a key feature of long-term processes of human development – the gradual shift in the balance of power between genetic restraints on behaviour and the cultural constraints that humans imposed on themselves and each other. Symbol formation was central to the process in which ‘biological evolution’ was replaced by cultural development’ – in which cultural transformations occurred without changes in the biological constitution of the human species.

The importance of symbol formation has been emphasised by the study of material symbols in recent archaeological writings and in reflections on ethno-symbolism. Those studies have discussed core features of human development such as the creation of symbolic boundaries between ‘insiders’ and ‘outsiders’ that have been interwoven with licensing acts of violence against enemies that have been prohibited in relations within the ‘in-group’.

Of particular importance have been the social construction of symbols that have defined societies that regarded themselves as civilized, and the creation of symbolic boundaries that separated the ‘civilized from the ‘savage’. But ‘society’ is not the appropriate level of analysis for the study of symbols of civilization and savagery that are central to any ‘civilizing process’. As Elias argued, societies are always bound together in international figurations. The symbols that unite and divide human groups can be analysed in different types of international order.

Throughout human history, exclusionary symbols that divided social groups have been more powerful than inclusive symbols that bind peoples together in ‘diplomatic communities’. In the modern period, nationalist symbols such as the ‘tomb of the unknown warrior’ have had greater emotional resonance than international symbols such as the Red Cross or the ‘blue berets’ of UN peace-keeping forces.

264

Territorial monopolies of power have been able to create national symbols; international agencies have been weak by comparison though elements of ‘post-national’ symbolism are not entirely absent.

The prospects for cosmopolitan symbols that are evidence of a global ‘civilizing process’ seem low. Will national symbols remain central to political responses to lengthening and deepening webs of interconnectedness? Are more inclusive, cosmopolitan symbols destined to lag behind? This paper discusses how the social sciences can develop insights into relations between intra- and inter-national symbol formations in the history of human groups.

265

Liston, K. and Maguire, J. On Being ‘Irish’: negotiating identity and the Involvement-Detachment balance in archival Research This paper will take up the call from Rojek (and others) in the 1990s for greater attention to be paid to the actual practice of involvement-detachment in figurational / process research (Elias, 2007). Drawing on material from ongoing archival research on the involvement of ‘Ireland’ in the British Empire Games, the authors will first outline their own ‘established-outsider’ biographies. This is important in light of their insights that follow on the challenges of maintaining relative detachment while immersed in archival material dominated by the ideologies of Pax Britannia on the one hand and political and cultural nationalisms on the other. Here, the involvement- detachment balance has fluctuated during the search for ‘the smoking gun’. The paper also reflects on the role of secondary re-involvement and the satisfaction to be gained from the generation of a ‘relatively adequate’ reality-congruent picture of the past, albeit one that is inevitably shaped by that which is both present and absent in various archives and historical sources.

266

Lotto, R. (Liverpool John Moores University) Supporting the researcher in sensitive research – a case study of patient and public involvement Background As a novice researcher, designing a qualitative study around the ethically sensitive and emotionally charged area of termination of pregnancies affected by severe congenital anomalies has proved challenging. Whilst there is a growing body of literature exploring the risks to the researched, the impact on the researcher is something that remains highly relevant yet under- investigated. This abstract provides a reflection on the impact of patient and public involvement (PPI) in reducing the risks to the researcher in the context of this highly sensitive research study. Patient and Participant Involvement There is a growing body of literature highlighting the impact of PPI in research. This has subsequently been translated into policy, where NHS guidance on research governance highlights the need for PPI at every stage of research. Whilst research funding is now often attached to the proviso that PPI is incorporated, this can risk tokenistic involvement. Despite the potential benefits, involvement is complex and challenging. There may be resistance from researchers who doubt the ability of participants to contribute in a meaningful way to academic research. In addition, participants themselves may not feel able to contribute authoritatively due to unfamiliarity of research methodologies.

Reflections Reflecting on my study, two main areas of impact were identified; navigation of governance processes and creation of opportunity to pursue the research interest; and benefits in preparing and supporting the researcher emotionally.

Navigation of Governance Issues raised by the Research and Ethics Committee (REC) focused on the unorthodox design of the information sheets, and recruitment strategies, in particular the appropriateness of approaching parents awaiting further testing where a severe congenital anomaly was suspected.

267

Evidence of the input of parents into both these aspects provided sufficient reassurance for approval to be granted without any changes. It has been suggested that sensitive research risks being ignored due to the inherent difficulties in accessing participants and gaining approvals. This experience highlights the positive impact that meaningful engagement has, in enabling studies that otherwise risk being vetoed, to be undertaken.

Researcher skills The opportunity to undertake practice interviews with parents who had ‘been through the system’ was particularly beneficial. Initial intentions were to engage with them to construct an interview schedule. However, I found myself being ‘trained’ by the parents. As I explored topics, unexpected issues were highlighted, from avoidance of particular terminology, to the use of silence. Of particular benefit was the insight I gained into how best to support my participants whilst desensitizing myself from the distressing content. Not only did this better prepare me to support participants, but I was able to absorb the experiences of parents with the benefit of an emotional distance.

Conclusion Within the context of this study, involving parents in the research process provided unique benefits which have not been previously documented. These related to ‘navigating governance’, where PPI enabled a divergence from traditional approaches, and ‘researcher skills’, where the unforeseen benefit of desensitising and emotionally preparing the researcher occurred through the presentation of scenarios and subsequent deconstruction of the interactions.

Key words: sensitive research, patient and public involvement, research governance

268

Love, G. (University of Sussex, England) Is every abortion a tragedy? Vulnerability and resistance in abortion research This paper employs Judith Butler’s (2014) reimagining of vulnerability and resistance to question the way abortion research is constituted by ethical review committees as ‘high risk’ and sensitive. It draws on doctoral research with women who have ended pregnancies in austerity Britain and the experience of engaging with medical institutional review boards (IRBs) as a qualitative social science researcher.

Hoeyer, Dahlager and Lynöe (2005) argue that bio-medical ethics locates the ‘sensitivity’ of research within the interpersonal relationship between researcher and participant, rather than between the research and its socio-political context. As a result, this paper argues, bio-medical IRBs encourage a framing of women who have had abortions as people researchers should expect to be emotionally distressed, and emphasise the researcher’s responsibilities towards managing this type of sensitivity. What is not often allowed for in bio-medical IRBs is the politically sensitive nature of abortion research; suggesting that women who have abortions should be expected to be emotionally distressed arguably reinforces anti-abortion discourses. As a feminist sociologist, framing participants in this manner in order to gain ethical approval feels uncomfortable.

This paper introduces a framework for abortion researchers to reconceptualise their ‘vulnerable’ participants buy using Butler’s reconceptualization of vulnerability as a ‘deliberate exposure to power,’ a condition that can facilitate resistance. In this way, the abortion story can said to be a performative act in that it at once invites naming – as a woman who had an abortion – and at the same time resists and complicates this designation. This paper argues that abortion researchers and IRBs should view participants through this double lens, as vulnerable yet at the same time agentic. They deliberately expose their vulnerability in order to take part in research, tell their stories, and resist societal injunctions that they remain silent about their experiences.

Reconceptualising vulnerability in this way offers a solution to the way in which social science researchers can be forced to conceptualise their participants in problematic ways by IRBs. It

269

enables research with hard-to-reach or hidden groups which recognises both their vulnerability, but also their agency.

References Butler J (2014). ‘Rethinking Vulnerability and Resistance.’ Talk given at the XV International Association of Women Philosophers, University of Alcalá, 24th June 2014. Hoeyer K, Dahlager L and Lynöe N. (2005). ‘Conflicting notions of research ethics: The mutually challenging traditions of social scientists and medical researchers.’ Social Science and Medicine 61(8): 1741-49.

Keywords: Abortion, ethics, vulnerability

270

Lübcke, T. (German Maritime Search and Rescue Service – DGzRS) Social research in the harm zone: How to follow in the wake of complex maritime search and rescue missions Exploration is a perfect use for qualitative methods (i.e. observations). First and foremost the exploration of new research fields requires special demands in preparation and research setup. To fill a pending research gap in organizational studies we took our first steps in the field of maritime Search and Rescue (SAR) in 2012. And indeed, there were several difficulties with the collection of data that we did not encounter onshore. On the contrary, this was an opportunity to refine our existing methods. One cornerstone of this research project was the observation of joint SAR exercises in the Baltic Sea held by the SAR organizations of all Baltic Sea neighboring states and France. We used the frame of the world’s largest international SAR exercise for our inquiries. Our focus was the (shared) decision-making during a huge mass rescue scenario that ran approximately six hours – thus the time frame for observations is very limited: one shot, one hit. Furthermore we had to keep track of actions on several vessels in a kind of “free game”: A general scenario was given, but the participating units were free to make their own operative decisions without any intervention by the exercise distaff. That issues into a maximum of contingency in progress. The complexity increased by parallel actions on changing locations. It’s superfluous to point out that our personnel resources – researchers are not always the best sailors – and recording equipment recordings were quite limited. Taking all these restrictions into account, we refined and extended the videography to be utilized in dynamic, unpredictable events to increase the research data outcome of short-term field visits. We pilot-tested our method first in a Danish navy exercise, modified the tool for the international SAR exercise 2014, and refined it again for an inquiry during the similar SAR exercise in 2015. However, we are going to expand our research in a new project – Mass Rescue Missions: In 2016 we started to come closer to the SAR missions practice. Guiding the rescue units/organizations in the Aegean Sea, the focus of our field research is away from the exercises on to hot SAR missions upon refugee aid in the Mediterranean. In this field we have “more shots” – several single

271

missions over a couple of hours – but other problems have arisen. Foremost among these are ethical issues dealing with an immense need for research. The researcher’s role must be ethical and positioned in this context: a researcher onboard a rescue unit limits its capabilities to save lives at sea. Otherwise the researcher has abandon the role of observer to become an active part of the rescue team. Both versions restrict the opportunities for ethnographic work in this field: either to influence the work(-ability) and the field by being present onboard, or to increase the bias of going native by being part of the team. This influence on the method needs to be handled – participant observation or observing participation. But there are more ethical aspects to the situation. The researcher has to take more than the rights of the refugees themselves into consideration. Rescue personnel are vulnerable to being traumatized on these missions – e.g. when children die in their arms or a rescue person has to decide who to save and who to leave in the water. We will discuss how trustworthy data of rescuers’ individual behavior under these restrictions are, as how to improve the research method to consider the entire ethical landscape. This is what our paper/presentation will discuss.

272

273

Lübcke, T. (German Maritime Search and Rescue Service – DGzRS) Videography in dynamic environments – restrictions on and opportunities for multiplicative GoPro camera usage Closely coupled to workplace studies, videography is also utilized in safety science and human factors studies. First and foremost, individual behavior and actions in complex socio-technical work environments are fruitful objects of research often finalized in case studies. The ongoing technological progress in recording devices have created new opportunities for field research in a minimally invasive manner. For instance, outdoor video cameras (action cams) are now small enough for disaster relief forces to wear on their missions. This offers new ways to bring researchers closer to the agent’s activities and to immerse themselves into action structures in ways that other observation techniques cannot allow. It is easy to minimize observer bias by using small wearable cameras. In so doing, the researcher has more capabilities for other necessary tasks, such as following simultaneously occurring actions that would ordinarily require a lot of staff resources. As promising as action cams are as a tool for field research, less intuitive is their usage in research practice compared to their use in leisure and sports. The reason is apparent: Action cams are designed for sports and fun – with a focus on aesthetics in motion pictures or photos. They were developed for filming under extreme and unusual conditions and above all to take beautiful pictures. The functional purpose of the devices is the collection of visual data, not on high-quality audio recording. Sportspeople may wear one or two cameras to record their own performance. More than two cameras at the same time were usually managed by professional media teams, specialized in the production of marketable action video content. If we want to utilize the capabilities of action cams for (social) research, the scientific user has a dilemma: The advantage of action cam technologies increases with the number of devices in operation. Conversely, the technological demands of the user are also increasing. At least, non- professional users – like most researchers – are capable of working with several action cams at once in unpredictable situations. We used action cams in a research project (2012-2015) on decision-making in maritime search

274

and rescue (SAR) missions to reconstruct the operational decision-making processes in an international SAR exercise. Faced with simultaneous challenges like shifting the operational language from English to Danish or switching from civilian to military procedures, we had to build up an expert based evaluating concept afterwards; based on a time-synchronized evaluation movie out of the recordings of four/six GoPro cams. We will discuss our experience, lessons learned and good practice from data collection to the preparation for analysis in the paper/presentation.

275

Mack, A. (Aberystwyth University) The Process of Migration: Identity, Affect and Established Outsider Relations This inquiry argues that the Winston Parva example, from Elias and Scotson’s The Established and the Outsiders provides a mode of analysis for understanding the process of migration. Elias’s theorisation of established and outsider relations is founded on an account of migration within the Winston Parva example. Although this was a survey of a small Leicestershire community in early 1960’s, it has a broader application to larger social units in their attempts to grapple with long-term social processes. Migration is a longstanding process of human movement, which brings into contact diverse groups of people into relations of interdependence that shapes their identities and feelings of belonging. It is a convergence that displays expressions of hostility and compassion, with the mobilisation, monopolisation and manipulation of affect, as demonstrated by the reactions of liberal democratic societies to the so called ‘refugee crisis’ in Europe and elsewhere. Ultimately, the inquiry illustrates that the so-called refugee ‘crisis’ is not a unique occurrence. The process of migration being witnessed in Europe and elsewhere is the meeting of established and outsider groups. Where the established demarcate the boundaries inclusion and exclusion between themselves and the outsiders, through the simultaneous creation and perpetuation of affect, through idealised group pride and the imposition of denigrating group disgrace.

276

Mahapatra, B. (Govt. of India, Deptt. of Statistics, (Retd. Indian Statistical Service officer)) Flexible Data Collection Methods and Data Quality In Survey Research Data collection on social aspects, is the process of gathering various information on some predetermined variables to be estimated, in a systematic manner. The data collection component of research is common to all fields of study including social sciences. While methods vary from variable to variable, the emphasis on accuracy of data and honest method of collection remains the same. The goal for all data collection is to capture quality information which then translated to rich data analysis and allows the building of a convincing and credible answer to questions pertaining to the field of social science or otherwise. Proper questionnaire is very important in any data collection work. Exhaustive, to the point, non-offending and related to the variable(s) under study should be prepared before taking up any data collection work. Data can be on qualitative characteristic or quantitative one. Quality of data for research on social science depends on many factors, like: Informant’s cooperation and correct answer, Enumerator’s understanding of the concepts and definitions, Correct transcription from primary data, Suitable tabulation method, Dual meaning of concepts and definitions, Correct observation particularly for qualitative data.

The methodological aspect on data collection on socio economic indicators can be done in the following manners: Personal interview method, Mail order enquiry, Web based data collection, Observation method, Transcription method, Through Apps using smart phone.

This paper discusses various issues and challenges relating to these methods.

Keywords: Data Collection methods, data-quality, questionnaire-survey

277

Mahapatra, T. Qualitative research My paper will present a theoretical study of different research paradigms, the difference between qualitative and quantitative research and interviews as a research methodology. The purpose is to enrich the philosophical and methodological stances of social science research, which might be suffering from their so called bias for quantitative research methodologies. Paradigms are defined as the basic set of beliefs (or metaphysics) or even a world view that guides a researcher or investigator, not only in choices of a method, but ontologically and epistemologically in fundamental ways (Guba and Lincoln 1994). Guba and Lincoln (1994) affirm that both qualitative and quantitative methods may be used appropriately with any research paradigm, but the questions of method/s are secondary to questions of paradigm. The nature of value free inquiry based on a “God’s eye view of reality”- a hall mark of quantitative research- is challenged by qualitative research (Denzin and Lincoln 2005). The word qualitative implies an emphasis on processes and meanings that are not rigorously examined, or measured in terms of quantity, amount, intensity, or frequency. Qualitative researches stress the socially constructed nature of reality, the intimate relationship between the researcher and what is studied, and the situational context that shapes inquiry (Denzin &Lincoln 1997). They seek answers to questions that stress how social experience is created and given meaning. But, Qualitative research though well developed in terms of validity is underdeveloped in terms of reliability (Perakyla 1997, p 216). Reliability, according to Kirk and Miller (1986), is defined as the degree to which the findings are independent of accidental circumstances of the research; whereas validity of research concerns the interpretation of observations. In contrast quantitative studies emphasize the measurement and analysis of causal relationships between variables, not processes. Inquiry is purported to be within a value-free framework (Denzin and Lincoln 1997). Quantitative research methodologies depend largely on tests, rating scales, questionnaires, and physical measures. It could be therefore concluded that while numbers are the end product of quantitative researches, flow diagrams and narrative description of events and process are the end result of qualitative researches (Landy & Conte, 2004; Struss & Corbin, 1990). There are different types of qualitative methods, which are ethnography/ participant observation, qualitative interviewing, focus groups, discourse and

278

conversation analysis, collection and qualitative analysis of texts and documents (Bryman and Bell 2003). Interview participants communicate the practices of their everyday life in manners that are recognisable and orderly giving a scope to the development of a reality that is ongoing and achieved through an interpretive accomplishment (Holstein and Gubrium 1997).

Key words- research paradigms, ontology, epistemology, methodology, qualitative research, quantitative research, semi-structured interviews.

279

Manderscheid, K. (University of Lucerne, Switzerland) A Field Approach to Spatial Structures: international comparisons of their structuring force on social inequality Although for some decades already, the so called spatial turn and the mobilities paradigm have pointed out the importance of space and movements as dimensions of the social, empirical social research in stratification and inequality hardly meets these theoretical concerns by still not including them into their quantitative analyses.

In my paper, I will present findings from my quantitative international comparative survey based analyses on inequality and mobilities: Drawing on the field concept, I will suggest the usage of multiple correspondence analysis as a statistical technique to include social as well as spatial structures into social inequality research. Using quantitative data, correspondence analysis allows us to extract patterns of practices which can then be plotted against structural characteristics. Yet, in contrast to most multivariate methods, multiple correspondence analyses does not assume specific relationships between independent and dependent variables but maps patterns of categorical profiles found within the data and interprets their positioning by projecting additional variables into the two dimensional plots. Bourdieu's applied this method for his concept of social space, identifying the volume and composition of cultural and economic capital as structuring forces of cultural practices in France of the 1960s.

In my research, I transfer these ideas to mobility and transportation and their interrelation with social inequality. These practices are understood as emerging within and shaped by the specific context of social and spatial situations, that is: household constellations and social network embeddings, cultural values, welfare systems, labour and housing market geographies, available transportation infrastructures etc. These shaping structures represent materialised policies and their histories which have different impacts depending on one's position and which differ between countries and states. On this token, geographies and spaces however are not equalised with two-dimensional territories but as socially constituted relational systems. The comparison of countries and the tracing of developments over time allows to qualify the

280

structuring impacts of geographical and infrastructural factors.

After discussing these issues conceptually, I will present empirical findings on the interrelation between spatial and infrastructural, socio-structural dimensions and mobility practices based on Swiss (micro census mobility and transportation 2005 and 2010) and German data (Mobility in Germany MiD 2008). My contribution will conclude with a reflection on what kind of variables we need in surveys in order to be able to better analyse the complex interrelations between social and spatial structurings of the social.

Keywords: Spatial Analysis; Mobilities Paradigm; Multiple Correspondence Analysis

281

Manzoor, H., Mollona, E., and Rashid, M. (Kohat University of Science and Technology, Pakistan, Universita' degli Studi di Bologna, Italy, and University of Peshawar, Pakistan) Digging Deeper: Researcher’s emotions as a ‘methodological tool in understanding board dynamics In the recent years, there appears to be a growing interest among the researchers in appreciating the role of researcher’s emotions in qualitative research although we are often advised to maintain an appropriate cognitive distance with the research participants in the field. Often, due to this pressure, there is a possibility that valuable data can be lost that could informs us possibly more about the context and the phenomenon under investigation, particularly emotions. This paper is drawn from a doctoral study on emotions in the public sector board governance, specifically the British National Health Services (NHS) boards, and aims to highlight the methodological challenges involved in studying the phenomenon of emotions of board members through an ethnographic study of the so called ‘rational’ board meetings, despite of the dominant discourses of openness and research on improving governance in the public sector boards. The dominance and persistence of rationality in boards created additional challenges for the researcher in gaining access to the emotions of the board members, which also became the main cause of the denial of access to the NHS boards. Built on personal narratives of the first author, this paper sheds light on how the emotions of the researcher in the field became a methodological device in accessing, understanding, and interpreting emotions of the research participants. The fieldwork was not only emotionally intensive for the researcher but also induced powerful emotions among the board members as it threatened their illusion of rationality in boards. The stories from the field are illuminating and reveal the extent to which the management of emotions of the first author became difficult due to emergent situations in the field that threatened the (re)negotiation process of gaining and maintaining access to boards. This paper concludes with an emphasis that researcher’s emotional experiences in the field could be saved and can be incorporated as a strong methodological tool in studying emotions in different contexts.

282

Martini, N. (Jagiellonian University, Poland) Grasping the Multidimensionality of Relational Space. A Strategy for Data Collection The paper addresses several questions concerning spatial analysis. Firstly, what methodological innovations can be observed within sociology in this particular research field?

Secondly, which of these innovations are best suited for producing knowledge about the threefold process of thinking, creating and experiencing space (Nina Baur et al., 2014)?

Thirdly, what types of data are appropriate for analyzing these dimensions of social construction of space? Finally, how can these data be effectively collected? The paper does so by discussing theoretical and methodological implications of using mobile and reflexive research tools with the goal of grasping the multidimensionality of relational space. The paper argues that the multidimensional character of space can be successfully captured by the concept of social practice being understood as “a routinized way in which bodies are moved, objects are handled, subjects are treated, things are described and the world is understood” (Reckwitz, 2002, p. 249). The paper evaluates the use of the go-along (also known as walking interview) accompanied by the participatory photography and participatory mapping, as well as the autoethnographic analytical approach as a methodological strategy particularly suited for exposing and exploring interconnected complexes of bodily and mental activities which form social spatial practices by providing access to the social actors’ discursive and practical knowledge, as well as to the embodied experience through various types of data (as it allows to collect both sociological and geographical information). The paper examines how spatial analysts can benefit from becoming “sensory apprentices” while learning to practice “the space” with research participants during go-alongs, combining geodata (tracklog data, point- location data) with social science data (textual, visual and sensorial), as well as from transforming their own lived experience into qualitative data while examining the process of acquisition of informants’ knowledge and practical skills. Drawing on the author’s own research experience (conducted studies on spatial behavior of socially excluded people; public space perception and utilization among urban elderly; memory of space among the residents of rural area) the paper examines the innovative

283

potential of the discussed strategy for data collection. The paper concludes with a brief discussion on the challenges associated with its practical application and considers possible further improvements and developments.

Keywords: spatial analysis, data collection, go-alongs

284

Marttila, H. T. (Institute for Sociology Ludwig-Maximilians-Universität Munich) Post-foundational Discourse Analysis: New Research Program in Relational Sociology While other discourse analytical approaches flourish, the so called Essex School in Discourse Analysis (here: Post-foundational Discourse Analysis/PDA), dating from Laclau’s and Mouffe’s post-Marxist discourse theory, has remained something of a laggard with regard to its limited use in empirical research. It is nothing of a surprise that PDA has gathered little attention in empirical research. This is largely because PDA’s “methodological deficit” (Keller 2011: 164) has implicated that there is hardly any information available about how the PDA can be operationalized in empirical research. This paper provides one possible solution for how this methodological deficit and explicates the analytical strategies and scientific methods that facilitate PDA’s operationalization in empirical research. This paper evolves in three steps. The first step explicates the ontological propositions underlying the PDA and explains why any socially meaningful phenomenon must be conceived to result from meaning generating relations called discourse. This relational ontology of the social constitutes “hard core” of a research program, which on the level of empirical analysis “tells us what paths of research to avoid (negative heuristic) and…what paths to pursue (positive heuristic)” (Lakatos 1968: 167f.). The second step poses the question what actually constitutes the research object of the PDA. Following the lead of Laclau and Mouffe the social cannot be subdivided into realms of discursive and non-discursive phenomena because every socially meaningful object is constructed in a discourse. However, in contrast to Laclau’s and Mouffe’s ‘flat’ epistemology of discourse, different discourses vary partly with regard to their composition of linguistic and non-linguistic elements, and partly with regard to the extent to which social subjects’ are actually aware of these elements’ meaning-contents. As a result of discursive sedimentation socially meaningful phenomena become largely decoupled from the meanings that once upon a time gave rise to them, legitimated their existence and motivated sets of related social practices. In order to identify the discursive origins and conditions of possibility of a socially meaningful object, we

285

cannot suffice with the reconstruction of meanings that social subjects consciously identify with this object. In order to identify underneath the level of subjectively accessible meanings located sedimented discourses; I argue that not discourse qua a relational system of meaning but “discursive regime” – a reciprocal system of relations between discourse specific meanings, subject roles and institutions – constitutes PDA’s research object. The third step elucidates scientific methods that render empirical visibility to discursive regimes. Considering that discursive regimes are not only made of meanings and meaning processing and generating linguistic practices, but consist also of sedimented institutions, technologies of power, routinized subject roles and “model practices” (Keller): PDA cannot be constrained to a study of linguistic structures and practices. It is argued that the “multimodal” character of discursive regime necessarily implicates that PDA needs to be conducted in form of a “multimethod” approach because only a combination of scientific methods with their different phenomenal focusses allow us to reconstruct entire discursive regimes. Keywords: Post-foundational discourse analysis (PDA), research program, relational sociology

286

Maslovskaya, O., Durrant, G., and Smith, P. (National Centre for Research Methods (NCRM), Social Statistics and Demography, University of Southampton)

Improving the Efficiency in Interviewer Call Scheduling: Analysing Final Outcome and Length of Call Sequences in a Longitudinal Context For interviewer administered surveys many survey agencies nowadays routinely collect paradata in the form of call record data. Examples of such data are recordings of the day, time and outcome of each call or visit and, for face-to-face surveys, information collected by the interviewer such as physical and social characteristics of the selected housing unit and neighbourhood. Researchers have increasingly become interested in how best to use and analyse such information. It is hoped that a better understanding of the calling patterns and the mechanisms leading to particular call sequences will help to improve different aspects data collection.

More specifically, for statistical agencies, investing time and effort into repeated calls and follow- ups is resource-intensive. From a survey management perspective, it seems desirable to avoid long unsuccessful call sequences to improve efficiency. The aim then is to identify early on in the data collection process cases prone to long and unsuccessful call sequences.

This paper models call record data predicting final call outcome and length of a call sequence in a longitudinal context. Separate binary logistic and joint multinomial models for the two outcomes are considered, where the models account for the clustering of sample cases within interviewers. Of particular interest is to identify good explanatory variables that predict final outcome and length of a call sequence, in particular characterising long unsuccessful call sequences. Further research questions that we aim to address in this study are: how can call record predictors best be incorporated into the model(s), as summary statistics or as individual outcomes?; how predictive are the models?; does their ability to predict improve once interviewer observation variables, more call record data, and information from the previous wave are available?

The study uses data from the first two waves of a large-scale longitudinal survey in the UK, Understanding Society. Implications for survey practice and fieldwork procedures are discussed. Outcomes of previous calls, in particular from the most recent call, are highly predictive. The

287

timing of calls and interviewer observation variables from both waves, although significant in the models, only slightly improve the predictive power of models.

288

Matthes, B. and Unger, S. (Institute for Employment Research (IAB)) Determining recall errors in retrospective life course data – An approach using linked survey and administrative data Autobiographical recall is imperfect and memories differ in characteristic ways from reality. Therefore, retrospectively reported event histories are at risk of containing a specific memory related error component (recall error). Literature shows that episodes and transitions are underreported, and memory is selectively worse for shorter and atypical events as well as for events that date back a long time. Because of the imperfections in recall, there is often skepticism about the usefulness of retrospective data.

One way to analyze retrospective recall errors is using linked survey and administrative data of the same individuals. We can draw on a dataset that contains information on all German employees only excluding self-employed persons and civil servants and link this to our survey data However, the reliability of administrative data has been distrusted, too. It is well-known that its quality depends on the importance of the recorded information for administrative purposes. For instance, information on occupation in the IAB employment data originates from the notification process of the German social security system. The information is unverified (except for simple value checks), and misreporting has no consequences concerning obligations or claims out of the social security neither for the employer nor for the employee.

Until now, the recall error of retrospective life course data has not been analyzed while simultaneously considering errors in administrative data. In our presentation we determine the recall errors in retrospective life course data by using the linked data-set ALWA-ADIAB, which combines interview data and administrative data from the same individuals. Assuming that the recall error does not play a role up to three month prior to the interview date, we analyze the “notification error” – the error which can be ascribed to the administrative notification process and by taking this into account, we determine the recall error in the retrospective reports of

289

survey respondents.

Using the example of information on occupation we test two hypotheses: First, we assume that the recall of information on occupation is increasingly subject of error when the retrospective interval gets longer. Second, we assume that the quality of survey data is influenced by the stability of job histories. That is, frequent job changes and short times of employment episodes coincide with more recall errors.

Our analyses show that the extent of the “notification error” is large and therefore has to be considered when analyzing linked survey and administrative data. In gaining information on the likelihood of error in administrative data as a function of job and firm characteristics, the recall error in survey data can be modeled. This also sheds light on the size and determinants of the recall error when reporting autobiographical events and transitions retrospectively.

Both data sources, administrative data and retrospective life course data, have more or less pronounced advantages and disadvantages. The challenge in analyzing such linked data sets is to combine the most reliable information of both sources. Considering the already relatively extensive literature about survey errors this calls for expanding the further research on the reliability of administrative data. Keywords: recall error; linked data; occupation;

290

Matjašič, M. and Vehovar, V. (Faculty of Social Sciences, University of Ljubljana) The Efficiency of Nonparametric Methods in the Detection of Response Time Outliers in Web Surveys In the last two decades, researchers have intensively used computerized methods for collection of different types of paradata (e.g. keystrokes, mouse clicks, response time) to evaluate and improve survey instruments, but also to understand how respondents behave during the survey process. Analysis of the response time, in particular, is a promising approach to gain this insight.

Within this context, majority of studies analyzed response times (also called time latencies) respondents need to complete certain actions. The basic problem in the corresponding analysis is the identification of the units that are staying too long or too short on a certain point in the process (i.e. on a certain page of the questionnaire). Here we are also talking about response time outliers or response time anomalies. Namely, it is very likely that these units violate the expected cognitive behavior of the respondent, so they might also provide a lower quality of survey responses.

Proper analyzing of response time is thus very important and requires a robust methodology. In this presentation we review and compare most frequent approaches (i.e. standard deviation, Tukey's method, median, absolute deviation from the median, speeder index), as well as some less common approaches (e.g. cognitive method reading rate) for identification of response time outliers per page. Using simulations we studied performance and robustness of these methods across various distributions (equal, normal, lognormal) and their parameters (skewness, kurtosis), as well as across various sample sizes (100, 500, 1000). The findings were then verified with empirical data (the real data set provided by open source application 1KA). The results show that among the most robust methods we find absolute deviation from the median, speeder index and reading rate. However, this is not a uniform rule, because the performance of various methods often depends on specific contexts (i.e. response time distribution).

291

Mavletova, A., Tholmogorova, T. and Gavrilov, K. (Higher School of Economics; National Research University Higher School of Economics) Risk communication among children: The effect of a gamified design The basis of patient-centered medicine is providing information of the benefits and risks of drugs, screening procedures, and treatments to the patients. The information is often provided as data with numbers and statistical probabilities. Patients should have required knowledge to understand this information and interpret data to make informed decision. Most experimental studies in risk communication focused on adults and fewer studies focused on children. In this presentation we measure if using gamified elements can improve accuracy of understanding of numerical information about medical treatments and procedures among adolescents 11-15 years of age. The experiment was conducted at a Moscow school in Russia. All children at the 6-th, 7-th and 8- th grade participated in the experiment and completed a web survey in the computer class. Children were randomly assigned to one of the two conditions: a traditional web survey and a gamified web survey. The experiment was conducted in April-May 2016. In total, 213 respondents completed the survey: 109 respondents completed a traditional web survey and 104 respondents completed a gamified survey. The mean age is 13.8 years of age (SD=1.1). 58% of the respondents are girls and 42% are boys. In line with the expectations Gamification was helpful for respondents in the questions describing difficult medical problems and in difficult calculation tasks. However the effect depended on the risk literacy of the respondents. Keywords: risk communication, gamification, children

292

Mayerl, J. and Stocké, V. (University of Kaiserslautern , University of Kassel) Moderation Effects of Response Latencies and Meta-Judgments on Response-Effects and Attitude-Behavior Consistency The aim of the paper is to test moderation effects of cognitive accessibility of the respective target information in respondents’ memory concerning a) moderation of attitude-behavior consistency, and b) moderation of the occurrence of response effects. The paper compares two alternative indicators of cognitive accessibility: self-reported response certainty and the time needed to answer the question. A second issue addressed in this paper is how response certainties and response latencies should be transformed prior to data analysis, in order to maximize their predictive power. Statistical analyses of data of two surveys include multiple group comparisons and interaction effects in linear as well as logistic regression analyses.

Respondents' answers about their attitudes and behavior are found to be affected by different kinds of response effects. This means that they differ considerably depending on situational cues, context information, wording of questions, or format of response scales. Furthermore, self- reported attitudes are found to be very differently predictive for behavior. It has been hypothesized that both, the degree of attitude-behavior consistency as well as the susceptibility to response effects are determined by the cognitive accessibility of the respective target information in respondents’ memory.

The first aim of the present paper is to test these moderation hypotheses concerning predictive power of attitudes towards behavior and occurrence of response effects using two alternative, individual level indicators for the cognitive accessibility of information. These measures are meta- judgmental indicators, in our case the subjects’ self-reported response certainties, and the time needed to answer the question under consideration as an operative unobtrusive paradata measure.

A second issue addressed in this paper is how response certainties and response latencies should be transformed prior to data analysis in order to maximize their predictive power for attitude- behavior consistency and response-effects. Accordingly, the predictive power of untransformed

293

measures is compared with that of logarithmic, square-root and reciprocally transformed versions. In addition, several types of data treatment of response latencies are applied and evaluated (e.g. correction of baseline speed, outlier treatment, and interviewers’ time validation).

To test these moderation hypotheses and data treatment procedures we use data from two different computer assisted surveys (CATI and CAPI) with random probability samples in Germany. Whereas the first study is used to test for attitude-behavior consistency of donating money to social welfare, the second study is based on a field experiment testing for the effect of low- versus high-frequency response scales for collecting information about the frequency of television consumption. Statistical analyses include multiple group comparison and interaction effects in linear as well as logistic regression models.

294

Mc Namara, C. (Trinity College Dublin, The University of Dublin) The time-saving benefits of using Grounded Theory Methodology for observational studies in the social sciences This paper considers a grounded theory study of betting shops carried out in Dublin, Ireland which utilised covert participant observation as the primary method of data collection. Grounded theory methodology and the specific tool of participant observation were utilised because the underlying logic of the methodology and the specific research tool were considered most appropriate for a study that was underpinned by a theoretical orientation that fell under the symbolic interactionist tradition. However, the focus of this paper is an unanticipated outcome of using grounded theory methodology; that is, that the use of this methodology, seen by some as labour intensive, actually made possible the significant reduction of time that had to be spent observing in the field. Particular consideration is given to the principal features of this methodology – theoretical sampling and the constant comparison of data – that can enable a researcher to spend less time in the field without reducing the quality of the research.

295

Menold, N. and Bosnjak, M. (GESIS – Leibniz Institute for the Social Sciences) Summarizing 60 years of Research on the Number of Response Categories in Rating Scales: First Result of a Meta-Analysis This meta-analysis investigates the effects of the number of categories in rating scales on reliability estimates. This analysis was conducted to address seemingly mixed findings, which have been reported in the literature for decades. We included 23 manuscripts, published between the years 1953 and 2015, each reporting reliability coefficients for different number of categories. Overall, approx. 200 reliability estimates along with the respective information about the reliability estimation method, the kind of the sample, scale labeling and scale orientation and other characteristics of rating scale implementations could have been extracted. The first results show that there is a curvilinear relationship between the number of categories and reliability. The results show also that other characteristics, e.g., labeling of categories, scale orientation and topic of requests were relevant in moderating this effect. We will discuss the implications of these findings for developing survey items, and will sketch some avenues for future research.

296

Menold, N., Landrock, U., Pellner, N. and Winker, P. (GESIS) The Impact of Payment and Participation on Interviewers’ Accuracy in Face-to-face Surveys: Investigations from a Field Experiment 1. In face-to-face surveys the interviewer is a key actor, who may affect the quality of survey data. There are some studies, which address interviewers’ impact, for example, by evaluating interviewer variance in secondary analyses. However, there is a lack of knowledge how the circumstances and working conditions influence the interviewers’ work. We implemented an experimental study, which investigates such influences. A motivation theory was used to generate hypotheses about the possible effects of interviewers’ payment, instructions and task difficulty on the accuracy of the interviewers’ field work. The number of deviations of an interviewer regarding the documentation of contacts and their outcomes (paradata), mistakes and other deviations from instructions were used as dependent variable. Two kinds of payments were compared: payment per interview and payment per time. In addition, an instruction was varied, which legitimated deviations in interviewers’ behavior. Finally, to be able to detect interviewers’ deviations, the potential respondents were instructed to behave in a certain manner: to cooperate, to refuse participation and to break off the interview. The results show that there were lower deviations if the interviewers were payed per time, while the variation in the instruction did not have an impact. In addition, there were more deviations by interviewers in the case of break offs, which is associated with a high task difficulty. The results are discussed against the background of interviewers’ motivation theory. Suggestions for the organization of field work are provided.

297

Mergener, A. (Federal Institute for Vocational Education and Training (BIBB), Germany) Testing Validity of hypothetical Recruitment-Decisions using a Factorial Survey In survey methodology, experimental designs, like vignette-studies or also called factorial surveys, have gained a broad attention in the social science. The advantages are obvious: Due to controlled settings, experimental designs allow causal interpretations and a higher internal validity. Also factorial surveys can be easily conducted as an add-on of established surveys and the mode of data collection can be internet-based, what minimizes costs. Furthermore, because of the multidimensionality of the vignettes, factorial surveys are considered to be less prone to provoke socially desirable responses in cases of sensitive issues (like discrimination). However, there are also critical voices. In vignette studies, respondents react on hypothetical descriptions of given scenarios (e.g. situations or persons). Therefore, it is arguable, whether this experimental design is valid in view of plausibility and realistic behaviour. This paper compares responses to single-item questions and vignette-judgements concerning recruitment-strategies of immigrants on the German Labour Market. Therefore, I use a merged dataset of a telephone survey (first stage) and a follow-up online survey including a factorial survey (second stage) where more than 5000 companies (Human Resource Managers) in Germany are included. Topics of the first and second stage survey have concentrated on recruitment strategies with a special focus on migrants and possible skill shortages/hiring problems in the companies. The analysis focuses both determinants that influence real employments of foreign skilled workers and those who influence hypothetical recruitment decisions. Because of the two stage process asking the same companies, it is possible to compare their responses and to gain on this way information about the validity of the answer within. Keywords: Factorial Survey, Validity, Recruitment process

298

Metzler, A., and Fuchs, M. (Darmstadt University of Technology) Using response time to predict survey break-off in Web surveys A key characteristic of self-administered surveys in general is that respondents obtain no additional support by an interviewer. While the absence of an interviewer has advantages when it comes to sensitive topics since social desirability is assumed to be lower when no interviewer is present, the lack of an interviewer motivating respondents to complete the questionnaire and to optimize during the question-answer process is seen as a serious disadvantage. Hence, careless responding and survey break-off are elevated in Web surveys. In recent years, interactive features have been incorporated in Web surveys to compensate for the absence of an interviewer. For example, messages concerning item missing, non-differentiation or speeding have been implemented successfully. So far however, this has not been achieved for survey break-off, which is dramatically higher in Web surveys as compared to other survey modes. In order to provide instant motivation and feedback to respondents of Web surveys, who are about to abandon the Web survey, it is essential to identify these respondents. In this paper, we assess changes in the speed of responding as a potential indicator of subsequent survey break-off, assuming that survey break-off can be identified right before respondents abandon the Web survey by analyzing changes in response times. Web surveys allow the collection of various paradata. Client-side paradata such as mouse clicks and keystrokes as well as server-side paradata such as time and user agent strings can be easily recorded. In the analyses reported in this paper, we used three large scale Web surveys among university applicants conducted in 2013 (n=7,395), 2014 (n=5,996) and 2015 (n=4,034) providing time stamps for each questionnaire page as a server-side paradata. We assessed changes in response times for respondents who completed the questionnaire, for respondents who abandoned the Web survey within the following three pages respectively to the page of interest and for respondents who abandoned the Web survey later in the questionnaire. In order to assess changes in the speed of responding we standardized response times and determined the relative speed for each respondent. Thus, we were able to detect changes in response times for each respondent across questions. Descriptive analyses indicated that survey

299

break-offs significantly speed up right before they abandon the Web survey. However, results also suggest that the false negatives (people who speed up but then stay in the survey) are still considerably high. Furthermore, logistic regression analyses were conducted to examine the effect of changes in response time on the likelihood of survey break-off. Initial findings showed that changes in response time have a significant effect on survey break-off. Survey break-off was more likely, if the relative speed of respondents increased. Further analyses of changes in response times might allow a more reliable prediction of respondents who are about to abandon Web surveys, thus, instant feedback can be integrated in Web surveys at the right moment motivating respondents to continue with the survey. Keywords: paradata, speeding, nonresponse

300

Meulemann, H. (Institut für Soziologie und Sozialpsychologie, University Cologne, Germany) Makes religion happy – or makes happiness religious? Religiosity and life satisfaction between early and late mid-life As religion gives meaning to misfortune and injustice, religiosity should increase life satisfaction – nomization hypothesis. As life satisfaction engenders a positive world-view, it should increase religiosity – optimism hypothesis. Furthermore, religiosity should depend positively on the inculcation of the religion of origin and negatively on the reflection about religious questions during youth – inculcation and reflection hypotheses. Finally, life satisfaction should depend positively on life success – success hypothesis. These hypotheses are examined in a panel survey of 1301 former German high school students first interviewed at age 16 and re-interviewed at age 30, 43, and 56. In such a panel, they constitute a causal system of two endogeneous variables (life satisfaction and religiosity) in three re-interview and four exogenous variables (inculcation and reflection at age 16, and life success in each re-interview). As the nomization and optimism hypothesis apply for age 43 and 56 and the success hypothesis applies for age 30, 43, and 56, hypothesis about their stability, their cross-over effects, and the changes of the impacts upon them are derived as well. The hypotheses are examined using structural equation models. Missing data are treated by FIML estimation. Unobserved heterogeneity is controlled for by a person factor in the equation such that fixed effects are examined. In contrast to fixed effects regression, a structural equation model has three advantages in such a model: (1) It allows the estimation of stabilities and cross-over effects between the two endogeneous variables; (2) it allows the examination of the equality of error variances over time points; (3) it allows tests whether specific exogeneous variables do not correlate with the person factor. – The nomization and optimism hypotheses are not confirmed; rather, religiosity and life satisfaction affect each other negatively. The inculcation and reflection hypotheses are confirmed partially, the success hypothesis throughout. The negative impact of religiosity on life satisfaction is explained by the long time lags between the panel waves during which the nomic utility of religion decreases and its costs increase.

Keywords: Life course, long term panel study, unobserved heterogeneity

301

Meyer, D. (Friedrich Schiller University of Jena) Social Network Analysis Meets Bibliometrics: Toward a Visualization of Academic Power in Economics, Political Science, and Sociology Due to the rapid growth of science, its transnational diffusion, and interconnectedness of its various fields, it is nearly impossible to track the development of academic disciplines, let alone study social connections among researchers or the distribution of academic power in today’s scientific community. In response to this problem, I seek to direct attention to the great potential of big data and modern bibliometrics for both historians and sociologists of science. Despite all criticisms, bibliometric data provided by Web of Science, Google Scholar, and other services continue to provide the best means of grasping the complexity and interconnectedness of contemporary science. By using the examples from economics, political science, and sociology, I here demonstrate how social network analysis and bibliometrics can be further integrated to gain a more profound understanding of recent developments in the social sciences. By drawing upon more than 50,000 articles from 30 top journals as indexed in Web of Science, I highlight major thematic and paradigmatic trends in the development of the three disciplines during recent decades. Using Bourdieu’s field-oriented approach, I examine and compare the concentration of departmental and personal power in the three disciplines, as well as their overall structure. Methodically, I apply citation-based similarity measures, and by using modern software, construct a network- based visualization that, similar to a video, shows how the fields have emerged over the years. In the paper, I also address methodological considerations such as Bourdieu’s notion of the relational and how it can be re-conceptualized in the age of (bibliometric) big data, as well as underscore drawbacks in combining social network analysis and bibliometrics, including dependencies on commercial data providers and the issue of complexity. Keywords: academic power; bibliometrics; social network analysis

302

Mika, T. (Deutsche Rentenversicherung Bund (German Pension Insurance, Research Data Centre)) The first twenty years of employment trajectories in Germany: A comparison of the birth cohorts 1935 to 1974 Cohort studies aim at the ongoing process of social change. The methodological challenge is the measurement of older and younger birth cohort at the same age and with the same analytical and empirical concepts. The proposal presents an example of a cohort study which analyses the extent of an increasing divide in the labour market for those born after 1955 compared to those born between 1935 and 1954. The time between the end of formal education and the successful entry into the labour market became more difficult for some members of younger birth cohorts. Periods of employment are more often interrupted and unemployment with and without income replacement and sickness pay fill some of those gaps. Personal income growth is also slower to come by and less steady for younger cohorts. The paper compares the changing pattern of the first twenty years of the employment careers for the birth cohorts 1935-1944, 1945-1954, 1955-1964 and 1965-1974. For this goal, a cluster analysis of all cohorts together is carried out with longitudinal data from the statutory pension insurance (‘Sample of the insured populations’ records’ (VSKT)) of the years 2002, 2005, 2007, 2009, 2011, 2012, 2014. The data make the life courses of those age cohorts at the same biographical period from age 15 to age 35 comparable, including the level of the gross income before taxes. However, a major challenge is the changing institutional framework, which requires a transformation of the biographical records in order to make them comparable. Key Words: Process-Oriented Methodology – Longitudinal Analysis – Cohort Analysis

303

Mikucka, M. and Valentova, M. (Université catholique de Louvain (Belgium); Laboratory for Comparative Social Research, HSE (Russia); LISER (Luxembourg)) Employed or inactive? Cross-national differences in coding parental leave beneficiaries in European Labour Force Survey data. In survey research the parental leave beneficiaries are usually coded as either employed or inactive. An exception is the European Labor Force Survey (EU-LFS), which includes parental leave among other forms of being employed but temporarily not working. This paper explores classification of parental leave takers in EU-LFS.We show that classification rules differ crossnationally: in some countries parental leave takers are considered inactive, in others – employed but temporarily not working. In particular in the Czech Republic, Estonia, Hungary and Slovakia the EU-LFS data classify the beneficiaries as inactive. We estimate the number of mothers on parental leave in these countries and show that EU-LFS employment rates of women aged 18-40 are biased downwards 2-7 percentage points; for mothers of children aged 0-2 the bias reaches 12-45 percentage points. Our study shows the limited comparability of EU-LFS employment rates and warns about possible bias in cross-national studies. Keywords: employment status; maternity leave; parental leave

304

Mohadjer, L. (Westat) Minimizing Total Survey Error in a survey of out-of-school children in developing countries The Programme for International Student Assessment (PISA) is a triennial international survey sponsored by the Organisation for Economic Co-operation and Development (OECD). The survey evaluates education systems worldwide by testing the skills and knowledge of 15-year-old students through a background questionnaire and a direct assessment, administered in classroom settings. To date, students representing more than 70 economies have participated in the assessment. To make PISA more relevant to developing countries, OECD has started an initiative, called PISA for Development (PISA-D) project, which currently includes seven countries. PISA-D includes two components; assessment of 15-year-old student in schools, and those that are not in school (referred to as out-of-school 15-year-olds). This paper is focused on minimizing total survey error in sampling out-of-school 15-year-olds in PISA-D. This project will help investigate various methodologies to identify out-of-school youth, and develop a better understanding of the issues that will need to be dealt with in implementing a full scaled nationally representative assessment of out-of-school 15 year-olds. The main objective of the nationally representative samples are to measure the extent and the reasons for 15-year-olds being out-of- school, their level of competencies, and to evaluate the effectiveness of various policies on the status of universal basic education in the countries.

More specifically, the goal is to develop a methodology and approach for incorporating out-of- school 15 year-olds in the PISA assessment through a household survey. Similar to other international comparative studies, the goal is to make inferences and comparisons across national populations on the basis of survey samples. To achieve this goal, PISA-D for out-of- school 15-year-olds will need to establish an overall set of Quality Assurance (QA) and Quality Control (QC) procedures covering all aspects of the study. Participating countries will be instructed to follow the standards to ensure the sources of survey variability are kept to a minimum and that the survey design and implementation processes produce high-quality and internationally comparable data. If successful, this could make it possible in future rounds of PISA for a measure of the competencies of out-of-school 15 year-olds to be introduced as a

305

complement to the assessment of 15 year-olds in school. This presentation will review the challenges involved in sampling a rare group (such as 15-year- olds that are out-of-school) through a household survey in developing countries, and provides an update on developing QA and QC procedures for such surveys.

306

Mohr, A.H., Lindsay, T. and Sell, A. (University of Minnesota) The Influence of Visual Design on Creative Divergent Thinking in an Online Survey Previous research has shown that the amount of text a participant enters in response to a survey question can be influenced by the size or number of text fields provided. However, there is less research on whether these effects extend to assessment-based questions, which attempt to assess not only what participants think, but also how they think. Tasks that use these questions typically measure aspects of the responses such as elaboration or detail, in addition to the semantic content, to assess some trait or ability. Since elaboration and detail are measured at least in part by counting words and themes present in responses, it is important to understand whether and how these questions are vulnerable to the effects of visual design. This presentation will discuss our recent work (Hofelich Mohr, Sell, & Lindsay, 2015) examining one type of assessment-based task, Guilford’s Alternative Uses Task, which measures an aspect of creativity called divergent thinking. In this task, participants are given an object and are asked to list as many uses for that object as they can in a set amount of time. Using a 2x3 design, we experimentally manipulated the visual display of the text boxes they used to enter responses: the type of text box (whether they appeared in list-like or essay-box formats) and the number of lines initially visible (5, 10, or 15 rows). We found that several measured aspects of creative ability were influenced by the visual display manipulations, including fluency, elaboration, and originality. These findings extend previous survey methodology and opinion measurement research, demonstrating that assessment-based tasks are vulnerable to visual design. Our results also attest to the importance of transparency in research methods; these visual design choices are often not reported in the literature, despite having an effect even on these assessment questionnaires. With growing concerns around the reproducibility of social science research results, it is critical to understand the level of detail required for reporting research methodology in order to produce consistent results. As our results demonstrate, visual design choices may have significant ramifications for online research administration. Reference:

307

Hofelich Mohr, A., Sell, A., & Lindsay, T. (2016). Thinking Inside the Box: Visual Design of the Response Box Affects Creative Divergent Thinking in an Online Survey. Social Science Computer Review, 34(3) 347-359. doi: 10.1177/0894439315588736 Keywords: visual design, online survey, creativity assessment

Mondo, E. (Université libre de Bruxelles, Belgium) The Role of Religion in the European Parliament: An Analysis of MEPs’ Vote on Abortion and Stem Cell Research (1998-2013) This paper studies how the members of the European Parliament (MEPs) vote on morality issues, i.e. issues that are framed in terms of (religious) values. The objective is to analyze the extent to which morality politics challenges the literature on party discipline and party loyalty. Internal cohesion and party leadership may indeed weaken when politicians are confronted to religion and values. The question is becoming more and more relevant in the light of the growing politicization of morality issues at the European Union (EU) level. For example, in recent years, MEPs have been debating the bioethical issues of abortion and human embryonic stem cell research (hESCR). The former is tackled through parliamentary reports on sexual and reproductive health and rights (SRHR); the latter is dealt with through the successive European research framework programs. The question arises: “How does religion influence MEPs’ vote on abortion and hESCR?” The paper analyzes four pieces of legislation: two parliamentary reports on SRHR (Van Lancker report– 2002 – and Estrela report – 2013), as well as two research framework programs (FP5 – 1998 – and FP7 – 2007). These study cases enable us to control both the topic and the time variables. The methodological approach relies on a qualitative discourse analysis of semi-structured interviews conducted with twenty MEPs. The results show that even though party discipline remains an important factor influencing MEPs’ vote, religion appears to be a driving force when it comes to morality politics. In other words, secularizing trends which dominate the European landscape have not diminished the political salience of individuals’ religious convictions.

308

Mueller, G. (University of Fribourg, Switzerland) Charting the Pathways Through the Space of Welfare Regimes: On the Use of QCA for the Construction of Topological Social Maps Topological maps like e.g. the network-plan of the metro of Paris are abstractions of the properties of the "real" Euclidean space, which surrounds us in our everyday life. Topological maps mainly focus on the non-/existence of direct connections between the points of the map and less on their quantitative distances. In spite of these deficits, topological maps are useful for planning journeys through the physical space, as demonstrated by the example of the plan of the metro lines. Moreover, they have the advantage that their construction is less demanding with regard to the available data: qualitative information is generally sufficient. This asset of topological maps is especially interesting for social research, where measurements are often only nominal or at best ordinal.

This paper focuses on the construction of topological social maps, which inform their users about possible transitions from a given state of a social system to another. It proposes to use for this purpose the modal logic of J. Lukasiewicz (1920 / 1970)1 and Qualitative Comparative Analysis (crisp-set QCA) of Ch. Ragin (1987)2. Modal logic is required for describing the possibilities and necessities of transitions, whereas QCA is used for determining the logical configurations of the conditioning variables, which enable or disable the mentioned possibilities of transitions. The result of the analysis has a pictorial representation borrowed from mathematical graph theory.

As an illustrative example, the author uses the mentioned method in order to explore the pathways of nations through the social space (P. Bourdieu 1989)3 of welfare regimes, which has according to G. Esping-Andersen (1990)4 three basic dimensions: the degree of socialism, liberalism, and conservatism of a social security system. The empirical analysis is based on a publication of L. Scruggs and J. Allan (2008)5, which presents additional data about the observed transitions of countries between 1980 and 2005 through this abstract social space. We attempt to explain these changes by the types of the considered welfare regimes, the public need for social security, and the deficit in the national household budget, which all together constitute

309

the logical configurations of the conditioning variables. The result is like the topological map of the metro of Paris: it informs about the next station in the history of a national welfare state and the possibility to change the direction of the collective social welfare.

Reference: 1 J. Lukasiewicz (1920 / 1970): Selected Works. Amsterdam, North-Holland. 2 Ch. Ragin (1987): The Comparative Method. Berkeley, Univ. of California Press. 3 P. Bourdieu (1989): Die feinen Unterschiede (La distinction). Frankfurt, Suhrkamp. 4 G. Esping-Andersen (1990): The Three Worlds of Welfare Capitalism. Cambridge, Polity Press. 5 L. Scruggs and J. Allan (2008): Social Stratification and Welfare Regimes for the 21st Century. In: World Politics, vol. 60, pp. 642-664.

Keywords: Spatial analysis, topological social maps, welfare regimes

310

Müller, J., and Garrido, N. (Institutional Affiliations: Universitat Oberta de Catalunya (UOC)) Sensor Based Data for Social Sciences Research. Using Sociometric Data for Microlevel Insights This contribution presents first insights from using sociometric badges for studying group level interaction processes in research teams. Sociometric badges are smart-phone sized, sensor packed devices that supply fine grained data for analyzing patterns of (face-to-face) social interaction, non-verbal communication (“Honest Signals” see Pentland 2008), and turn-taking. They have been used in studying forms of collaboration (Kim, McFee, Olguin, Waber, & Pentland, 2012)⁠, the flow of knowledge in organizations (Fischbach et al., 2010)⁠, social interaction patterns (Onnela, Waber, Pentland, Schnorf, & Lazer, 2014)⁠ or measuring social capital, creativity and trust in teams (Gloor et al., 2012)⁠. The type of data gathered cuts across established research methodologies and data paradigms: it provides more fine grained insights on non-verbal communication in real-life settings for example than traditional observational methods can gather; simultaneously, the amount and wealth of data produced in off-line networks enables a new precision in the analysis of network formation and evolution, difficult to achieve with self-report studies. Sociometric, sensor based data thus cuts across typical quantitative and qualitative paradigms. The current research is embedded in a H2020 project (GEDII, 2015-2018) on the impact of gender diversity on research performance. The project specifically aims to develop a methodology for using sociometric badges for capturing the effects of gender-role status beliefs in research teams on team collaboration. Although we use sociometric badges, a specific technology available from sociometricsolutions.com, our reflections regarding this technology are broader given the pervasiveness of powerful sensors in today's smartphones and other electronic devices. What is needed is a general discussion regarding the possibilities and limitations of high resolution data – neither “big”, nor “small” - which capture almost every parameter of our (social) lives. The presentation will focus on elaborating the specificity of sociometric data (e.g. available data dimensions) and how it dovetails existing social science research methodologies. Furthermore,

311

practical concerns will be exposed such as concerns of research ethics and data privacy, the quality of the data and possible approaches for analysis. The presentation will close with a discussion on the advantages and pitfalls of using sociometric data for progressive social change, i.e. uncovering and confronting (gender) bias in work environments. Keywords: sociometrics, sensors, data analysis, social network analysis, gender bias

312

Müller, S. (GESIS – Leibniz Institute for the Social Sciences, Germany) Studying the influence of the direct living environment: examples with the georeferenced German General Social Survey 2014 Individual behavior and attitudes are influenced by environmental factors such as exposure to environmental strain, the socio-demographic composition of the neighborhood and the accessibility of infrastructures. Therefore, not only in recent years, an increasing amount of studies explicitly explored the relationship between e.g. environmental burdens and socio- economic status. In turn these studies are the result of a great demand for spatial data in the social sciences. Our current project “Georeferencing of Survey Data” at the GESIS Data Archive for the Social Sciences implements a spatial data infrastructure for the Social Sciences in Germany and is, among others, concerned with the analysis potential of spatial data. While the project especially aims at linking social science survey data with small-scale spatial data on respondent’s coordinates level linkage in regard to E.g. neighborhood characteristics is considered as well. In a pilot we used two main sources of spatial data attributes that have been linked to survey data: environmental noise and German Census data. The environmental noise data that were collected in the context of the EU directive 2002/49/EG are in great part available through the Central Data Repository of the European Environment Information and Observation Network (EIONET) (http://cdr.eionet.europa.eu/) and are our example for spatial data on the respondent’s level. They consist of measurements for road, rail as well as air traffic noise. These data were added to the German General Social Survey 2014 (GGSS), so that every respondent of the GGSS was assigned to an environmental noise value which was measured at the façade of their dwelling. As the GGSS comprises questions concerning the subjective perception of environmental noise, differences between actual and subjective noise can now be examined. Furthermore, it can be explored which role subjective noise perception play when mental and physical health impacts of actual environmental noise are investigated.

313

The freely available German Census 2011 data consist of a variety of demographic measurements and are our spatial data on the neighborhood’s level (1km² raster grids) (https://www.zensus2011.de/). By adding these data to the GGSS 2014, we provided information about the neighborhoods respondents are living in. Besides the simple extraction of grid information from where the respondent is located, we also retrieved information from surrounding 1km² grids by conducting focal analyses. For example, these data can now be analyzed in respect to variation in demographic factors between different neighborhoods and their effect on attitudes towards migration. In our talk we will at first present the georeferenced GGSS data that were prepared in our spatial data infrastructure, which includes services for geocoding survey data as well as combining these with spatial data. The main part will consist of research examples we were able to explore by implementing this infrastructure. These include analyses in the area of health research as well as the investigation of attitudes influenced from neighborhood compositions. We will finish with a debate about the results implications in regard to the advantages and possibilities of georeferenced survey data. Keywords: Spatial Analysis, Georeferencing, German General Social Survey

314

Müller, S., Schweers, S., Kinder-Kurlanda, K., Siegers, P. and Zenk- Möltgen, W. (GESIS – Leibniz Institute for the Social Sciences) Studying the influence of the direct living environment: Examples with the georeferenced German General Social Survey 2014

Individual behavior and attitudes are influenced by environmental factors such as exposure to environmental strain, the socio-demographic composition of the neighborhood and the accessibility of infrastructures. Therefore, not only in recent years, an increasing amount of studies explicitly explored the relationship between e.g. environmental burdens and socio- economic status. In turn these studies are the result of a great demand for spatial data in the social sciences. Our current project “Georeferencing of Survey Data” at the GESIS Data Archive for the Social Sciences implements a spatial data infrastructure for the Social Sciences in Germany and is, among others, concerned with the analysis potential of spatial data. While the project especially aims at linking social science survey data with small-scale spatial data on respondent’s coordinates level linkage in regard to e.g. neighborhood characteristics is considered as well.

In a pilot we used two main sources of spatial data attributes that have been linked to survey data: environmental noise and German Census data. The environmental noise data that were collected in the context of the EU directive 2002/49/EG are in great part available through the Central Data Repository of the European Environment Information and Observation Network (EIONET) (http://cdr.eionet.europa.eu/) and are our example for spatial data on the respondent’s level. They consist of measurements for road, rail as well as air traffic noise. These data were added to the German General Social Survey 2014 (GGSS), so that every respondent of the GGSS was assigned to an environmental noise value which was measured at the façade of their dwelling. As the GGSS comprises questions concerning the subjective perception of environmental noise, differences between actual and subjective noise can now be examined. Furthermore, it can be explored which role subjective noise perception play when mental and physical health impacts of actual environmental noise are investigated.

315

Munro, C. and Couch, P. (Heatlth e-Research Centre, University of Manchester) Extending DDI to support the harmonisation of cohort variables: a new metadata registry In order to gain sufficient statistical power, the analysis of cohort phenotypic data increasingly requires the pooling of data across multiple cohorts. The cohorts are often recruited from different populations and the data are collected and managed in different ways. Before analysis, the data need to be harmonised, a process that involves mapping similar variables across the cohorts onto ‘standard variables’. We are developing an approach and supporting software that allows the creation of reporting rules for these standard variables. These rules, known as Variable Report Models, specify the information (properties) that must be available when a primary measurement or outcome is made available to an analyst. For example, this additional information may include qualifying factors that provide the context of the data needed for the analyst to correctly interpret the results. We adopt the DDI concept of a scheme, and support the sharing of Variable Report Models through the exchange of Variable Report Model schemes. Within the metadata registry we have applied OWL (the web ontology language - a formal way to describe taxonomies and classification network). The categories, properties and their relationship to each other, which form the (Variable Report Models) are modelled as OWL constructs. We have created a new metadata registry that can be used to manage Variable Report Models, along with additional metadata required for the harmonisation process. DDI 3.2 is used to model cohort data and the registry provides support for DDI groups, study units, datasets, questions, variables, control constructs, code schemes and response domains. The HeRC e-Lab is a secure online research environment allowing scientists to collaborate with data in a social network style, audited environment. The e-Lab is unlike many other data management systems as it is designed to operate on any data. Data are stored in a domain independent way such that the software can be easily extended and applied in different contexts. E-Lab modules have been developed to integrate data sets gathered across cohort studies. Data sets can be imported into the e-Lab and associated with data descriptions which are developed by the data managers of the cohorts and provide detailed information about the data. This

316

includes information about how data are represented, the semantics and additional context required for correct interpretation. This metadata supports the integration of the data, allowing research questions to be asked across multiple data sets, created by different communities. The data descriptions make reference to metadata managed by the registry and the Variable Report Models are used by the e-Lab to ensure that sufficient information about the data are provided by the data managers when data are imported into the e-Lab. Here we will report on our experiences applying Variable Report Models for the integration of data during the STELAR project. The models cover clinical measurements such as skin prick tests in addition to questionnaire data. Keywords: Metadata; Repository; Analysis

317

Muraleedharan, S. (Department of Politics and international Studies, SOAS, University of London) Visualizing Democracy: Mapping ‘Dalit’ subjectivity in India This paper aims to focus on the interconnected triad between identity, experience and social relations by bringing in the ‘visible’, ‘performative’ and the ‘ocular’ in the context of analyzing democratic politics in India. By using interdisciplinary methodological approaches like ethnographic interviews, participant observation and most crucially visual ethnography through photographs, I would like to propose that experiential research methods like using photographs from the site / of the subjects can elicit knowledge and the self-experiences of the respondents, in this case caste-based assertion in a more holistic way than the other traditional methods used to study politics. Politics needs to be analyzed as a ‘visual field’ and so does the discipline need to engage with the ‘visual’. This paper will argue to give the ‘visual’ a critical role, by introducing it as an alternative way of understanding and a route to knowledge about social and political phenomena. Here particularly I am referring to how photographs can be used as an effective tool for accessing the unarticulated and embodied views of individuals and groups in the research process. They contribute a lot in understanding about how people – respondents – see and interpret reality. Locating visual ethnography within the qualitative research field would then mean considering a ‘photograph’ becoming a field text in itself. Interdisciplinary approaches contribute thus in bringing the rational and the affective of the self and their worldviews. More importantly it enhances in understanding the inter-subjectivities of the respondents. Finally, it would try to raise issues with mainstream analysis of democratic theorization largely based on ‘speech’ or ‘debates’ than on the ‘vision’ or the ‘ocular’. This paper would hence make use of a few selected photographs from the fieldwork sites to try and explore as to how Ambedkar’s iconography is reflective of performance of defiance. Visual ethnography hence not only adds to reflexivity but also powerfully contributes in depicting temporalities, moments of social interactions, worldviews and ocular empowerment. Keywords: Photographs, Dalit identity, ocular democracy

318

Murphy, P. A Re-appraisal of Elias’s ‘Scenes from the Life of a Knight’: A case study of the influence of heteronomous preoccupations on methodology. Norbert Elias, with the subsequent explicit or implicit support of a number of figurational sociologists, placed considerable emphasis on the contribution that his case study ‘Scenes from the Life of a Knight’ made towards underpinning his explanation of the transition from the figuration in which the traditional knightly class held sway to the one dominated by court societies. This movement is a central dimension of his broader civilizing thesis. His case study is almost entirely dependent on his interpretation of a selection of drawings taken from The Medieval Housebook. The scope of the monograph (eBook) is extensive. However, today I can only offer a flavour of its contents. I will limit myself to an appraisal of Elias’s case study with the aim of highlighting the influence of heteronymous preoccupations on his methodology.

319

Musik, C. (St. Pölten University of Applied Sciences) An Ethnography of Trade Fairs as Field-Configuring Events The main purpose of trade fairs is to bring sellers and buyers from different parts of the world together. It is the place to socialize, to communicate and to contract. Simultaneously trade fairs also serve a crucial function in forming and spreading the industry’s cultural, i.e. cognitive basis: What is the specific business about, how can we tell good products from poor and which business practices are most appropriate? As such trade fairs „attract all major players in an industry in a single, generally confined place and over a limited period of time.“ (Moeran 2011: 8). The consequence of this high accumulation of relevant actors at one specific place and point of time is that „…the single most important ethnographic opportunity for participating in and observing a field in action is the industry trade fair.“ (ibid.) Because trade fairs are clearly limited in time they are exceedingly applicable to short-term ethnographic field work. Nevertheless, in cases where the focus of research is on the organisation of trade fairs, long-term field work might be the more appropriate approach as planning, scheduling, preparations, and postprocessing of trade fairs may take place throughout the whole year. Short-term field work is particularly relevant if the focus of research is on trade fairs as field-configuring events (Meyer et al. 2005; Lampel & Meyer 2008). In this sense trade fairs act as temporary and spatially bounded social organisations that relate to a more permanent field context. Specific fields and their values are both mirrored and reconstituted when trade fairs and markets are taking place. The paper presents the ongoing research project „Trading Cultures. An ethnography of international trade fairs for television programs, music and books“. The project aims at an ethnographic account of trading cultures in its twofold meaning: the trade in cultural goods and the cultures of trading. Picking up insights from new institutionalism and field theory it is assumed that selling and buying decisions by cultural intermediaries rest on shared beliefs, routines and rules. That is to say, a taken-for-granted trading culture which diffuses throughout global organisational fields. Especially in the cultural industries where future success is hardly to predict and the costs of failure are high, people tend to imitate highly legitimate models of

320

dealing with and in cultural goods. Trade fairs are places where we can observe this process of imitation and negotiation at work. The project is especially but not only grounded on short-term field work participant observation. We thoroughly explore and compare annual trade fairs and markets in three different cultural industries: The TV & Film industry is included by participating in the TV content market MIPCOM in Cannes and in the European Film Market (EFM) in Berlin. Book industry is respresented by the Frankfurt Book Fair and the Bologna Children's Book Fair. And finally, to get a read of the music industry we visit Midem, the international business-to-business music market, in Cannes and also attend the music platform Eurosonic Noorderslag, taking place in Groningen (NL) each January. The paper also deals with emerging methodological and practical problems. This most notably refers to the question of how to access business-to-business only trade fairs and markets. Due to research interests and strategic consideration we decided to focus on the buyer’s perspective. Because buyers are those actors with most influence on the selection and marketing of cultural goods. That means, we especially wanted to experience how acquisition executives of domestic media companies (inter)act at international trade fairs and markets. As we currently are in the middle of short-term field visits the plan was and still is to follow these actors (Latour 2005) on all their ways throughout the fairs and markets. The first field work experience at the TV trade show MIPCOM in Cannes showed that it was only possible to follow in very constrained ways: On the one hand, as it was previously arranged with our acquisition executive in several interviews, it was undesired to attend face-to-face business meetings, on the other hand many events were „by invitation only“. Nevertheless, the connection to the acquisition executive enabled us to attend major events, to meet important players in the field, and to collect very rich data in only a couple of days. Through subsequent interviews with other acquisition executives and the analysis of the trade show blog, website and social media appearance it was possible to connect the temporal bounded event to the wider context of the field.

321

Naldini, M., Santero, A. and Torrioni, P. M. (University of Turin- Department of Cultures, Politics and Society, Italy) Gender Dynamics within Couples using nested longitudinal qualitative data The arrival of the first child changes the equilibrium within the couple. The need to balance care and work responsibilities often requires a re-negotiation of gender roles. In particular, the transition to parenthood is a turning point in which the previous division of responsibilities within the couple undergoes a profound re-definition. But when does this negotiation process start? What are the mechanisms behind the construction of parenthood which result in couples' adaptation and solution? How can the researcher identify the main processes and mechanisms by analysing longitudinal interviews with couples? The negotiation processes and gender dynamics around transition to parenthood within highly- educated dual-earner couples with gender-equal orientations are particularly interesting phenomena to be studied in the Italian familialistic context (Keck and Saraceno 2010), in which “traditional” gender expectations towards mothers and fathers tend to be persistent. The aims of the paper are: 1) to illustrate how it is possible to reconstruct the negotiation processes and the dynamics of doing and undoing gender around parenthood during the transition to the first child in the Italian context using an "integrative approach" to the study of gender construction and the perspective of gender as practices (Risman, 2004; Connel, 1987); 2) to discuss the use of a tool – the couple’s synopsis – for handling the nested data structures in longitudinal interview analysis and its potential for comparative analysis. The study is based on longitudinal interviews with 21 dual-earner Italian couples, living in Turin and in its surrounding area. The partners were interviewed separately before and after their first child birth. The majority of respondents were university graduates. Hence, the sample was intentionally biased towards "gender oriented" couples. Most women enjoyed a similar position than their male partner in terms of education and labour market status. The corpus of interviews was subjected to content analysis (Smith, 2000). A double process of analysis was carried out: first, a coding process performed with ATLAS.Ti, Computer Assisted/Aided

322

Qualitative Data Analysis Software (CAQDAS); secondly, construction of a synopsis for each interview. The synopsis is a short longitudinal report on the main issues regarding the couple: it is structured in order to support the theory-building process. Three key issues affecting qualitative couple interviews analysis can be addressed through the synopsis: 1) intersubjectivity; 2) synthesis without losing the richness of information; 3) emerging theories. The paper will discuss these three issues within the interpretative process of constructing the team analysis project. Moreover, the paper will discuss a set of indicators suitable for analyzing the mechanisms of "doing gender" (gender ideology) or undoing gender in the Italian context and for selecting those institutional factors that lead the couple in unexpected directions as they negotiate. This discussion may point out the most relevant topics to investigate comparatively. Keywords: longitudinal qualitative analysis, transition to parenthood, gender

323

Nascimento, L. (Federal University of Bahia – UFBA) From codex to autocoding: reflections about CAQDAS, Digital and Figurational Sociology The characteristics of culture and society were profoundly altered in the face of “ubiquitous” or “pervasive computing” that modified the way we search for and produce information regarding all aspects of social life. However, the social sciences still analyze this phenomenon in a partial manner, neglecting the effects of digital technologies in society and especially in sociology itself. Following Elias’s Figurational Sociology and Roger Chartier’s proposition that the transformation of the production and transmission of texts is also a fundamental epistemological mutation, the paper argues that the use of CAQDAS by sociologists affects the process of producing sociological knowledge. The possibility of inspecting a vast amount of documents (texts, images, music and videos) that are the objects of sociological research has a considerable effect on the sociological imagination. Despite claims that we are undergoing a crisis of sociological methodology, the article argues that the advent of digital technologies represents a historical moment for an upgrade of sociological tools, especially investing in the use of CAQDAS in digital social research.

324

Nešporová, O. (Research Institute for Labour and Social Affairs in Prague, the Czech Republic) Qualitative panel data – three waves of interviews with parental couples As a part of the TransParent Research Network project, three waves of interviews were conducted with couples in the Czech Republic between 2011 and 2016. A total of 16 couples were interviewed concerning their future plans and, subsequently, regarding their everyday experiences with parenting. The semi-structured interviews focused on the topic of combining parenting with paid work and the division of gender roles between parents; both partners were interviewed separately. The first wave of interviews took place during pregnancy and the second wave of interviews was conducted once the child had reached around one-and-a-half years old. The third wave of interviews was conducted when the (first) child was around three-and-a-half years old. Thus, the research team monitored the experiences, plans and notions of the parental couples for a total period of four years. In this session, I propose to address the three most challenging parts of the project based on the team’s experiences from the conducting of the research itself and the analysis of the data gathered: 1) the retention of the respondents interviewed for research project purposes for the subsequent wave of interviews; 2) the storage of the data; 3) the coding process and the analysis of the transcribed interviews. Due to the fact that the coding process itself forms a key part of the analysis, I intend to focus principally on the third point. Types of codes and their families will be introduced with regard to coding. At this stage, the research team was faced with significant problems caused by temporality. Firstly, interviewees referred to a large number of time periods/levels even in a single interview which in fact produced statements of varying quality (references to the past, present and future) which had to be reflected in the coding process. Secondly, the longitudinal character of the research rendered the coding process increasingly more complicated with each wave of interviews (the changing nature of plans and reality in time). As far as the analysis is concerned, one of the key issues consisted of the choice of the most suitable units of analysis for specific topics and research questions. The most interesting results were provided by the within- couple and international comparisons. Finally, I intend to summarise the problems involved in the reduction of the rich and broad collection of data and the process leading to the

325

identification of key topics. Once more, the time question was crucial; should the data from all three waves of interviews be employed or should one or two waves only be chosen? Keywords: Longitudinal research, qualitative interviews, coding

326

Neumann, R. and von Hermanni, H. (Technische Universität Dresden) Charitable giving in the field - evidence from a quasi-experiment at bottle refund machines in Germany The study of the emergence of a social norm of helping strangers by donating money to charities or non-profit organizations has been scrutinized from different perspectives. Theoretical approaches like warm glow behaviour or feelings of social obligation to help others have mostly been tested using either laboratory settings that come with the usual drawbacks or as field experiments during survey interviews that were conducted to mobilize donors. Additionally, a few field experiments have highlighted the importance of framing decision contexts by providing descriptive cues about the social expectations to enhance contribution to a collective good. Using a unique research design from a supermarket chain in Germany, we systematically alter the information provision of the decision context at bottle refund machine that has the option to donate the refund. The talk will address problems of implementing the intervention, it will present the designs that were used to control for potential biases and will address the challenges that arise regarding the choice of the appropriate matching strategies to allow for the identification of causality. Keywords: charitable giving, social norms, matching

327

Nigel, F. (University of Surrey) Changes and continuities in online social research The affordances of online technologies for social research have increased sharply since the arrival of Web 2.0. We have grown used to talk of ‘the data deluge’, ‘Big Data’ and ‘transformative technologies’. The new research environment undoubtedly makes for change but longer-standing realities still lie behind ostensibly new practices. The interaction between new practices and established realities marks both our daily research routines and the wider political economy of social research. Emergent trends include the growth of citizen research, the challenge from Indigenous Knowledge to western-centric methodological assumptions, and renewed struggles over the purposes and status of social research. Some fear such developments as a threat to our disciplinary authority, while others applaud the end of yet another professional monopoly. This talk considers what has changed, and what has not, in four domains: Analyzing Online Materials; Collaborative Research; Citizen Research; and the Role of Social Research in the New Environment.

328

Nkolo, J. C. (Economic Development Initiatives – EDI) Using CAPI to Improve and Evaluate the Quality of Socio-Economic Surveys Socio-economic research has been gradually moving from paper questionnaires to electronic surveys. Computer Assisted Personal Interviewing (CAPI) is undoubtedly the most advanced progress as CAPI handles complex, multi-topic and dynamic questionnaires, linking hundreds of variables and dozens of tables. The surge in CAPI, especially in developing countries, has been allowed by the availability of cheaper ultra-mobile PCs and tablets, and great efforts on the part of developers to make a technology at the service of research and focused on the quality of the data collected. In developing countries, a wealth of data is collected through face-to-face interviews. And a large strand of the socio-economic literature stresses measurement errors arising from these surveys (e.g. implausible responses, poor interviewer performance, etc.). Recent studies illustrate the superiority of CAPI for data collection. Indeed, using CAPI reduces many errors by adding automated routines (skips), validation checks (to detect inconsistencies), etc. In this paper, we provide insights on how CAPI technology can be used by researchers and survey practitioners to evaluate survey measurement quality. Although, there is now a general consensus that CAPI significantly improves data quality and reduces measurement errors, research on how CAPI can be used to validate survey finding remains sparse. To address these questions, we build on our experience in developing a CAPI software, Surveybe, and in implementing dozens of large-scale surveys in East Africa. These developments are the results of regular interactions between software developers and researchers and have led to several improvements allowing to evaluate survey measurement quality. Firstly, aspects of CAPI that reduce errors during the actual survey itself will be discussed. Secondly, features of CAPI that allow an overall evaluation of the quality of data, both during and after fieldwork, will be considered. Keywords: CAPI ; Surveybe ; Questionnaire design; Paradata; Images; Time stamps; Measurement; Quality Control; Developing countries

329

Nkolo, C et al. (Economic Development Initiatives, surveybe) Experiences and Challenges in Data Collection Monitoring from Bukoba (Tanzania) to High Wycombe (UK) Experiences and Challenges in Data Collection Monitoring: From Bukoba (Tanzania) to High Wycombe (United Kingdom) This paper draws from the experience of EDI Group (made up of two sister companies based in Tanzania and the UK respectively, EDI Limited and Economic Development Initiatives (EDI) Ltd) in setting surveys in Tanzania and other East African countries. The focus is on how best practices in terms of monitoring data collection are implemented to provide high quality data within an inclusive and sustainable partnership between our offices in the United Kingdom and in Tanzania. The goal of this contribution is to add to current discussions about international settings for monitoring data collection in developing countries. Usually, an international firm is in charge of the technical expertise and partners with a local organization for the local knowledge. Our own experience shows that combining both within a unique organization is possible and sustainable. Our contribution is twofold, we show (i) how the tools we use enable us to produce high quality data (ii) how our organization's setting promotes constant improvement of current international best practices in data collection. We have a collaborative approach at every stage of the data collection process which is allowed by our team structure and the use of our Computer-Assisted Personal Interview (CAPI) software, Surveybe. Surveybe allows our teams to work simultaneously during the preparation of survey tools (questionnaire design, skip patterns, translation, etc.), during field work and quality control activities. Every stage of the data collection process is jointly operated by UK-based teams and Tanzanian teams, both having the technical expertise to use Surveybe. Moreover, our systems allow both teams to be alerted of any issue during field work and react swiftly. Thanks to the technical expertise of both teams and their proficiency in Suveybe and data analysis and statistical softwares, quality control activities and data cleaning are jointly operated in a cost-efficient and inclusive way. Having UK based CAPI experts working hand in hand with local teams is key to ensure high quality data especially during quality control activities. Their use of paradata facilitates the process.

330

Over the past 14 years, we have successfully implemented dozens of large-scale surveys mainly in Tanzania and other East African countries. The way we address many challenges considerably decreases the risk of measurement errors.

Keywords: developing countries, CAPI, surveybe, paradata, quality control

331

Nosek, V. and Netrdova, P. (Faculty of Science, Charles University in Prague) Individual and regional determinants of unemployment in Czechia: multilevel analysis The regional and spatial variability of socio-economic indicators, their determinants, and evolution in time, are one of the biggest research topics in geography. In this study, we focus on unemployment as a complex indicator of socio-economic performance of regions and we examine the relative influence of individual and spatial/regional factors in determining the risk of unemployment. The main research questions of this study are: to what extent does unemployment in Czechia vary based not only on personal characteristics of individuals, but also according to specific areas in which they live? What is the role of “space” in determining the risks of unemployment? To answer these questions, we use multilevel regression approach, which provides an appropriate and comprehensive framework for analysing individual and contextual influences on the risk of unemployment. It enables us to analyse data with nested or clustered organization. This is typical for geographical data, which are hierarchical or multilevel in nature – people are nested in localities such as municipalities, municipalities are nested in regions etc. The multilevel analysis overcomes disparities between models constructed on micro-level and macro-level. At the same time, multilevel modelling deals with the basic problem every geographical research is facing, i.e. the choice of appropriate hierarchical scale of analysis. Using data from 2011 Czech census, the individual and regional determinants of unemployment among population at working age are studied. To reveal the real and long term contextual effects we work only with population living in the same region where they were born. All undergoing analyses are two-level with one individual (micro) level and one regional (macro) level. The regional levels differ in respective analysis (municipal level, NUTS3 level, NUTS4 level), thus by comparing the results, some conclusions can be made regarding the significance of different regional levels in determining the risk of unemployment. The individual data include the information about gender, age, education, marital status etc.; the regional data are partly aggregate from individual data (unemployment rate, demographical and educational structure) and partly gained from official statistics (e.g. GDP per capita).

332

Nóżka, M. (Jagiellonian University Institute of Sociology, Poland) Mental maps of people experiencing exclusion: A study on the effect of subjective experience of exclusion on the way of valuing and mapping physical space. Mental maps are a rich source of spatial knowledge. For many years there have been implemented projects aimed at identifying information gathered in them that allow to understand the process of perception, human needs, social relationships related to being and functioning in certain areas. Sociologists focused on exploring this kind of knowledge, likewise psychologists, use the term representation for an adequate description and explanation of the behaviour of people exploring the world and solving various cognitive issues (cf. Nęcka and others. 2011: 27-28). One of the incentives for applying cognitive mapping in research entitled The social closing of space. The influence of social exclusion on perception and cognitive schematisation of physical space and territorial behaviour of excluded people, the results of which will be presented as a part of the presentation, was to identify new areas of analysis and deepen knowledge of the phenomenon of exclusion. Equally important was the attempt to determine whether schematic maps may be a source of knowledge about the consequences of being excluded, social dynamics and spatial relationships. As a result of the applied methods - schematic maps of the area and mental walk - was identified the potential of knowledge behind them. Possible and relatively consistent way of analysing the empirical material collected by means of them was proposed as well. The interesting thing was not so much to identify such or other ways of "drawing" the neighbourhood by its residents, but appropriate for people ways of thinking about space and being in space, hidden in those drawings. Schematic maps with the accompanying narration revealed layers of meanings, but also the lack of meaning, invalidity, insignificance of an inhabited space or its individual components. The schematic maps developed by the researchers turned out to be an interesting diagnostic tool, allowing to get to know the different ways of organising spatial knowledge and the people living in a given space, their relationship with the place, relationships with other people, their own relationship to each other. In the framework of the presentation the following issues will be

333

discussed: the research method used, fixed variables differentiating the appearance and content of schematic maps of the environment, as well as - proposed on their basis - the typology of schemes of mental maps on: [1] a road network with objects, [2] a road with objects, [3] a road network without objects, [4] home based systems, [5] objects without roads, [6] fragmentary systems. The proposed division of schematic maps served to systematise the collected empirical material, it was also to indicate the potential of applying this type of tool to study social reality.

334

O'Connor, H. and Goodwin, J. (University of Leicester, England) Fieldnotes, Marginalia and Paradata in Youth Employment Restudies, 1960-1985 Drawing on our experience of secondary data analysis of both qualitative and quantitative historical data sources, this paper highlights the importance of fieldnotes, paradata and marginalia when revisiting studies from the past. We will use examples from three historical studies of youth employment, from the 1960s and 1980s, to illustrate the importance of these by-products of social research to the secondary analyst.

335

O’Dwyer, A. (University of Roehampton) CDA: Reflecting on problems associated with interpreting language & contextualizing discourse This abstract intends to explore methodological problems associated with the Critical Discourse Approach (CDA), in particular how as a social scientist we can as Foucault states analyse discourses as “practices of power/knowledge and meaning production” rather than just focus on “language in use” (Foucault, 1972; 1973). The critique of CDA and language interpretation is set within the context of my research into the process and techniques of performance management policy in the Further Education (FE) sector, and the challenges the policy may present for teachers with Specific Learning Difficulties (SpLDs). CDA, has been used in a number of ways, however, it is argued by Martin and Wodak (2003), that it is ‘fundamentally interested in analysing opaque as well as transparent structural relationships of dominance, discrimination, power and control manifested in language’. However, one of the key criticisms of CDA is that although it can assist in the discovery of ‘obscured’ patterns in language use, in particular what ‘knowledge patterns are produced and reproduced’ it fails to address the question of ‘why?’ the patterns exist in the first place and ‘why?’ the ‘knowledge patterns’ continue to be reproduced; in short it does not comprehensively address the ‘root’ of the power and control which manifest in language and therefore the ‘root’ of why a discourse(s) continues to perpetuate an unequal ‘subject’ remains unclear (Mengibar, 2015). In the context of my research, critically analysing discourse within the text and language of ‘performative’ and ‘equality’ policy documents alone, will not uncover the economic, political and social context in which the discourses are produced. The Sociology of Knowledge Approach to Discourse (SKAD) which draws on Foucauldian ‘tools’ and places them in the wider Berger & Luckmann’s framework, could assist social scientists to move the analysis beyond the ‘analysis of a single set of data’ (Keller, 2005). Moreover, Fairclough (1985) states, often ‘interconnections’ and ‘chains of cause and effect’ are unclear and somewhat contorted, the “critique” is what helps to make the ‘interconnections’ clear (Fairclough (1985) cited in Wodak & Meyer, 2001: 2). It could be argued however, that language contextualisation is subjective and imposed by the researcher themselves, the

336

“critique” and ‘interconnections’ will only illuminate the ‘chains of cause and effect’ that are evident within the specific language analysed. My own research will take place in two spaces, the first ‘space’ will be within the teaching environment of two FE colleges and the second ‘space’ will consider other relevant agencies and organisations that are part of the policy cycle process. I intend to address the issue of the historical economic, political and social forces in which the discourses are produced (Mengibar, 2015). This abstract has attempted to briefly highlight some of the limitations of the CDA approach and the methodological problem of uncovering the power relations in language. Although, CDA is undoubtedly a useful framework for deconstructing texts, analysing language and understanding the ‘construction’ of knowledge, it falls short on why patterns in unequal power/knowledge relations exist in the first place (Mengibar, 2015). Keywords: Critical Discourse Analysis (CDA), critique, economic/social/political context

337

Odella, F. (University of Trento) Longitudinal network data and time sequencing of structural changes. An analysis of serialization and competition in the organization of a cultural market. The paper compares methods for analysing multilayer relations and illustrate results from the analysis of a longitudinal networks dataset. The aim is to discuss two particular issues. First, to review procedures of structure reduction in complex network database and second to evaluate applications of MCA to two-mode networks. Examples will rely on a longitudinal dataset concerning mass media market in the nineties and specifically TV serial production in the private/public broadcasting sectors. Cultural production markets are usually described in terms of affiliation networks or chains of collaborative relations; several analysis highlighted innovation and cooperation emerging from such organizational structures and put in evidence how market structure and organizational strategy shape the diversity and form of popular media products. Macro level factors such as technological and institutional changes, however, are also capable drivers of change in cultural markets and their influence could affect alliances and hierarchies in the networks that link external producers to the creative personnel. The analysis proposes to explore the competitive dimensions of serialization in cultural markets through time and the multilayered relations between broadcasting companies, tv serial producers and media professionals focusing on dependence of network structural configurations. Multiple Correspondence analysis (MAC) is thus applied to obtain a trajectory for changes in inter- organizational relations, identify specific time intervals that characterize networks structural changes and shed light on the dualities of relations between media production companies.

References Borgatti S. and M.G. Everett, ‘Network analysis of 2-mode data’, Social Networks, n.19, 1997, pp.243-269. Cattani G. and Ferriani S. ‘A core/periphery perspective on individual creative performance: social networks and cinematic achievements in the Hollywood film industry’, Organization Science, vol.19, n.6, 2008, pp.824-844. D’Esposito, M., De Stefano D. and Ragozzini, G. ‘On the use of Multiple Correspondence Analysis to visually explore affiliation networks, Social Networks, 38, 2014, pp. 28-40. De Nooy W. ‘Networks of action and events over time. A multilevel discrete-time event history model for longitudinal network data’, Social Networks, n.33, 2011, pp.31-40.

338

Everett M.G. and Borgatti S., ‘The dual projection approach for two-mode networks’, Social Networks, in press, 2012. Faust K. ‘Using correspondence analysis for joint displays of affiliation networks’, in Models and Methods in Social Network analysis, by Carrington P., Scott J. and S. Wasserman, 2005, Cambridge University Press. Jones C., ‘Co-evolution of entrepreneurial careers, institutional rules and competitive dynamics in American film, 1895-1920’, Organization Studies, vol.22, n.6, 2001, pp.911-944. Jones C., Hesterly W., Fladmoe-Lindquist K., and S. Borgatti, ‘Professional Service Constellations: how strategies and capabilities influence Collaborative stability and change’, Organization Science, vol. 9, n.3, 1998, pp. 396-410. Roberts J. M.’ Correspondence analysis of two mode network data’, Social Networks, n.22, 2000, pp.65-72. Keywords: affiliation networks; market structure; time intervals

339

Ogresta, J. and Ružojčić, M. (Department of Social Work, Faculty of Law, University of Zagreb) The Quality of Retrospective Reports about Occupational History in Web Survey Imperfect memory during retrospective reporting is one important source of response error that compromises data quality. Literature review revealed that calendar instruments have demonstrated considerable potential to enhance the quality of retrospective reports. Although methodological studies show positive results regarding calendar effects on data quality in off-line research, there is still lack of empirical evidence about effects of web-based calendar instruments. The aim of this study was to compare occupational histories of young adults collected via calendar instrument in an off-line research with those collected using calendar instrument within self- completed web survey. We evaluated the effect of calendar mode for four quality measures: response/break off rates, completeness and consistency of retrospective reports and interview duration. The occupational history data provided information about young adults’ employment status in each year from the year of completing their highest level of formal education to the survey year. The data were collected on a sample of young adults who grew up poverty. They were recruited from families registered as welfare recipients during their children’s high-school education period in the Social welfare centers in Croatia. The results will be discussed in terms of their implications for tailor elements of web-based calendar instrument. Key words: calendar instruments

340

Onaka, F. (Japan Women's University) Examinations as a Pivot of Social Processes Examinations have been a focus of sociological studies in Japan, yet there have been drastic changes in how they are perceived. Examinations were previously understood—for example, in the 1950s, as “a way of fairly distributing personnel” (Yoshihiro Shimizu) and, in the early 1980s, “an objective measurement of ability” (Ikuo Amano). However, in the late 1980s, influenced by Robert John Montgomery’s view in the 1960s of examinations as “an administrative weapon” and by Michel Foucault’s view of examinations as “a center of procedures which construct individuals as effects and objects of power” in the 1970s, I proposed a political viewpoint: “examinations as ritual.” Subsequently, influenced by Pierre Bourdieu, Takehiko Kariya argued that examinations are “mechanisms of stratification” in the 1990s and Takayasu Nakamura, influenced by Anthony Giddens, argued that examinations were “reflections of society” in the 2000s. I tried to integrate these discussions into a new reformulation, the “socio-cultural network” in the 2010s. This paper attempts to proceed this new reformulation and argues that examinations comprise a pivotal socio-cultural network of social processes. This formulation emphasizes that examinations produce social processes that are either static or dynamic. In fact, examinations contribute to maintaining the basic structure of society and examination reform has induced huge changes. This viewpoint highlights the methodological significance of examinations: examinations are among the essential foci in the investigation of social processes.

341

Ortmanns, V. (GESIS – Leibniz Institute for the Social Sciences) Can survey characteristics help? - Trying to explain inconsistencies in the distributions of the education variable with survey characteristics The educational attainment variable is a key variable and it is often use to identify respondents’ position in the social structure of the society. In quite a large number of analyses the education variable is inserted as control variable which often turns out to be significant. Consequently, respondents’ attitudes, beliefs, values, and behaviors are influenced by their education. What underlines the importance of this variable is that it is covered in (nearly) every survey. We had a look at the education distributions of seven cross-national surveys, including official, politically- motivated, and academic surveys for the time period from 2008 to 2012. The different data sources are compared to the official data of the EU-LFS, which are used as reference data, by calculating Duncan’s Dissimilarity Index. Using this index we identified inconsistencies across surveys, across time and, also across countries and we now want to explain those. The research question of this paper is: Can survey characteristics explain some of the observed inconsistencies in the distribution of the education variable? Duncan’s Dissimilarity Index is used as dependent variable in the analysis. We try to explain the identified inconsistencies through different survey characteristics, those serve as independent variables. Survey characteristics from the measurement and from the representation side of the Total Survey Error framework are taken into consideration. On the measurement side, we, for instance, assume that the survey instrument is a likely candidate for explaining the identified inconsistencies in the education variable across surveys. The number of answer categories on the showcard of the education question should vary strongly across countries, because of the different educational certificates and systems. In one survey we observe a quite stable number of answer categories across all countries - that is pretty unusual. This sheets substantial doubt whether these categories consist of appropriate and understandable terms for the respondents of the different countries. Thus, different measurement instruments could perhaps explain some of the inconsistencies. Factors on the representation side we take into consideration are, for example, the response rates and the survey mode. Response rate is a most commonly used indicator for representativeness of a

342

sample. A high response rate is often linked to a more representative sample but this is only the case when there is no systematic shift between the target population and the surveyed participants. Regarding the survey mode this is nearly the same: If a certain property with specific characteristics is not covered a bias is likely to occur regarding the representativeness of the sample. This is e.g. the case with the online mode which is likely to over represent higher educated and to underestimate lower educated people. Both survey characteristics, the response rates and the survey modes, can possibly explain some of the inconsistencies in the distributions of the education variable. Further survey characteristics, we will insert in the analysis are the sampling frame, proxy reports, obtaining data from population register, fieldwork time, and the distribution of male and female respondents of heterosexual couples in the sample by using Sodeur’s method. The data and survey characteristics will be analyzed using some fixed- effects and regression models.

343

Otte, G. (Johannes Gutenberg University Mainz, Germany) Using mixed methods to investigate social inequalities and symbolic boundaries in a local club and discotheque market In many sociological portraits of young people’s leisure time practices and consumption patterns adolescents are depicted as free to choose from a “supermarket of styles”. Youth is often seen as a moratorium in the life-cycle that dispenses with the status hierarchies typical for adult life in class-based stratification systems. This paper uses the under-investigated case of the urban nightlife economy to study processes of symbolic boundary-construction and structures of social inequality in young people’s music scenes.

In a large multi-methods research project, music scenes were accessed via clubs and discotheques in the city of Leipzig, Germany. The central building block of the data collection are quantitative audience surveys in twelve sites varying in size, organizational form and musical genre. These data are used to study the composition of audiences in terms of socio-demographic attributes, music and bodily capital and additional behavioral and attitudinal measures. Based on the survey data and using techniques of multiple correspondence and regression analysis, a picture of an urban nightlife market emerges that is segmented along multiple hierarchies of social inequality.

These data do not show, however, if and how objective inequalities are reflected in the subjective perceptions of the participants. Therefore, group discussions are used as a second component in an integrated research design. Based on the statistical results of the audience surveys three clubs were identified for further study. They can be considered extreme cases in the two-dimensional market space. A few months after the standardized interviews typical visitors were recruited in these clubs for in-depth investigation. Their typicality was defined on the basis of the socio-demographic audience profiles, bodily appearance and frequency of attendance. The selected regulars were invited to group discussions along with their clique they usually go out with. The group discussions are a suitable method to study the discursive

344

construction of symbolic boundaries in urban nightlife. According to our interview material, young people use three discursive perspectives to gain a positive identity: they identify with their favorite club and its audience; they look for symbols of subtle distinction within the in-group audience; and they draw boundaries against devalued out-group audiences.

The combined analysis of quantitative and qualitative data follows a double intention: on the one hand, the specific advantages of both methods are used to produce results that complement each other; on the other hand, the data are used to validate each other. Substantially, the paper shows how traditional bases of social inequality (especially class, gender, and age) and scene- specific forms of capital (especially music and bodily capital) interact and, together, produce a market of urban nightlife that is hierarchized along multiple lines. As this topic can be seen as one instance of the general challenge of integrating socio-structural and cultural perspectives as well as quantitative and qualitative methodologies, implications are drawn for future research.

Keywords: Mixed methods, survey research, group discussion

345

Oud, J., Voelkle (Radboud University, Nijmegen, Netherlands; Humboldt University, Berlin, Germany)

SEM Based CARMA Time Series Models for Arbitrary N This presentation consists of two parts. We begin the first part by reviewing that the SEM continuous time procedure is appropriate for N = 1, as it is for the large N case (Singer, 2010). Although software to do so is now readily available (Driver, Oud, & Voelkle, in press), this is not well-known and applications are scarce. We then introduce the general continuous time autoregressive moving average CARMA(p,q) model and discuss its relations to the discrete time ARMA(p,q) model. A simulation study will be presented in which the same model is estimated for an entire sample of N cases and for each of the N = 1 cases separately. The second part of the presentation relates to the increasing interest in the antonym of inter-individual and intra- individual research, particularly the question whether results found on the group level are reproducible on the individual level and vice-versa (Molenaar, 2004; Molenaar & Campbell, 2009; Song & Zhang, 2014). In a combined analysis of an N = 1 within-person model and an N – 1 between-person model, Voelkle, Oud, von Oertzen, & Lindenberger (2012) designed and performed a test to compare the results on the individual and group level. In the simulation, this test will be performed to find out its validity for the continuous time case. We end with an empirical example on the relationships between mood at work and mood at home, to illustrate the potential of SEM-based CARMA time series models for arbitrary N in the social sciences. Keywords: CARMA, Time Series by SEM, Arbitrary N

346

Pennell, B. – E., Hibben, K. C., Lyberg, L. Mohler, P. and Worku, G. (University of Michigan) A Total Survey Error Perspective on Comparative Surveys Although the TSE paradigm is increasingly used as an organizing framework in the design and evaluation of one-country surveys (Biemer, 2014), it is rarely cited in reference to comparative or what are increasingly referred to as multinational, multiregional, and multicultural surveys (which we refer to as “3MC” surveys for brevity). In a one-country survey, the design challenge is the optimal allocation of resources to minimize TSE for key measures, keeping in mind constraints such as costs and time, among others factors. In the context of 3MC surveys, the question of optimal allocation is a much more complex set of decisions with many more possible constraints on design options. Pennell et al. (2010) and Pennell & Cibelli Hibben (forthcoming) detail the numerous aspects of survey context affecting survey design features across cultures and nations, which make 3MC surveys highly complex undertakings. These features include varying levels of access to up-to-date or good quality sampling frames, challenges in designing questionnaires that are culturally relevant within and across countries, accommodating multiple languages (some possibly unwritten), the ‘survey climate’, locating respondents in sparsely populated regions, locating ‘hidden’ populations (e.g., homeless), refugees, internally displaced populations, migrant populations, conducting interviews in fragile states or after natural or man-made disasters, religious and other taboos, gate keepers at the state, regional, village, or household level, differing levels of telecommunication and internet access, lack of research infrastructure, and different research traditions, among many others. Although one might encounter any of these challenges in a one-country survey, the probability of encountering several or many of these challenges in quite varying arrangements is much higher in a multi-country survey. The TSE framework we present in this talk links error sources to the key stages of the survey process: design; implementation; and evaluation. We outline representation error – including coverage error, sampling error, nonresponse error, and adjustment error, which are indicators

347

of how well survey estimates generalize to the target population and then turn measurement related error – including validity, measurement error, and processing error - which are indicators of how well survey questions measure the constructs of interest. We also incorporate the dimensions of cost, burden, professionalism, ethics, and other design constraints that frequently impose constraints on 3MC survey design and have an important influence on 3MC survey quality. Finding the optimal balance among these sometimes competing goals is at the heart of comparative survey methods research. Finally, the framework identifies, for each error component, key potential sources of error that may contribute to TSE in individual populations and may present particular challenges in standardizing design and implementation (or establishing suitable localized equivalents) across populations, thereby potentially increasing comparison error. We discuss these key potential contributions to error in our discussion of design and implementation challenges across the main stages of the survey lifecycle.

348

Pereira, M. J. (Paraná State Univesity, Brazil) Capturing Emotion in the Statements of German Political Prisoners in Paraná during the Second World War The Social Sciences has taken openness to analysis of the political, economic and cultural , however, our concerns focus on how to analyze the emotions are perceived within the political field. This article presents the ramifications of institutionalized action of Paraná political police , represented by the Bureau of Political and Social Order ( DOPS / PR ) , seeking to find in the documentation investigated signs of political and social resentment toward individuals of german origin. Also intends to clarify what pipelines and emotional behaviors served as sources provocative and undermined relations between the Brazilian community and immigrants of germanic origin between 1942 and 1945. Based on the analytical assumptions proposed by Norbert Elias , we aim to understand the repressive action taken by the political police was only an offshoot of the nationalist project of Vargas , it stemmed from the common suspicion in principle between countries that are on opposite sides during the Second World War or , they had indications that a process of denial that german element for reasons that go through establishment of issues , control of drives , manifested through reactive feelings like resentment, bitterness and hatred . The sources consist of Autos statement obtained by the police during the repressive process Paraná between 1942 and 1945, documents kept by the Paraná State Public Archives, since they keep documentation produced by DOPS / PR and also by individuals who have been detained, personal documents such as quotes, charts and reports that allow us to clamp elements to think about the goals in this article. Keywords: Emotions, Germans, Political Police.

349

Petschick, G. (TU Berlin) Researching social processes combining ethnography and interviews This paper mentions to show how a combination of ethnographies and interviews can be used to investigate social processes which occur during a period form several years combining the efforts of these two methods: Many issues investigated in ethnographies are complex and lengthy processes. The method of ethnography allows in principle to accompany such a process in-the-making, including its fractions, critical moments as well as the envisaged alternatives and "dead ends". So tacit knowledge about practices and routines can be experience at first-hand. At the same time real- time studies of processes with long field stays are very resource consuming and not feasible for many ethnographers. Furthermore, such processes are often not localized in space and time, i.e. not limited to a specific place or group of people. This confronts the ethnographer with the unsolvable situation that he cannot be in several places at the same time and thus may not entirely follow possibly parallel decisions. In interviews, however, non-observable aspects, such as interpretations, reflections and thoughts about situations and decisions can be asked. The possibilities of purely interview based studies are limited, as they lack the detailed knowledge of the field stay and only the explicit knowledge of the process is ascertainable. The presented method of attempts to combine the advantages of these two methods for investigating processes. The author proposes Ethnographic-Panels as a mixed-method of repeated, short participant observation and retrospective interviews. The time spend in the field can thus be reduced. While intensification further allows a process accompaniment. The present article outlines the method and illustrates the research approach with an example, the social construction of scientists: The participatory observation were realized over a period of three years and supplemented by interviews. Participating at the everyday work of young researchers, especially PhD candidates and postdocs, I tried to identify mechanisms of exclusion and inclusion and thereby to understand how the academic field operates. Supposing that there is an academic personality how is socially constructed, the aim is to understand how young researchers become like this when successfully trained through the academic career, especially in excellent research fields.

350

The focus of this investigation are the everyday practices of researchers, since I assume that a number of including and excluding practices and their incorporated implicit attributions, are not made consciously by the actors, but happen interactive and are visible in scientific practices and in habitual actions.

351

Petschick, Grit (TU Berlin, Methods of Social Research at the Department of Sociology) Problems arising by comparing two Case Studies The present study addresses the questions of how scientists are made in the time of the Excellence Initiative and which dimensions of social inequalities, such as gender, nationality and social background of the actors, play a role in the field. In two ethnographic case studies the construction of the scientific personality as well as the production of differences were investigated at the level of everyday operation, i.e. scientific practices and habitualized action. Assuming, that the actors are not aware of a number of differentiating and discriminatory practices and therein incorporated implicit estimations but that these are rather made interactively, the presented research focused on the everyday practices of the researchers. Therefore, two isolated cases studies of two scientific working groups involved with their research in an excellence cluster were conducted using an iterative approach. Young physics and chemistry researchers, especially PhD students and postdocs were accompanied ethnographically in their academic everyday life during a period of three years. By contrasting and comparing the two case studies the work routines of the researchers, their beliefs, their attitudes and expectations of the academic system in general and their current work and research situation in particular were compared in order to make the relevant rules and resources of the academic system visible as well as its mechanisms of inclusion and exclusion. But what if the two cases studies, as in the presented case, through the iterative process develop differently and an asymmetry in the observation time and thus in the amount of data is produced? To what extent a contrasting of the two cases is possible? Related to which categories? Are generalizations possible in this case? These and similar questions will be discussed in the present paper. Keywords: Ethnography, Comperative Case Study, Science Study

352

Petzold, K. and Eifler, S. (Catholic University of Eichstätt-Ingolstadt, Department of Sociology, Germany) Presumed Behaviour and Real Behaviour in a Normative Everyday Situation Survey experiments are increasingly used in sociology. Benefits are often seen in a high flexibility as well as in the usability for the investigation of sensitive issues. Compared to conventional direct measurement survey experiments provide a more detailed description of the situation and are thus closer to reality. It is widely believed that the higher precision provides less room for interpretation and evokes therefore less bias by social desirability. The method was even developed originally, to collect more valid responses on sensitive topics (e.g. via fictitious acting other people). However, systematic studies on the validity of the measurement of behavioural willingness with survey experiments are rare and inconsistent. While some authors indicate a high correlation between reported willingness and real behaviour, others have some doubt regarding the validity of behavioural measurement, especially in everyday situations. The present study compares the results of a survey experiment with the results of a field experiment. To simulate an everyday situation a so called horn-honking experiment has been carried out in a German city, twice, in field and survey. Here, subjects are blocked in their cars randomly at a traffic light by an experimental car. Following former versions of this experiment, status and origin of the frustrator car have been varied by car size and licence plate. While the status effect of previous studies could be replicated in the field experiment, the survey experiment yields an effect of the regional affiliation of the car. These opposite results suggest that the application of survey experiments in normative situations may rather lead to the report of a behavioural willingness that is consistent with relevant norms instead of with real behaviour. This supports the idea to use survey experiments primarily for the measurement of norms and values instead of for the measurement of behaviour, especially in their application for sensitive topics. Keywords Survey experiment, Field experiment, Behavioural validity

353

Plugor, R. (University of Leicester, England) Opportunities and challenges in using administrative data for research purposes: two case studies Administrative data are derived from any data source containing information that was not primarily collected for statistical purposes. This paper focuses on two sources of administrative data: one obtained from several National Health Service Trusts operating in England relating to their workforce, and the other one obtained from a Police Force relating to domestic violence cases in the area. Drawing on the experiences from these two projects, the author presents the opportunities and challenges that arouse in the process of working with the data. The main areas discussed are: obtaining the data, cleaning and preparing the data for analysis, using the data for statistical analysis, data quality and limitations. The paper concludes by providing summarising thoughts and further suggestions for using administrative data from these two sources.

354

Pollien, A. and Stähli, M. E. (FORS - University of Lausanne, Geopolis, 1015 Lausanne, Switzerland)

Promises and failures of nonresponse surveys In order to analyse nonresponse bias, the survey methodologists have become used to conduct non-response surveys (NRS). These surveys are supposed to detect systematic errors that occur because individuals differ in their propensity to participate in a survey according to their characteristics and attitudes. This kind of measures of non-respondent’s attitudes conveyed promise to solve issue of representativeness of surveys sample. The Swiss ESS team conducted several NR follow-ups along with representative face-to-face international surveys: ESS, EVS and MOSAiCH (ISSP). NR follow-ups consist of a paper questionnaire picking questions in the face-to- face survey assumed to be related to the nonresponse mechanism. Shortness and change of mode are supposed to commit a large number of non-respondents to participate. NR follow-ups have two main challenges: The first is that there may still be a group of non-respondents that do not participate in the nonresponse follow-ups: follow-ups extend the participation but don’t provide information about the whole sample. Therefore, when working with results of NRS, researchers assume that there is a continuum of participation, respondents to NRS having characteristics being in between respondents to the main survey and non-respondents to the NRS. In order to account for response mechanism, the NRS must imply the same process of participation but commit more respondents to participate. The second challenge is that the survey variables must be sensitive to response mechanism, but not be sensitive to measurement effects. The variables included in the NRS should be relatively insensitive to time-frame, mode and context (sponsor, layout, surrounding questions, questions order) to avoid the detected differences between respondents and nonrespondents to be due to survey design effects. A paper published in 2015 attempt to identify the pertinent variables across four NR follow-up (Vandenplas, Joye, Stähli and Pollien). The researchers find the selected attitude variables have low levels of reliability. The answers given by the control group in the main survey and in the follow-up are less correlated than expected. Furthermore, they predict weakly the survey

355

participation. Since this study, two additional non-response surveys were conducted. A survey tested the stability of the measurements through the use of batteries of 3 items. Another survey includes an experiment designed to disentangle time and mode effects on measurements: a drop-off was left behind to a first control group of respondents right after the face-to-face interview and a follow-up was sent to another respondents control group at the same time as the nonresponse questionnaire. The results show that extraction of dimensions from batteries only partially stabilizes measurements and that the time lap experiment shows that the temporal shift between face-to- face and paper follow-up survey produces more discrepancies that the difference between respondents and non-respondents. The questions that are reliable are also questions that do not differentiate between respondents and non-respondents. And the questions that differentiate between respondents and non-respondents are not reliable across time. Response mechanism and attitudinal discrepancies interrelate differently in each survey context, so it seems difficult to assess for non-respondents from a NR follow-up. Keywords: nonresponse surveys – measurement effects - reliability

356

Pretto, A. (University of Trento, Italy) Bipolar photo elicitation The aim of my presentation is to propose an interview technique which is still little known, bipolar photo elicitation. The particular feature of this technique is that a set of two images – and not one single photo - is shown to the interviewees. In general, photo elicitation is considered a very suitable technique for the study of experiences and is often used when the researcher is interested in the individuals' interpretations and/or perceptions of specific circumstances or events. So, bipolar photo elicitation is a valuable tool for studying 'specific areas' that touch on themes that could be 'delicate' for the interviewees. Bipolar photo elicitation is very useful when the research object can be explored, basically, from two perspectives, when the object of study can be examined through a comparison, and when the researcher wants to understand or investigate opinions, attitudes and/or perceptions organized around two poles that he/she is unable to perceive directly. In particular, there are three strengths that I have found in this technique compared to photo elicitation with one image: 1) it can reduce the embarrassment of the interviewee; 2) it can increase the equilibrium between the weakness and the structuring of the stimulus; 3) a possible increase of consistency in the analysis phase. Keywords: Non-Moving Visual Data, Comparison, Double Perspective

357

Purwaningrum, F. and Fakhri, A. (The Institute of Asian Studies Universiti Brunei Darussalam, the University of York – UK) Distant or Participant Observation? Navigating the Field amidst Time Restriction in Brunei Darussalam and Indonesia The paper examines the kinds of observation that are helpful for fieldwork in light of time restriction. Restrictions are not unforeseen in ethnographic research. They can range from time, funding, to access to the field. Observation, particularly participant observation, is one of the predominant metaphors in anthropological research (Clifford 1986). Observation is both an active and passive engagement in the field. As method it carries weight as it accords ethnographers an emic perspective, which in fact suggests a subjective, informed, and influential standpoint (Headland, Pike, & Harris, 1990 as cited in Kanuha 2000). Corbetta (2003: 246 as cited in Tjora 2006) identifies different elements that can be incorporated in an observation namely; (1) the physical setting, (2) the participants and their roles and tasks, (3) formal interactions, (4) informal interactions, and (5) the social actors’ own interpretations (by informal conversation and formal interviews). Observation has its associated fieldwork methods, which range from distant or passive observation, informal discussions (storytelling), discussion with key informants in the field, to participant observation of meetings. Different methodological approaches adjusted with particular situation and issue do characterised ethnographic research (Lueders 2002). Yet, entering the field and carrying out observation are not an easy task. Access is oft limited due to a multiplicity of factors, such as; cultural disparity (Ezeh 2003), construction of sexual identities (Brown-Saracino 2014), or local languages (Shtaltovna and Purwaningrum, forthcoming). What are the kinds of observation that ethnographers may find helpful in their fieldwork bearing in mind time restriction? Furthermore, what are the fieldwork strategies that can be mobilised to reduce time yet at the same time maintain such depth and breadth in ethnographic research? The proposed paper will attempt to answer these questions. Topics that systematically investigate the types of observation, how they can be applied in ethnographic research, and their complementing fieldwork strategies are under-researched in Sociology. Interview as a method including its challenges was discussed by Michele Lamont and Ann Swidler (2014). Observation, nonetheless, remains liminal. The paper aims to fill the gap and offers

358

empirical reflection based fieldwork conducted by the authors in two countries: Brunei Darussalam particularly in Tutong in 2014-2015; and, in Bekasi, Indonesia in 2010-2011. The authors used a variety of observations in their ethnographic fieldwork in two different countries in Southeast Asia. The two country case studies presented will highlight the author’s experiences with regard to observation and the kind of fieldwork strategies used to collect data that were advantageous in their research.

359

Quinlan, C. (De Montfort University, England) The Four Frameworks Approach: A new tool for teaching research methods This paper contributes to the development of a pedagogical culture for teaching research methods by showcasing a new approach; the four frameworks approach (Quinlan, 2011, and Quinlan et al 2015). I developed the four frameworks approach from my teaching practice. I developed it specifically to provide students with a model, a simple guide, to help them with their research projects. Using the four frameworks approach, the (student) researcher uses their very well developed research statement or question, the conceptual framework of the research project, to direct the development of the research project. This first framework, the conceptual framework, guides and directs the development of the other three frameworks, the theoretical framework, the methodological framework, and the analytical framework. Drawing on examples of student work, in this paper I will demonstrate how the four frameworks approach works in practice. I will show how the four frameworks approach gives students a model with which they can map the boundaries of their research project. This helps to bolster student confidence, so they can avoid feeling overwhelmed by the scope of the undertaking. The four frameworks approach helps them maintain an appropriate focus while they undertake and complete their research project. The first framework in the four frameworks approach is the conceptual framework. This is contained in the one-sentence research question or statement which the researcher develops from reflection on their research idea and from reading the literature published in the area of their research. Each key word and phrase in that one sentence is a key concept in their research project. The second framework is the theoretical framework. This is contained in the literature review undertaken and written for the research project. The key concepts in the conceptual framework help to guide the researcher towards the content of, and an appropriate structure for, the literature review. The third framework is the methodological

360

framework. This is contained in the research methodology chapter of the thesis/in the research methodology section of the report of the research. This is a complete and thorough account of how the research was carried out. The final framework is the analytical framework. This framework is contained in the data analysis chapter of the thesis/in the data analysis section of the report of the research. It contains all of the detail of the analysis carried out for the research. The researcher is guided in the work of carrying out and presenting data analysis by the conceptual framework, contained in the one sentence research statement/question, the theoretical framework, contained in the literature review, and by the methodological framework, which provides detail on the means through which data for the research project are to be gathered and analysed. The four frameworks approach is explained in detail in this paper. It is a simple useful tool for teaching and learning research methods. It is a tool that can aid in the development of a pedagogical culture for teaching research methods.

361

Quinlan, C., Baldwin, L. and Knight, V. (De Montfort University, England) Methodological approaches to research on emotions in criminology and in criminal justice studies There has in recent years been an increased interest in the study of emotions in social science research. At De Montfort University, we have a particular focus on this through our Emotion and Criminal Justice Cluster. Two of the key aims of this research group are as follows: to explore the topic of emotion within the criminal justice field, and to expand knowledge, research and teaching on this topic. We recently held our first annual Emotion and Criminal Justice Conference. This was attended by academics and practitioners from around the United Kingdom and it provided unique insight into the value and the richness of the research on emotions currently being conducted in criminology and in criminal justice. In our quest to develop a broad understanding of research on emotions in criminology and in criminal justice studies, the kinds of research being undertaken and the perspectives and frameworks within which emotions are being studied within the disciplines, we undertook a review of the literature. A broad survey of the literature was undertaken with a view primarily to highlighting the ways in which emotions feature in research conducted in criminology and in criminal justice. This literature review exposed the different ways in which emotions come into focus in criminology and in criminal justice studies. Our findings with regard to the complex range of contexts and experiences in emotions research in criminology and criminal justice, the essential focus of this literature review, is currently being prepared for publication. In this presentation, we focus on critically examining the methodological approaches that were used in the emotions research that featured in this literature survey. We present our findings regarding the methodological approaches used and our evaluation of the methodological approaches used with regard to their relative strengths and weaknesses. We present an overview of the methods and methodological approaches used in emotions research in criminology and in criminal justice studies, and we critically examine their use and efficacy in the context of research on emotions in criminology and in criminal justice studies.

362

Reeves, L. S. (University of Queensland, Australia) Challenging hierarchies: The role of visual research methods in participatory development communication Recent years have seen a broad, interdisciplinary move towards participatory and collaborative research methods, particularly those involving visual materials. Alongside this trend, visual research methods have become increasingly popular within the international development sector, and advocates for participatory approaches particularly favour methods that incorporate visual materials. These methods are frequently used across the sector with little theorising or critical reflection. As such, several scholars argue that this has led to multiple cases where participatory development communication (PDC) approaches have been implemented based on what appears to be the assumption that visual automatically equals participatory, and participatory automatically challenges existing power structures. This research modestly challenges this assumption, and aims to identify some of the potential ways that visual research methods practice can align with key participatory development communication principles in order to encourage more critical and reflexive practice. The research focuses on a specific case study around a workshop held in rural Papua New Guinea in 2013. The workshop employed a visual multi-method approach over four days with the purpose of designing a local community development plan. Twelve local male farmers, fisherman and community leaders participated in the workshop. The workshop was broken up into six sessions, each incorporating a different visual research method. These methods included participatory mapping, 3D model building (based on the LEGO® SERIOUS PLAY® method), participatory photography using both digital and instant-film cameras, site visits (similar to transect walks), album making, and a group discussion with a matrix ranking activity. I used participant-observation of the workshop to retrospectively determine whether the use of visual research methods demonstrated evidence of challenging existing power structures, both between participants themselves and between participants and the research team.

363

I obtained 26 hours of video footage, 1543 still images, 54 pages of field notes, and almost 16 hours of audio recordings (from debriefing meetings, segments of the workshop, and other discussions). These observations were complemented by a series of semi-structured interviews with a community leader who was also a participant, and with one of the facilitators of the workshop. Using a pattern-matching or congruence method, it was expected that the visual research methods employed throughout the workshop would demonstrate observable evidence that the visual aspects of these sessions shifted power dynamics. Perhaps unsurprisingly, I found that while the visual research methods used in Papua New Guinea demonstrated evidence of shifting some power structures, this was not necessarily because of the method or methods themselves. I argue that these findings demonstrate the need for greater reflexivity within participatory development communication practice, and posit that visual does not necessarily mean participatory, and participatory does not necessarily mean equality. This research builds on previous work by Dr Lauren Leigh Hinthorne and Dr Lilly Sar, and I am very grateful to both of them. Keywords: Visual research methods, participatory development, Papua New Guinea

364

Redd, R. and Veltri, G. (University of Leicester, Leicester) The Analysis of Cultural Processes and Social Identity Formation in the Re.Cri.Re Project This presentation focuses on the methodological challenges that underpin the Horizon 2020 funded Re.Cri.Re project. The main purpose of Re.Cri.Re. is to understand what kind of changes are affecting the way European citizens represent themselves and their life context, how these changes are associated with the 2008 financial crisis, and what these changes in local understandings mean for policy implementation in the EU. The project is grounded in social psychological and sociological theories, using social representation theory to link cultural processes and cultural change to individual social identity formation. It further argues that social representations (cultural understandings) matter for how policy is understood and engaged with by constituents, which means that understanding these processes is important for crafting policy. To explore these questions, the project utilises a series of mixed methods, and here we focus on a novel adaptation of multiple correspondence analysis to analyse cultural change over time in mass media and then link cultural change to local patterns of individual symbolic universe formations and understandings of culture. One of the key parts of this study is the longitudinal mapping of semantic fields of a number of highly relevant societal issues (e.g., immigration, health and wellbeing, racism) during the economic crisis and relate them to geographical areas in Europe. We approach cultural dynamics in terms of the evolution of semantic relationships in a representational space that is identified by means of latent semantic analysis of local newspapers across the EU. We then combine this analysis with survey data to obtain a diachronic and synchronic understanding of the social representations of these processes. The presentation will problematise this linkage, and explore whether quantitative methods prove sufficient for mapping cultural processes and social identity formation. This approach will be discussed also in the context of future development and use of other sources (e.g. online data).

365

Respi, C. and Sala, E. (Dipartimento di Sociologia e Ricerca Sociale, Università degli Studi di Milano-Bicocca) What is the impact of using personalized salutations on survey participation and data quality? Results from an experiment on Italian graduates. Till recently, standardization was considered the norm in surveys. According to standardization rules, respondents are to be contacted using standardized procedures, interviewed by reading out questionnaire question as scripted and using standardized survey “tools”. However, there is evidence that suggests that, in some contexts, this approach may not be the most effective and alternative approaches to data collection, such as those based on “tailoring” survey instruments to respondent characteristics (i. e., flexible interviewing), may obtain better results. Little is known on the effects of using flexible data collection methods on data quality. In this paper we report findings from an experiment that evaluates the impact of personalized vs. standardized SMS invitations on two aspects of non-sampling error, i. e., non-response and measurement error. Previous research on this topic has found that personalization of e-mail salutations significantly increases response rate when surveying university students (Heerwegh 2005; Joinson, Reips 2007). However, we are not aware of any studies that also looked at the impact of different ways of phrasing salutations on measurement error. This study is meant to contribute to expand the knowledge in this research field. We use experimental data collected in the context of a follow-up of a national web survey on occupational outcomes of Italian graduates in Social Work (Sala, Decataldo and Respi 2015) that was carried out in 2015. To invite sample members to take part in the follow-up, short text messages (SMS) were used (telephone numbers were available from the graduates’ administrative records). Sample members were randomly assigned to two groups: the “standardized” and the “personalized” salutation group. The text of the “standardized” SMS was: “Dear graduate, we are carrying out the second wave of the survey Occupational Outcomes of Social Workers. Take part in the study! Link to the 6 questions [url]”. The text of the “personalized” SMS was different only in the incipit: “Dear [graduate’s name], we are carrying out...”. The invitation was sent on May, 13th 2015.

366

To assess the impact of the different types of salutations on response and measurement errors we consider different indicators of survey participation and data quality. Indicators of the former include response rates and speed of response; indicators of the latter include item non response, the time taken to fill in the questionnaire, and social desirability. To analyse our data, we will adopt both bivariate and multivariate analysis. So far we focused on the first research aim and analyzed the impact of tailoring the salutations of the SMS on survey participation. Results from a preliminary analysis show that there is a difference of 5 percent points in response rates between the two groups; sample members who received the personalized invitation (Dear Jon, for example) are more likely to respond than those sample members who received the standardized invitation (significant at 10% level). In the paper we intend to reflect on the psychological mechanisms that lead these differences.

Keywords: standardization, flexible interviewing, non response error.

367

Rho, Eugenia (UC Irvine) Emergence of Membership, Naturalization, and Boundary Objects in the Personalized Social Networking Site With the advent of mass consumer-generated data, the proliferation of personalization technology has become heavily embedded in various forms of products and services. We see personalization in the form of recommendation systems, advertisements and even wearable devices all pitching themselves with the commercial rhetoric of making our lives easier and better. Moving away from the mass-customization of products to individually personalized goods and services comes with the notion of enriched consumer well-being. The personalization metaphor has historically traversed in close proximity with the relational association between technology and improved life standards, a concept that is hardly new. What is relatively novel, however, is that personalization technology is beginning to emerge in our everyday lives in both noticeable and unnoticeable forms of behavioral tracking classification systems that “learns” and categorizes its users in the climate of Big Data. By bringing the concept of personalization into the terrain of classification theory, I will explore the different forms of relationships, practices, and phenomena that develop in the personalized SNS environment. Classification concepts, such as membership, naturalization, and boundary objects can help unpack the various social and political ramifications of personalization. Through our membership, we become a network of mini self-classifiers that empowers personalized social networking sites, such as Facebook as a larger classification infrastructure. Personalization is the operative engine of this system that effectively directs, shapes, and naturalizes our relationship with the social networking site. By personalized participation, we become our own boundary objects on the platform, lengthening the runway of our curated lives within the networked community. The often invisible symptoms of naturalization that emerge through the practice of classification deserve attention for their substantive and material impact on the way we perceive ourselves and others. Using a personalized social networking site (PSNS) as the site of examination, the paper will aim to bring attention towards the various forms of classification practices that remain largely invisible, yet have real, substantive impact on people.

368

Rivas, C., Tkacz, D. and Cruickshank, D. (University of Southampton, England) Automated ‘real time’ thematic analysis of large volume survey freetext responses: methodological issues in using an intelligent system to capture the human voice Large-scale surveys of patient experience typically include free-text responses. An example is the national Cancer Patient Experience Survey (CPES), which produces over 70,000 freetext comments from patients annually. Such data provide significant insights beyond those produced through survey closed questions yet are under-utilised, due to the resource-heavy processes involved in analysing them. CPES freetext for example is provided in unstructured form to service providers who then structure it in ad hoc and inconsistent ways that take so long that feedback to staff often occurs when the data are out of date. There has been discussion about the utility of these comments in the absence of an effective system to analyse them within a useful timeframe for service improvements. Text mining, a possible solution, which we have tried, still has many limitations. We have therefore developed an artificial intelligence type approach to automated analyses of survey freetext comments, using ‘text engineering’, that addresses the main issues involved in processing freetext Big Data. Our process provides literal themes that are suitable for service improvement and could also be used by researchers to develop more conceptual themes. Automation brings its own issues, and we have worked with the different stakeholders - patients, commissioners, managers and clinicians – to ensure that our analyses remain sensitive to the patient voice and retain a human ‘touch’. This paper will discuss the ways we have attempted to bridge the human-automaton divide to move beyond a Big Data-Small Data polarisation and develop a system for the future that can be transferred beyond the patient experience survey, using CPES as our first case. Our approach can encompass divergent epistemologies and is designed to cope with the rapid pace of change in textual content.

PRESENT is funded by the National Institute for Health Research Delivery Research Programme (project number 14/156/15). The views and opinions expressed herein are those of the authors and do not necessarily reflect those of the HS&DR, NIHR, NHS or the Department of Health.

369

Reed, K. and Ellis, J. (Department of Sociological Studies, University of Sheffield) Creating maps of the mortuary: adopting mobile and sensory methods to examine the role of imaging in foetal and neonatal post-mortem Sociological research has often focused on early life loss - on parents’ experiences of miscarriage and neonatal loss. A significant amount of research in this area has centred on memorialisation practices, examining the role of material artefacts in this process. Despite the recent epistemological and methodological shift across the social sciences towards a focus on mobility, sociologists have yet to draw on spatial methods such as maps in their explorations of reproductive loss. Furthermore, there is a limited amount of research in this area which uses the mortuary as a potential research site. The mortuary - a hidden and taboo space - is a place where bodies and biological matter are dissected and analysed and where various technologies are applied. It incorporates a complex network of relationships and interactions between families and different types of professional practices. It provides an important context therefore with which to explore issues relating to death, bereavement and memorialisation, and to think about how organisational practices shape and mediate parental encounters with the physical reality of their loss. This paper draws on an ethnographically-informed mapping technique used in a study exploring the role of MRI technology in foetal and neonatal post-mortem practice. The paper will explore the processes involved in producing these ‘maps’ and discuss the role they and the ‘mortuary’ play in the research. By creating maps in this study we were able to examine the complex relationship between MRI technology, professionals, families, objects (memory- boxes) and other technologies used within the mortuary context. The paper seeks to illustrate the ways in which these maps provided important background context for in-depth interviews with parents and professionals, as well as providing the research team with a rich source of data themselves. It will conclude by highlighting the value of a taking a mapping approach in research on foetal and neo-natal loss, as well as the importance of using the mortuary as a case site. In doing so it seeks to extend the methodological focus of existing work on reproductive loss - and offers an original contribution to emerging conceptual and methodological debates on mobility and the sensory turn across the social sciences.

370

Keywords: *mortuary; *mapping; *MRI post-mortem

371

Roberts, J. (University of Nottingham)

The ontology of early labour (and the difficulties of talking about it) This paper presents reflections on the process of collecting interview data about fathers' experiences of 'early labour'. Early labour is the first phase of labour, defined in textbooks by regular contractions and cervical dilation of up to 4cm. Women are typically encouraged to stay at home during early labour and only travel to hospital when they are in 'active labour'. Maternity services (and other providers of antenatal education) devote a great deal of attention to educating parents-to-be about the phases of labour and about how to recognise the 'right time' to travel to hospital but 'early' admission remains a problem. Interviewing fathers about their interpretations of early labour was challenging and in this paper I will argue that this was partly due to the ontological tensions between my own interpretivist position and the realist tradition from which the concept of 'early labour' stems. I will reflect on the difficulties of talking about 'early labour' without reifying the concept, my concerns about appearing to 'test' fathers on their knowledge of labour, and the importance of finding ways to explore and value the meanings expectant-parents attribute to their embodied and relational experiences. Keywords: ontology, interviews, early labour

372

Robinson, T. (University of Auckland) How do health researchers in New Zealand use quasi-experimental methods to evaluate new programs? A review of published studies. Evaluating outcomes of health programs is important, but often true experiments cannot be undertaken. Quasi-experimental methods may provide an alternative way of assessing program impact, with good internal validity. We wished to assess whether, and how such methods were used in outcome evaluation in New Zealand. We searched for health service outcome evaluations undertaken in New Zealand between 2005 and 2015. 52 evaluations were identified from academic journal or grey literature resources. These were assessed against a framework that assessed designs used, the amount of control the evaluator achieved over important study constructs, and whether significant threats to internal validity were present and addressed. Our main findings we that very few studies could be called quasi-experimental. Most were purely observational in that the evaluator rarely influenced implementation of the new service in a way that strengthened causal inference, and stronger quasi-experimental methods were infrequently used. Over half (27) were exclusively uncontrolled pre-test post-test studies, and nearly all had significant validity threats, particularly selection (18) and history threats (20) and regression to the mean (20). We believe that when a true experiment cannot be used to evaluate a new service or policy we need to think more carefully about quasi-experimental methods to try to achieve estimates of impact that are valid. Keywords Evaluation, quasi-experimental design, health services

373

Robinson, T. (University of Auckland)

The use of three methods to evaluate a health intervention to reduce readmissions to hospital in older people. An exploration of quasi-experimentation and managing threats to validity. In 2012 a two hospitals in Auckland, New Zealand implemented a service to reduce the frequency of readmission among people over the age of 75. A RCT was not supported by governance but we had the opportunity to apply strong quasi-experimental methods. Patients were selected for the intervention on the basis of a scoring over a threshold using a predictive risk model that estimated risk of readmission. We were therefore able to use three methodologies to analyse the impact of the intervention on 28-day readmission rates and emergency department attendance. These were regression discontinuity analysis, interrupted time series, and difference-in-difference analysis All found no effect for both outcomes. We examine some of our successes and failures in the use of quasi-experimental design and analysis, and how different study designs were more or less susceptible to different internal validity threats. We also show how additional analysis aimed at examining more complex patterns can address some of these threats.

Key words Hospital readmissions, quasi-experimental design, regression discontinuity

374

Roche, M. and Dimbath, O. (University of Augsburg) Sampling an urban district: lessons from Augsburg’s "Universitätsviertel" Theoretical sampling is now considered to be an established method for data selection in qualitative research. It was introduced to fulfill the need for interpretive methodology to have access to a more adequate alternative to representative, random sampling and more robust possibilities for explanation than dimensions selected according to demographic criteria. Its main characteristic is that data selection takes place reciprocally with the process of interpretation. This way of proceeding makes it difficult to plan projects within fixed parameters, since the end of the empirical process is unknown, occurring only after so-called theoretical saturation has been achieved. It is only during the research process that the full range and complexity of the initial research questions are revealed. It is also precisely this aspect that makes theoretical sampling uniquely suitable for developing an interpretive approach to data. Our submission concerns itself with discussing the theoretical sampling strategy of data selection in the context of an urban sociology research project. In due course, we intend to show how the (implicit) decisions of which data to sample at any given moment come about and examine what ‘theoretical’ saturation means in a practical sense. Rather than frame theoretical sampling in a space completely void of empirical data, we would like to realize this opportunity to share our experience of concrete research events, situations, and decisions as a basis for discussing the impact of sampling strategies on the research process. For us, this culminates in a careful inspection of how this kind of gradual data selection functions in a practical context, resulting in an attempt to develop a more comprehensive strategy schema. Specifically, we refer to our recent project that studied an urban district in the city of Augsburg, Germany. With an express interest in following and exploring Grounded Theory research methodology, and in examining the potential of ‘classical’ theoretical sampling strategies, the study focused on generating interviews (inspired by biographical research and narrative interview methods) with the population of this district (Universitätsviertel, University Quarter). The study was carried out in the form of a so-called “teaching-research” project at the University of Augsburg. This context makes it suitable for exploring the often abstract notion of theoretical sampling, since it required the development of explicit guidelines and conventions on how to

375

implement theoretical sampling in practice and afforded room for feedback on their effectiveness in practice within the context of coordinating the group’s encounters with the field. This does not mean that the experiences and results were exclusively positive. They do, however, provide an interesting foundation for reflecting on aspects of qualitative sampling that are often ignored: the role of institutional contexts, the sometimes paradoxical effects of inductive sampling on research design and scope, and the necessity of experiencing failure and ‘dead-ends’ during the research process, not as a weakness, but as a strength of theoretical sampling. In bringing our experiences into alignment with the ongoing discussion on qualitative sampling as a whole, we hope to provide new insights on recurring problems in qualitative research.

Keywords: theoretical sampling, Grounded Theory methodology, qualitative research in practice

376

Rosen, C. (Goethe Universität Frankfurt, Germany) “Accessing the political field of Berlin and Madrid from a spatial perspective” In the following I want to propose a paper dealing with my experiences during ethnographic field research on the national political fields of Berlin and Madrid.

Based on the ideas of Pierre Bourdieu’s construction of social space the political field is a social subsystem consisting of actors following a set of certain rules and using different kinds of capital to fight about positioning within the hierarchic field. My question is: What meaning does space have in these processes?

Throughout different stages of ethnographic field research I collected a big corpus of data of different types addressing questions about the two dimensions of space in social sciences. First, the meaning of space as a given variable structuring field participant’s daily life and work was of my interest. The relevance of distances and proximities, the quality of locations and the relation of time and space were central questions here. Second, the construction of political space through social action inside the field was important to me, too. Here the main question was about what happens when field members come together on different occasions at different places. What are the differences between official political places like parliamentarians offices and a bar or restaurant? Is it just a more informal setting and the same talk or is there a change in rules of making politics happen when actors are exchanging the spatial setting or even creating their own?

To find out about the view of field participants concerning the relevance of space I conducted 45 guideline based interviews, had several informal talks, did participant observation and collected documents in Berlin and Madrid.

Besides my findings, pointing in the direction that space, as a social construct and at the same time as a frame for social action, is an important factor for field participants to act successfully within the field, my work also tells about challenges concerning field research and the accessibility of space as a research dimension. Therefore I would like to address two questions in my paper:

377

1. How can we handle access problems and matters of confidentiality when we are confronted with an exclusive field like a state’s national political field?

2. How can we ask about the relevance of space and how can ethnography as a method help us find out about it?

Both questions are addressing mayor issues during my field research and are still a topic of high priority now analyzing the material and formulating results. While the first question addresses the sampling strategies in my specific cases the second question reflects on the general choice of ethnography as a method but also as a research paradigm for spatial analysis.

378

Rüttenauer, T., and Best, H. (TU Kaiserslautern, Department for Social Sciences) The Causes of Environmental Inequality: Evidence from Spatial Time-Series Analysis in Germany The analysis of environmental quality and socioeconomic composition on the level of spatial units has become a standard approach of empirical research on environmental inequality. The majority of prior studies relies on cross-sectional spatial data and hence cannot adequately study the causal mechanisms leading to the unequal distribution of environmental. Thus, it remains a puzzle whether selective move-in, selective move-out, selective siting by the industry or a combination of these factors lead to an uneven distribution of environmental hazards across different socio-economic groups. In addition, most of the research has been conducted in the US and empirical results from continental Europe and especially Germany are rare. In this paper, we study the spatial distribution of air pollution emissions and the connection to socio-economic factors in Germany between 2001 and 2012. The analysis relies on a combination of data from the European Pollutant Release and Transfer Register (EPRTR) and socio-economic data on the level of 4,521 German municipalities. The amount of pollution from the EPRTR facilities is matched to the communities within a 5 km radius of the facility location proportionate to its spatial overlap. To investigate the causal mechanisms producing the unequal distribution of environmental pollution, we use a fixed-effects time-series approach with a correction for spatial autocorrelation (FE-SAR). The results identify selective siting and selective move-in as a cause of the unequal distribution of environmental harm, while contradicting the selective move-out explanation. However, we need to be aware of potential problems due to the use of spatially aggregated data. Keywords: Spatial Analysis; Environmental Inequality; Spatial Fixed-Effects Model

379

Roberts, C. (University of Lausanne) Do mode effects matter in cross-national surveys? An assessment of measurement invariance across data collection methods and countries Mixed mode data collection is becoming increasingly popular in survey research in response to pressure to reduce fieldwork budgets. Mode choice in survey design can influence who is able to participate in a survey (coverage); who chooses to participate (nonresponse); and how respondents answer questions (measurement), meaning that the structure of errors affecting an estimate varies as a function of how the data were collected. The use of a combination of modes can, therefore, potentially hinder analysts’ interpretation of differences between subgroups of interest, if selection effects are confounded with measurement effects. In multi-nation studies, the use of different modes in different countries is similar to the use of multiple modes in a single-nation study in terms of its confounding effect on the comparability of estimates across subpopulations. Where multiple modes are envisaged both within and between countries, the question of how to assess measurement equivalence and manage the additional data complexity poses an important challenge for the field of comparative survey methodology.

The literature on mixed mode data collection has predominantly been focused on whether differences can be observed between modes of data collection, yet a more relevant question for data users concerns how important differences between modes are in terms of their potential impact on point estimates, and the relationships between variables of interest. This question of whether mode effects ‘matter’ has remained largely unanswered, partly because of the technical complexity of disentangling measurement differences from selection effects, which makes simply detecting mode effects problematic to begin with. One solution appears to lie in a total survey error approach, however, which attempts to evaluate the relative amount of error from different sources in survey estimates. In a comparative context, for example, it makes sense to ask whether the impact of differences between modes is greater or smaller than the impact of other methodological differences between countries that may affect measurement equivalence.

380

In this paper, we attempt to address this challenge by presenting the results of an analysis of measurement invariance across countries and across modes of data collection using Confirmatory Factor Analysis. Using data collected as part of a methodological experiment conducted during round 3 of the European Social Survey, we will assess the implications of mode choice (face-to-face versus telephone surveys) on measurement equivalence of multi-item measures of subjective wellbeing in four participating countries (Germany, Hungary, Poland, Switzerland).

381

382

Rhodes, B. (RTI International) Monitoring data collection from afar: Methods and lessons learned from data collection in Pakistan Collecting data in foreign countries can pose many challenges for researchers. Using an in- country data collection organization, with local knowledge and ready staff availability, is one method for overcoming some of these challenges. This method, however, comes with its own challenges. Primary among these is that the researcher may not have direct control over the data collection methods being employed on the ground. Methods such as a prior review of the in-country data collection procedures, quality control measures, and field materials can help ensure a sound data collection methodology. Experience has shown that these procedures are not always implemented as the researcher anticipates. This paper draws on experience from a recent project to evaluate a teacher education program in Pakistan. Two in-country organizations were responsible for all aspects of in-country data collection (staff recruitment and training, contacting schools, in-person data collection, and data delivery). An international firm was responsible for review of the data collection instruments and procedures, as well as, review of the collected data and ultimately data analysis and reporting. Survey data were collected from teachers and administrators and students were given an achievement test. All data were collected using paper and pencil forms that were later enter into a central database. This paper describes the project data collection procedures, quality control methods implemented, and data review techniques. The paper describes the successes and challenges of working with the in-country data collection firm from the perspective of the international research organization. It explores specifically issues related to how procedures were misinterpreted or not appropriately implemented during the actual data collection, how these issues were detected, and the methods used to remedy them. In addition, the division of labor between the various organizations meant that for the international organization the data itself was often the best indicator of the data collection procedures being implemented on the ground. The paper also describes the data monitoring methods used to identify procedural errors from afar.

383

keywords: field data collection, Pakistan, quality control

384

Sakshaug, J. (University of Manchester, England)

Quantifying the Prevalence of Analytic Errors in Secondary Analyses of Complex Sample Survey Data Investigators from many scientific fields commonly perform secondary analyses of survey data collected from large probability samples. These probability samples are often drawn using complex sampling designs involving stratification, clustered units, and unequal probabilities of selection. All of these design features can affect the validity of any inferences that arise from survey data if they are not appropriately account for at the analysis stage. Unfortunately, not all data users possess the relevant training and/or software tools needed to properly account for these features when analyzing survey data. This can diminish the quality of any population-based inferences that are made and could produce misleading conclusions from the research. We refer to these problems as “analytic errors” in the sense that users apply incorrect survey analytic techniques when performing secondary analyses of complex sample survey data. There is virtually no data on how prevalent this problem is, particularly in the published literature where secondary analysis of survey data is common. To address this knowledge gap, we carried out a meta-analysis to quantify the frequency with which researchers appear to use incorrect analytic techniques when publishing secondary analyses of survey data arising from complex samples. In addition, we examined characteristics of the journals in which these articles were published (e.g., impact factor, presence of statisticians on the editorial boards, analytic guidelines for authors, etc.) to see whether they were related to the prevalence of various errors. The results of the meta-analysis suggest that several types of apparent analytic errors are in fact quite prevalent, including inappropriate subpopulation analyses and a failure to use appropriate software for survey data analysis. Analysts also frequently failed to incorporate weights or compute standard errors reflecting sample design features. Furthermore, we find that descriptions of analysis results and inferences may tend to mislead readers about the scope of the inferences (i.e.,

385

population vs. sample). We also find that most peer-reviewed journals, including those with large impact factors, fail to emphasize the use of specialized survey procedures for secondary analysis of complex sample survey data in their guidelines for authors. These findings suggest that academic journals and survey organizations could do more to emphasize the types of analysis methods that would be appropriate for a given complex sample survey data set.

386

Sakshaug, J., Schmucker, A., Kreuter, F., Couper, M., and Singer, E. Evaluating Active versus Passive Consent Procedures to the Transfer of Federal Contact Data to a Third-Party Data Collection Agency Quality issues can occur in the entire process of collecting survey data. Often the first obstacle to break through is getting valid contact information (e.g. addresses, telephone numbers, email addresses). Many public authorities record extensive contact data suitable for sample selection and recruitment but privacy laws generally prohibit them to provide this information to researchers. Sometimes there are some legal exceptions and the data can be transmitted if some legal restrictions are met.

In Germany, for example, the Federal Employment Agency stores address data of all employees liable to the social security system, all recipients of welfare benefit, registered job-seekers, and participants in active labor market programs. Transfers of this contact information to third-party institutions for purposes of conducting research projects are generally exempt from consent requirements. But concerns recently expressed by relevant data protection authorities prompted the evaluation of the impact of requiring consent on survey participation rates, selection error, and differences in the resulting survey estimates (relative to a control group).

Therefore we conducted an experiment by applying two different procedures (active and passive consent) to obtain consent from subjects for transferring federal contact data for prospective survey participants to a data collection agency for purposes of conducting a subsequent telephone survey. A control sample reflecting the status quo (no consent required) was also utilized for comparison purposes. Many researchers argue that active consent is the only type of consent that accurately reflects the true wishes of the subject, and is closer to the informed consent ideal than passive consent procedures. Proponents of passive consent argue that active consent procedures hurt study participation rates and increase the risk of selection bias in the study estimates to a greater extent than passive consent procedures. Empirical evaluations of these claims are rare given the lack of studies that experimentally assign subjects to different consent procedures and utilize a control group (requiring no consent) to facilitate comparison.

387

Our findings show that the passive consent procedure does a better job of minimizing self- selection error and maximizing the validity of the survey estimates (relative to the control group) than the active consent procedure. However, neither procedure is ideal: both consent procedures increase the total selection error and reduce the sample size – the passive consent procedure just to a lesser extent.

Key words: self-selection bias, nonresponse, opt-in/out consent, data privacy, survey error

388

Sand, M., and Kolb, J-P. (GESIS – Leibniz Institute for the Social Sciences, Germany) Using Spatial Information for CATI-Samples In contrast to other countries, the structure of Germany’s landline telephone numbers does not allow a typical Random Digit Dialing (RDD) - approach when sampling CATI-Surveys. Since the length of a German landline-prefix varies between three and six digits, and the total length of a telephone number usually lies between 9 and 13 digits, sampling via classical RDD is not feasible due to the range of numbers that it would create and its accompanying low hitrate. Therefore, a conventional approach to construct a landline sampling-frame usually employs the method of Gabler and Häder (1997), which allows the sampling of telephone numbers with equal probabilities based on unique banks of 100 subsequent numbers that are derived from numbers that are listed in a telephone book.

A further problem, especially when conducting regional CATI-Surveys in Germany, is that the prefix area is often not congruent with an administrative area, which might either lead to erroneous estimates or to a higher burden for those conducting a survey, since certain areas must be screened out to avoid such biased results. Regarding that problem, this paper addresses such obstacles by mapping the CATI-samples of specific surveys we have been counselling on, using the open source data of e.g. Open Street Map (OSM) and comparing these areas to the data of administrative areas.

Furthermore, this paper will investigate the potentials of combining spatial data and landline surveys. Hence, two distinct questions are of interest. The first is whether spatial information can be used to identify specific patterns in landline numbers that allow us to link specific numbers of a distinct prefix area to a singular administrative area. The other question more generally focuses on how to use spatial information to identify areas of interest and therefore prefix areas that should be included in a CATI-sample regarding distinct research questions.

389

Sanduk, M. (University of Surrey, England) Can the society be simulated as plasma fluid? Picture of Holism and Individualism in social science is quiet similar to that of fluid physics (of particle and fluid models). Connection between holism and individualism is improved via computer simulation techniques, and this modern technique is quite common in modern computational physics. Many sociology researches adopt the concept of mechanical fluid in computational social simulation, especially in crowd phenomenon.

The fluid is a continuous medium; this characteristic feature is found in sociology as well in ‘social fluid’ (Muller and Pollak , 2001) or the unstable social fluid (Kaplan,1994). Not only is the terminology of fluid studies used in modern sociology, but many other relevant terms of fluid dynamics like diffusion, mobility, instability, bunch…etc have also been adopted. The usage of these terms of fluid dynamics in sociology does not mean that the theoretical logic of fluid dynamic controls that medium (society). These terminologies are scattered arbitrarily in different research articles without mathematical unification (logic) as that of fluid dynamics. The concept of physical fluid may not fit the simulation of society very well, or it has a quite limited validity and may need modifications.

In modern physics, the forth state of matter or the plasma may have rather similar behaviour of the society. The plasma scientist D. Bohm (1917-1992), saw in plasma a model of society (Wilkins,1986), but he did not go into the details. Plasma is the fourth state of matter, and may be defined as an ionized gas composed of charged and neutral particles that exhibit a quiseneutrality and collective behaviour. Macroscopically, plasma looks like a sort of fluid called MagnetoHydroDynamic (MHD) fluid, whereas microscopically, plasma is considered a grope of free charged and neutral particles and obeys the statistical kinetic theory.

Plasma fluid has other properties than those of ordinary fluid like collective behaviour, shielding, high response for external effects of fields like electric and magnetic fields, and diffusion through a confining field. These properties arise owing to the charged structure of the plasma, and the rather free units (free charged particles).

390

Both society and plasma fluid are composed of active, interactive, and free, individuals. These individuals respond to any internal and external effects, and exhibit collective behaviour. According to this structure, there are a wide range of similarities between plasma fluid and the society. The nature of fluidity of plasma arises from the interaction of its free interactive charges; therefore, the society may behave like a fluid owing to its free interactive individuals. This fluid model may explain many social phenomena like social instability, social diffusion, social flow, social viscosity, and hence the society behaves as a sort of intellectual fluid. The equation of fluid dynamics is a modified equation of motion of a single particle. For the purpose of computer simulation technique, the random walk of a number of steps is intro-duced in addition to the equation of motion of a single particle. To adopt the same method of building up the fluid equation for society, one may start with the considering the force in case of social individual. The acting force on the person is a summation of many types of forces. They are probably like: force of instinct, force of ambition, life needs force, social forces, be-lieving force, media forces, etc. These are confronted with problems of quantisation.

References Kaplan, R. (1994). The Coming Anarchy. Atlantic monthly, 273 (2). Muller, W. and Pollak, R. (2001). Social Mobility in West Germany The Long Arms of Histo-ry Discovered?, Meeting of ISA research committee 28, Berkeley. Wilkins, M., (1986). Oral history interviews with David Bohm, 16 tapes, undated transcript (AIP and Birkbeck college Library, London), 253-254.

391

Santos, F. (Freie Universität, Germany) Spatial Analysis at the Fringes of Europe: Field Notes from French Guiana When social scientists study the European Union and its borders, they rarely consider the EU’s overseas territories that lay scattered around the world. It was only recently that scholars have begun to stress the need of researching these blind spots of Europe’s spatial, post-colonial configuration. By highlighting the case of French Guiana – an integral part of today’s European Union on the Latin American continent –, this paper aims to readjust mainstream sociology’s map and to address methodological questions of how researchers can adequately study these forgotten fringes of Europe. The French- Brazilian border – a historically contested, yet literally fluid border – is of particular interest. Based on first empirical explorations and findings from this region, the article argues that ethnographic fieldwork offers invaluable insights especially for microsociological approaches drawing on a relational and processual conceptualization of space. Thus, the non-essentialist study of spatial changes experienced and triggered by local populations (and migrants) in EU overseas territories and border regions requires an open and dialogic research perspective. However, the article also mentions pitfalls of ethnographic methods, particularly with regard to the challenge of grasping the French-Brazilian border’s cultural diversity.

392

Santorso, S. (University of Padua) Researching Social Process in Prison: Consideration on Mixed-method and Analytical Dualism This paper is the result of my researches on the prison community and social change, conducted between 2011 and 2013. The research aims to study the way in which prison community and prisoners’ social organization changes in the age of mass incarceration. The paper takes into consideration some issues on the epistemological and methodological approach arisen in the analysis of the complexity of the social process within the prison community. Therefore, moving for some considerations on the empirical research, the paper wants to underline the relevance of a mixed method data collection and a multilevel analysis (Maruna S., 2009) for the criminological research. In fact, in opting for a mixed method, I use a sequential research design, able to “add qualitative profundity to quantitative findings” (Huber 2007: 179). In fact, I decided to articulate the research design using three different methodologies, in the following order: time series analysis – biographical interviews – ethnographic observation. As premises is necessary to underline as the detention experience is a peculiar interaction between time and space (Goffman E., 1968), and for its nature totalitarian nature also between micro and macro level. On this regard, the paper focuses its attention on the “qualitative profundity” of the research method, discussing the study of social process inside the prisoners’ communities. The paper focuses its attention on two issues: the dichotomy agency-structure and the dichotomy time-space. The paper debates how the use of different methodological tools generates a dialogue between agency and structure, as dialectic implicated and mutually constructive of each other. This interaction potentially creates a continuous exchange that does not lend itself to a rigid categorisation, rather to a flow of data analysis and redefinition of the categories produced by their interplay. Therefore, the paper discusses the use of a methodological trick to overcome this situation: the analytical dualism (M. Archer, 2003). This, applied to the study of prison communities, allows to separate the data and to keep them constant, facilitating the analysis. In addition, the presentation argues that this approach, especially for its focus on

393

transformation, needs to grasp the time space relations inherent in the constitution of all social interaction (Giddens A., 1979). Thus, the paper discusses the use of a diachronic level of comparison (past-present) to understand the social process and their conservative/transforming impact. Even in this case, thanks to the use of an analytic dualism as technique the dialectic comes to a standstill, facilitating the analysis. In the conclusion the paper present some consideration on the relevance of a multilevel analysis to study the social process. Reference: Archer M., Structure, Agency and the Internal Conversation, Cambridge Press, 2003 Maruna S., Mixed method Research in Criminology: why not go both ways?, Springer, 2010 Giddens A., Central Problems in Social Theory, University of California Press, 1979 Goffman E., Asylum, NYP, 1968

394

Saramago, A. (Department of International Politics, Aberystwyth University) Social Scientific Orientation in the Next Phase of Humankind’s Global Development Informed by Norbert Elias’s process sociology this paper discusses how the social sciences might provide more adequate ‘means of orientation’, here understood as theoretical frameworks on the basis of which human beings might come to better understand their conditions of existence and how to exercise a greater degree of conscious and collective control over them. It makes the argument that the present levels of global interdependence of the species, combined with the global impact of humankind on the planet’s biosphere, increasingly demand a widening of social scientific perspectives to encompass very long-term processes of change and humanity-divided- into-survival-units as their main units of analysis. In this context, the paper shows how a process sociological approach can help social scientists navigate through this ‘return to the long-term’ in a manner that avoids the teleological and evolutionist character of earlier grand narratives on human development and that aids survival – both human and that of other earthly species – in the next phase of humankind’s global development.

395

Sarracino, F., and Mikucka, M. (Statec (Luxembourg), Université catholique de Louvain (Belgium), Laboratory for Comparative Social Research, HSE (Russia) Estimation bias due to duplicated observations: a Monte Carlo simulation This paper assesses how duplicate records affect the results from regression analysis of survey data, and it compares the effectiveness of five solutions to minimize the risk of obtaining biased estimates. Results show that duplicate records create considerable risk of obtaining biased estimates. The chance to obtain unbiased estimates in presence of a single sextuplet of identical observations is 41.6%. If the dataset contains about 10% of duplicated observations, then the probability of obtaining unbiased estimates reduces to nearly 11%. Weighting the duplicate cases by the inversion of their multiplicity minimizes the bias when multiple doublets are present in the data. Our results demonstrate the risks of using data in presence of non-unique observations and call for further research on strategies to analyze affected data. Keywords: duplicated observations, estimation bias, Monte Carlo simulation

396

Savinskaya, O. (National Research University Higher School of Economics Moscow, Russia)

1 Exploratory sequential design: toward the inductive and deductive transfer of knowledge Problem statement. “Exploratory sequential design” (Creswell, Plano Clark 2011, p.71), or “QualQuant sequential contribution” (Morgan 2014), or “instrument development design” (Creswell, Fetters, & Ivankova, 2004) suppose a sequential timing. In brief, it builds on a first phase of the collection and analysis of qualitative data and a second, quantitative, phase which focused on testing or generalization of the initial findings. The first one is ascending from everyday language and knowledges to theoretical terms and assumptions about their relationships. And the second is vice versa: the measurement tools in colloquial language are elaborated from theoretical conceptions about subject under the study. It can take research iterations to connect the two levels as a part of a “substantive theory” (Glaser, Strauss 1968). These two processes allow transferability (Lincoln & Guba, 1985) as the criterion of research quality. Mixed methods design here is stitching together two types of social reality representation – unique experience and generalized practices. The main goal of the design is “to demonstrate that the core concepts uncovered in the qualitative research apply more generally, beyond the original collection of specific cases” (Morgan 2015, 2). The quality of the exploratory sequential design could be related to the reflection of how knowledge transfer at every step is moving «up» the and then “down”. To achieve the consistency and convergence between data of different levels/types, I developed the model of “knowledge transfer” in method sequences.

Research design. To show how that model could be used, I turn to my research on parental qualitative evaluation of preschool education providers conducted in 2014-2015. The sequence of qualitative and quantitative methods was used QUALQUANT in Morse’s (1991) notations. Then I argue how knowledge accumulated transfers “up”(inductive) and “down” (deductive) during the research. I illustrate it with categories/themes of qualitative analysis, identified in the first, qualitative, stage.

397

Phase 1: The qualitative phase of the research included 30 individual in-depth interviews with mothers of 3- to 6-year-old children attending different public municipal kindergartens. The stage of qualitative analysis is requires sequential coding and saturating the initial codes with unique personal meanings that during the constant comparative analysis (Glaser, Strauss 1968) are becoming more general and obtain the status of category. At the transitional stage the categories are transformed into the statements for summative rating scale.

Phase 2: The on-line survey with 340 parents was taken during the quantitative phase. Three types of scales that were elaborated during the first phase have been tested for statistical consistency by Cronbach’s Alpha and by split-half by Spearman-Brawn. The interpretation of scale values required qualitative data as in the beginning. The strategy of descending or “grounding” the qualitative data was taken in the elaboration of generalized statements about routine practices in a kindergarten for the summative rating scale.

Dicsussion. The analysis of the qualitative stage responding to the need of scale elaboration is more rigorous, which helps to develop focused interpretation. The quantitative research is more responsive to the unique research problem through close contact with qualitative data and analysis.

Keywords: exploratory sequential design, knowledge transfer, preschool education

1 The article was prepared within the framework of the Academic Fund Program at the National Research University Higher School of Economics (HSE) in 2015 (grant № 15-05-0012). The project “Mixed methods research: needs, design and procedure implementation”.

398

Sauerborn, E. (Freie Universität Berlin) Understanding Emotions in Hermeneutic Processes Understanding emotions in hermeneutic processes often diminishes a specific consideration of the research topic. Due to a lack of distinction between social meaning and social interpretations that underlie emotions, the scientific understanding of emotions often relies on vague methodological assumptions. In order to present a more specific methodological reflection of the examination of emotions, I want to firstly outline in what way emotions constitute a different object of investigation than social meaning does. My focus here is on hermeneutic sociology of knowledge. Secondly, I want to stress and analyse the chief role of the researcher’s affective socialization for the understanding of emotions in qualitative data.

In order to analyse the hermeneutic processes that are needed to understand emotions in a scientific way, qualitative researchers have to clarify their research topic. To do this, it is crucial to first distinguish between everyday interpretations of emotions and those that are followed by rules. Second, we need to analyse methodological assumptions about the differences between reconstructing meaning and reconstructing emotions on the one hand and how meaning is concerned in hermeneutical analysis of emotions on the other.

Generally, the aim of scientific hermeneutic analysis is the reconstruction of meaning, “inner conditions” of individuals and functions of social interaction’s conditions. Even though emotions apparently rely on meaning, represent inner conditions and lead to interactions, it would simplify qualitative interpretations if we would use the same tools for emotion research as we use for the reconstruction of meaning without methodological reflection. To begin with, scientific research processes for the understanding of emotions influence the research question. Considering constructivist and symbolic interactionist approaches on emotions, researchers would not address emotions at first glance but rather through narratives, social interpretative frames and meanings that signify the foundation of understanding. In this respect, I propose that the reconstruction of meaning is one indispensible criterion for the research aim but not to be compared with the reconstruction of emotions per se.

399

Moreover, subjective views of the researchers are by definition meaningful in hermeneutic processes. I argue that the researcher’s affective socialisation is even more important in emotion research than it is in other qualitative interpretations. Using nostalgia and lovesickness as emotions that are learned later in life as an example, I want to show how the perception and memories of individual emotions have a huge impact on the scientific interpretation and recognition of emotions in qualitative data. By doing this I will show how misleading empathy proves to be since hermeneutic interpretations request skills to comprehend but not necessarily to feel the emotions of others. This leads finally to the assumption that different readings of the data whilst necessary are often not sufficient to identify emotions if the narrative backgrounds are missing.

400

Sawert, T. (University of Potsdam) A ‘Dead Language’ as Profitable Investment? A Quasi-experimental Study on the Influence of Learning Classical Languages at School at Point of Career Entry Even after controlling for central influencing factors of educational achievements, there still is an effect of the social background on labor market chances. To explain this persistent effect, research on the influence of horizontal educational inequalities has risen in Germany. One field that has not attained any attention until now is the impact of the foreign languages learned at school on chances in the labor market.

On their way to attain the highest secondary degree in Germany (Abitur), students must decide for at least two foreign languages. Since globalization is proceeding, foreign language skills become increasingly important for chances on the labor market. In this context, it is a surprising phenomenon that about one third of all students attaining the German Gymnasium choose Latin as foreign language. As previous research shows that the choice of Latin is dependent on having parents with a high educational background, the assumption arises, that the effect of social background on labor market chances is in part mediated by the chosen language profile. This assumption has not been tested yet, nor have data been collected which allow for a testing.

Based on different theoretical approaches, hypotheses are derived about the way in which the language profile at school affects the decision making processes of human resource managers, thereby influencing the labor market chances. To test these assumptions and to measure the direct causal effect of different language profiles, a quasi-experiment has been conducted in which applications have been sent to 130 job advertisements for executive positions. For this purpose, applications were generated that differed only in the extent of Latin classes chosen at school (no Latin, Latin as a first vs. Latin as a second language). As the dependent variable, the proportion of invitations to job interviews was measured and compared over the different application conditions. Allowing for a deeper analysis of the mechanisms behind the influence of the language profile, further information about the application process was collected and a short survey has

401

been conducted, asking the people in charge for the advertisements about their own language knowledge.

The results clearly indicate that there is a strong positive effect of a strong classical profile on labor market chances, which is dependent on the language profiles of the human resource managers.

The research conducted clearly shows the potential of quasi-experimental designs in analyzing causal influences, since high rates of internal and external validity can be assumed. This is the case because the quasi-experiment took place in the natural setting of an application procedure where high levels of attention of the respondents can be expected. It can be assumed that there is no relevant selection bias for the group under investigation, because answers to 98% of the applications where received. Furthermore, the results demonstrate the potential of combining survey data and data from quasi-experimental studies, allowing for a more detailed analysis.

Keywords: Quasi-Experimental Design, Labor market inequalities, Foreign language skills

402

Schmucker, A. (Institute for Employment research (IAB) Quality, analytic potential and accessibility of linked administrative, survey and publicly available data Longitudinal data continuously gain in importance especially for life course studies. However, surveys increasingly face the problem of unit-nonresponse due to increasing data protection concerns and panel attrition or declining reachability and cooperation of respondents. Quality issues arise with item- nonresponse or misreporting, especially when recall error in retrospective interviews occurs. Particularly longitudinal interviews lead to high costs and response burden (Groves 2011).

At the same time more and more process-produced data sources like big data or administrative data emerge and can be examined regarding their value for research (Kreuter/Peng 2014). One potential remedy for quality and costs issues thus is the linkage of survey data with administrative or publicly available data. The purpose of their collection may initially have been other than creating research data but they usually offer precise and reliable information covering long periods of time. Data linkage thus potentially results in higher cost efficiency and data quality. Linked data also provide higher analytic potential for substantive analyses than their separate parts, either by combining their sets of variables, by adding observational levels (e.g. employees within establishments within companies) or following respondents after their panel drop out. Moreover, research on the quality of either data source gets possible by applying validation, unit- or item-nonresponse analyses or by examining the selectivity of consent to and success of record linkage.

However, when conducting data linkages, we also face challenges: first, we need to have an appropriate matching key for the data linkage. Often we have to deal with error-prone and non- unique matching variables within record linkage procedures (e.g. names, addresses, birth dates). Ridder and Moffitt (2007) describe potential bias induced by imperfect linkage. Second, we have to consider legal restrictions for data linkage and data access. This can potentially lead to lengthy process of getting permission from ethics boards or data protection commissioners.

403

Our presentation will focus on the potential, quality and accessibility of linked data of the Research Data Centre of the German Federal Employment Agency. On the one hand they comprise administrative data on a daily basis with exact information on the income and receipt of benefits since 1975. On the other hand these data can be linked to several survey data sets on individuals, households or establishments which focus on fields of research like poverty, education, intergenerational mobility or lifelong learning.

References: Groves, R. M. (2011). Three Eras of Survey Research. Public Opinion Quarterly, 75(5), 861-871. Kreuter, F.; Peng, R. D. (2014): Extracting Information from Big Data: Issues of Measurement, Inference and Linkage. In: Lane, J.; Stodden, V.; Bender, S.; Nissenbaum, H. (Eds.), Privacy, Big Data, and the Public Good. Frameworks for Engagement, pp. 257-275. Cambridge: Cambridge University Press. Ridder, G.; Moffitt, R. (2007): The Econometrics of Data Combination. In: Heckman, J. J.; Leamer, E. E. (Eds.), Handbook of Econometrics, Vol. 6, Part B, pp. 5469-5547. Amsterdam: Elsevier.

404

Schnable, S. A. (University of Florida) Pregnant with Possibility: The Importance of Visual Data in (Re)Presenting Queer Experiences of Reproduction Lesbian women who are navigating reproduction are building their own visual archive comprising virtual pregnancy journals, photographic blog posts, and image-driven conversations. Although these women are increasingly using visuals to document and share their experiences as a form of resistance and community-building, sociological research has yet to attend to lesbian women’s reproductive choices and obstacles by basing methodologies on visual data. The lack of such research that visually collaborates with the queer pregnant subject presents a disjunct between what many members of the LGBTQ+ community see as important and what researchers are choosing to study. Lesbian women are exercising agency by documenting their reproductive experiences visually and thus proclaiming this as a key site of subjecthood, so why have sociologists failed to engage with this topic as a fruitful place for analysis? This presentation considers two forms of data being constructed for the researcher’s doctoral dissertation in order to argue that the visual archive of lesbian reproduction persists as a crucial site of recovering agency and documenting queer histories. It follows Rose’s (2012) three core tenets of a critical visual methodology by taking images seriously for both their contexts and their own effects; considering the sociohistorical conditions and consequences of visual representation, especially practices of inclusion and exclusion; and interrogating the researcher’s own way of looking as it is grounded in social location and purpose. Visual data on this topic disrupts traditional ways of knowing by centering the queer body as a valid form of knowledge and experience, thus contesting the habitual and limited disciplinary ways of understanding the social world. Further, it challenges “controlling images” (Collins 2000) imposed on pregnant bodies by the medical model as well as those imposed on nonheterosexual women by larger society by empowering women to represent their own unique reproductive experiences. The collective quality of reproductive pathways indicates a deep need for queer preservation, as women are documenting their experiences not only for themselves but also for the purposes of forming relationships and interactions around the uniqueness of nonheterosexual reproduction.

405

Their negotiations with, conflict over, abandonment of, and resistance to existing scripts are (inter)actions most fully explored within the visual realm, where queer bodies can be un- silenced, un-erased, and un-ashamedly reproductive. By Bourdieu’s (1990) account, that which is not photographed is that which is not photographable; thus, by denying visibility to lesbian mothers-to-be in social science research through erasure, marginalization, or forced assimilation, these women become unrepresentable. In producing a visual archive in collaboration with researchers, lesbian women can recoup the subjecthood that they have systematically been denied in this arena. Any social science of reproduction – inherently a topic centered on the needs of the female-bodied in society – should be attuned to these dynamics of power and exclusion. The use of visual data holds significant potential for the (re)presentation of bodies and experiences that have been largely ignored in studies of reproduction, especially where such research is concerned with co-constructing images of unphotographable subjects like queer women on their own terms. Keywords: queerness; reproduction; visual sociology

406

Schneider, S. (GESIS - Leibniz Institute for the Social Sciences) The shortcomings of long list showcards: measuring education with an adaptive open question and semi-automatic coding Education is one of the most frequently used variables in social science research. Research on the reliable and valid measurement of an individuals’ educational attainment is accordingly of vital importance for high quality research using survey data. Traditionally, educational attainment is measured using one or a few closed questions employing showcards listing a country’s educational attainment categories. The use of showcards facilitates the retrieval of autobiographical memory, and results in data that can be readily analysed. However, the use of showcards for measuring educational attainment has several disadvantages: Firstly, due to rising differentiation of educational systems, showcards used for education questions either become more abstract or get very long, both resulting in increasing item difficulty for respondents. Secondly, a high level of education is today a highly desirable characteristic, and the use of showcards facilitates socially desirable responding. Thirdly, long showcards in which individuals still cannot find their often highly specific educational qualification (which they are often eager to report) are demotivating for respondents. Finally, migration results in the presence of individuals with foreign educational qualifications in survey samples. They are usually asked to ad hoc assign themselves to ‘equivalent’ national education categories, rather than providing more culturally sensitive adapted measurement instruments. Thus, measurement problems in this commonly used variable should be the rule rather than the exception. Also, the measurement of education in cross-national surveys continues to be difficult, mostly because different studies inconsistently harmonize educational attainment data. This paper asks whether it is possible to replace closed questions with showcards by open questions integrating an international database of education in computer-assisted surveys, to overcome the problems outlined above. To respond to this question, we analyse data from 3 studies: Firstly, the GESIS Panel Pilot, a German online panel with about 500 respondents for which data were collected in late 2013 employing a split-ballot design with four different open educational attainment questions. Secondly, the SOEP Migrant Sample in summer 2015 and

407

thirdly the Dutch LISS panel in January 2016, the latter also including a split-ballot experiment, fielded an open educational attainment question with integrated database search, complemented by a closed question as fall-back option. For all three studies, we have independent education measures from earlier panel waves using ‘established’ questionnaire items. This provides the opportunity to compare respondents’ educational attainment resulting from the open and closed questions, addressing the following questions: Are different types of questions (closed or open) associated with different amounts of item- nonresponse? Are different types of questions (closed or open) associated with shorter or longer response times? How do the substantive results of the open questions compare with those of closed questions? Taking furthermore the results from two cognitive interview studies (one in Germany, one in Venezuela) into account, we will also try to find out which question format produces the most valid data.

408

Schnell, R. (City University London)

Putting people on the map without revealing their location The use of geo-referenced data in the social sciences is seriously hampered by privacy considerations. Nearly no information below clusters of about 20 households is available in Europe. In many scientific-use files, no geographical information below large administrative units (NUT-Level 3) is available. The statistical justification of this kind of data protection is simple aggregation. However, especially in medical research settings, a growing literature discusses other methods of privacy protection in spatially referenced data. The presentation will mention a few of the available approaches and will concentrate on a recently developed mathematical technique for the anonymisation of distance matrices (Kroll/Schnell 2016).

In general, research on anonymisation has focused on point locations. The disclosure risk arising from publishing inter-point distances and methods of their anonymisation methods have not been studied systematically. Recently, we suggested a new anonymisation method for the release of geographical distances between records of a microdata file -- for example respondents in a survey. For a data release scheme in which microdata without coordinates and an additional distance matrix between the corresponding rows of the microdata set are released, the newly proposed method offers specific advantages. In contrast to most other approaches this method preserves small distances better than larger distances. The presentation will demonstrate the technique, show how distance matrices can be used for re-identification and how the proposed Lipschitz-embedding protects against respondent disclosure. Finally, an implementation of the technique in the programming language R is explained.

409

Schnell, R. and Thomas, K. The Effect of Formal Question Characteristics on Design Effects: A Cross-national Study Us- ing the ESS The widespread use of the Total Survey Error (TSE) framework in recent textbooks reflects the common knowledge that precision of a survey estimate not only depends on sample size and nonresponse problems, but on non-sampling errors as well. In practice, TSE models can rarely be applied since the individual effects can hardly be separated given the available data. Despite this limitation of standard survey data, indirect estimations of some elements of TSE are possible. One approach is the use of design-effects. Design effects are the in-crease in estimated standard errors by comparison to a simple random sample of the same size. Since design-effects can be estimated by using interviewer intraclass correlations, available design-effects permit the estimation of homogeneity of responses within the set of respondents assigned to an interviewer. Of course, for a single survey it cannot be de-cided if the homogeneity is due to spatial (social) homogeneity of the PSU or the inter-viewer. Previous research (Fowler (1991), Fowler/Mangione (1990), Schnell/Kreuter(2005)) has shown, that variance due to an interviewer is at least partially due to question characteris-tics. Examples are non-factual questions (e.g., question about attitudes), questions that are more difficult for respondents to answer as they require an increased cognitive effort (e.g., recall questions asking about longer time periods in the past), question formats capturing sensitive items (e.g., embarrassing or socially undesirable behaviour or attitudes), and open questions (e.g., absence of explicitly articulated response categories) are more likely to be affected by survey interviewer effects than their factual, easy, non-sensitive, or closed counterparts. Therefore, we developed a classification scheme for question characteristics which might make deviations from standardized interviewing more likely. We applied this classifi-cation scheme to all questions for the first seven rounds of the ESS. The variables of this classification are used in multi-level model with interviewer-homogeneity as dependent variable. The presentation will give details on the classification and it’s predictive power concerning interviewer related measurement error due to question characteristics.

410

References Fowler, F.J. (1991). Reducing Interviewer-related Error through Interviewer Training, Su- pervision, and Other Means. In Measurement Errors in Surveys P.P. Biemer, R.M. Groves, L.E. Lyberg, N. Mathiowetz, and S. Sudman (eds), 259–278. New York: John Wiley and Sons. Fowler, F.J. and Mangione, T.W. (1990). Standardized Survey Interviewing: Minimizing In- terviewer-Related Error. Newbury Park: Sage. Schnell, R. and Kreuter, F. (2005): Separating Interviewer and Sampling-point Effects. In: Journal of Official Statistics, 21, S. 389- 40.

411

Schoonenbooma, J., Pullen, A., Griffioen, D., and de la Croix, A. (University of Vienna, Austria; University of Applied Science Amsterdam, Netherlands; VU University Amsterdam, Netherlands) Integrating qualitative observation data and personal characteristics: overcoming the dilemma of sample size In this presentation, we discuss a problem that occurs with observational, explorative inquiry into social interaction. Our real-life case runs as follows. For two days, 27 university students collaborated face-to-face in nine small groups on a joint proposal for the design of a new product. The inquiry was aimed at the spontaneous role division between the students, which displayed itself in the number and nature of the initiatives that each individual student took to initiate a new discussion topic. Our two research questions were: 1. Is it possible to distinguish different collaborative role profiles on the basis of the number and nature of the initiatives, and if so, what are they? 2. What is the relationship between a student’s collaborative role and his or her personal characteristics regarding excellence? Qualitative data consisted of 27 hours of video material, including 2831 initiatives. Quantitative data consisted of standardized measures of various aspects of ability, personality, and motivation, among which were creativity and conscientiousness. In a first qualitative analysis, four role profiles were identified: Driving Force, Content Builder, Task Completer and Follower (Research Question 1). Subsequently, we aimed to link these role profiles to the personal characteristics that had been collected quantitatively (Research Question 2). The integration problem of this inquiry relates to sample size. The number of initiatives was large enough (N=2831) and rich enough to derive solid profiles, and thereby answer Research Question 1. The number of students (N=27), however, was too small to establish correlations between personal characteristics and the number and/or nature of the initiatives, and thereby answer Research Question 2. This led to the following dilemma. For the qualitative analysis, we needed rich data for each individual (in the sample on average one hour of video data per individual). For the quantitative analysis, we would need many individuals. Given our limited resources, it would not be possible to acquire a sample that would be large enough to perform the correlational analyses for answering

412

Research Question 2. Linking the qualitative data to the personal characteristics of a different and larger quantitative sample did not provide a solution, because the aim was to establish how the personal characteristics of one individual relate to the actual collaborative behavior of that same individual. In this paper, we discuss our two-step strategy out of this dilemma. Our strategy was based upon the idea that the personal characteristics, while they did not allow for statistical analysis, could be used to identify possibly interesting areas for inquiry. In our study, Driving Forces scored higher on creativity, while Task Completers scored higher on conscientiousness, something which is quite plausible theoretically. This conjecture was subsequently elaborated in an analysis in which creativity and conscientiousness were not viewed as continuous variables, but in which each level of creativity or conscientiousness was viewed as a group that was characterized by a specific frequency ratio of certain types of initiatives, namely the generation of ideas and task completion. Keywords: Mixed methods; integration; sample size

413

Schuman, A. (Center for Scientific and Social Studies) Punctuated Ethnography: using social time to plot fieldwork plans Time is a problem for ethnographic work, especially for those researchers who have academic responsibilities. The neoliberal university model, with its increasing use of part time professors and its orientation to efficiency and technical production, has little tolerance for immersive approaches to generating understanding of social settings. Researchers may have multiple projects and other responsibilities concurrently, making traditional, long term ethnography difficult or impossible to carry out. While long term immersion may remain the “gold standard° for anthropologists and allied social science disciplines, it is often only the period of dissertation research that affords the extensive time the methodology prescribes. There are alternatives, however. This paper seeks to illustrate, through the description of a project in process as of this writing, how periodic (“punctuated”) fieldwork can be structured so as to capture the essence of social processes without long term residence in the community of interest. This method relies on the researcher´s sufficient understanding of the setting to plot a schedule of fieldwork visits that coincides with important points in the social time(s) of the community. This can be achieved through prior knowledge or a “sensitizing” preliminary encounter with the community of interest. This strategy should not be confused with Rist´s blitzkrieg ethnography (1980), as it contributes only to the initial outlines of what becomes a long term, though discontinuous, involvement with various sectors of the community. Social Time Sorokin and Merton´s now classic article (1937) remains among the most useful considerations of the concept of social time, noting that “The need for social collaboration is at the root of social systems of time” (p. 615). They propose “social periodicities” (recurring economic, spiritual or communal practices) as the measure of duration of social time. Earlier (1912), Durkheim proposed that the concept of time itself derives from the “rhythm of collective life” (p. 23, note 6). An activity based phenomenon, social time is social motion. Zerubavel (1981) notes that events have a sequential structure. Within this, duration, temporal location (when they take place) and frequency of recurrence are all ingredients of social time.

414

Within his study of temporal patterns, Zerubavel identifies physiotemporal, biotemporal and sociotemporal orders, the last of which is of primary concern to him. Within the sphere of the work presented here, the sociotemporal overlaps with the biotemporal, given that a large part of the population of the area of interest is dependent on agriculture, hunting and harvest of wild products, both for food and for the production of handcrafts for sale. Indeed, this portion of the population may be said to be more influenced by biotemporal realities than strictly social ones. Alternatively, this social realm can be conceived of as resulting from the dialectic between the bio and the sociotemporal. In contrast, the other part of the population of the township is more governed by calendrical time, as its primary economic base is tourism. Although there is a regular flow of visitors, they are dependent on three “high seasons” that have specific durations that are determined by federal and state functionaries, as well as the religious calendar in two of the three instances. These are the Christmas/New Year holiday, the two week Easter break and the summer school vacation period. Through presentation of this on-going research in the southern Mexican state of Quintana Roo, the strategy of punctuated ethnography is exemplified. The overarching concern of the project is socio environmental conflict. Through reliance on fieldwork periods that conform to the social time(s) of the actors, a methodological approach is developed that is economical in its use of researcher time and faithful to the lived experience of the participants.

415

Schwarzenbach, A. (Max Planck Institute for International and Foreign Criminal Law) Youth-Police interaction in a Multi-Ethnic Society The German-French POLIS Research Project (Police and Adolescents in Multi-Ethnic Societies) delivers interesting insights on the youth-police interactions in a multi-ethnic urban environment. It is widely known that young people frequently interact with the police and therefore, studying the relationship between juveniles and the police is of pivotal interest. Thereby, the fact that some countries over time report more conflicts and more tensions in the youth-police interactions than others is of particular salience. The relationship of young people with the police is inherently shaped by the context in which interactions take place and affected by historical developments and national policies of immigration and integration. At the same times, in today’s multi-ethnic cities youth-police relationship faces new challenges. By retrieving on a unique comparative large-scale youth-survey (N=ca. 20.000) the factors that affect juveniles’ interaction with the police in Germany and France can be disentangled. No significant overall disparity in the frequency of police contacts in the two countries exists, but there are important differences by ethnic minority status. Other than in Germany, where no significant differences by ethnic minority status can be attested, in France adolescents of North African origin report significantly more police-initiated contacts, more physical violence during those contacts, and correspondingly express less trust in the police. Next to the ethnic minority status the impact of cultural identity, social ties, social deprivation and experiences with delinquency on juveniles’ probability of reporting police- initiated contacts in Germany and France is discussed. Additionally, results from the analysis of the impact of neighborhood effects on juveniles’ relationship with the police are presented. Finally, limits and challenges of large-scale comparative research on youth-police relationship are addressed.

416

Schwengler, B. and Kropp, P. (Institute for Employment Research, Germany) Stability of functional labour market regions During the last decades the delineation of functional regions based on commuting flows has been a topic of many studies in several countries. Most of these studies focus on one method and mostly on data of one year. As the number of commuters and the distances of home-to- work travel have increased in all modern countries during the last decades this study extends existing research on how these changes influence the delineation of labour market regions. Based on a former study where we proposed a new approach for delineating functional labour market regions (Kropp/Schwengler 2014), this paper focuses on the temporal stability of such delineations. This analysis uses commuting data between German municipalities from the German Federal Employment Agency statistics for all employed persons in Germany who are subject to social security contributions from 1993 to 2013. These are between 26.060 and 29.145 million employees in Germany. The Three-Step Method of our previous study was applied to the data of each three-year group. In a first step all bidirectional flows were computed among approximately 12,000 municipalities in Germany and their relative share compared to the resident employed labour force. If a small region was connected to a larger region then the largest share was labelled as a dominant flow. Regions were merged if the dominant flow exceeded a certain threshold value. In a second step commuting flows were recalculated between individual regions, between individual regions and aggregated regions, and between aggregated regions with the same thresholds. Then, both steps were iterated until no more changes occurred. Depending on the threshold values and the number of iterations, many possibly meaningful delineations of labour market regions were generated. The best solution was identified by applying the modularity approach of Newman and Girvan (2004) as a quality measure and choosing the delineation with the best value. Comparing the delineations with the best algorithm over time, we classify regions with a stable labour market attribution as core regions. Regions that sometimes form a labour

417

market on their own but belong to one particular other labour market region otherwise are called related regions, and regions which belong to different labour markets over time as overlapping regions. The main finding is that regions around important labour market centres form stable labour markets. Over time, most employees belong to the same labour market region or parts of it. However, increasing mobility reduces the share of the regional labour force living and working within labour market regions. As a result, the number of labour market regions has decreased over time as well as the quality of the delineations.

References Kropp, P.; Schwengler, B. (2014): Three-Step Method for Delineating Functional Labour Market Regions. Regional Studies. Journal of the Regional Studies Association, Online First, 17 S. DOI: 10.1080/00343404.2014.923093 Newman, M. E. J. and M. Girvan (2004). Finding and evaluating community structure in networks. Phys. Rev. E 69 (2004) 026113.

418

Sendelbah, A. and Vehovar, V. (University of Ljubljana) Effects of questionnaire characteristics on web survey page response times: analysis of an archive of web surveys Several authors have investigated effects of questionnaire characteristics on response times: Couper & Kreuter (2013) in a CAPI survey, Olson & Smyth (2015) in a CATI survey, and Yan & Tourangeau (2008) in four CAWI surveys. Overall, these studies have confirmed that response times are associated with question length, complexity and type. Our study extends existing research in several directions. Namely, our data source is an archive of web surveys created and hosted on the 1KA survey tool. Archive contains over 12,900 different web surveys, created by users from professional survey data collection organizations (academic, government and public sector), as well as by the so-called do-it-yourself users (Callegaro, Lozar Manfreda, & Vehovar, 2015). We use this vast and varied dataset to explore relationship between response times and questionnaire characteristics in different settings. Moreover, in contrast to published studies that have been focused on question- or item-level response times, we conduct our analysis on page- level response times. While page-level response time offer less detailed information about the pace of responding, they are are more widely available across different web survey platforms due to technical reasons. We have finished extensive process of data preparation. Our page-level predictors include word count, relative position inside the survey, frequency of all questions and items, as well as frequencies of specific types of questions (e.g. single-choice, multiple-choice, open-ended, table etc.). In addition, our set of page-level predictors also includes information on actual responses (i.e. on how many questions has respondent actually answered) and in case of open- ended items also information on the length of the answer. Preliminary analysis with simple correlation measures and linear models has shown some expected effects of several factors (e.g. higher word count leads to longer response times, as well as inclusion of open-ended questions). Following recommendation by Yan & Tourangeau (2008), our next step is to build a multi-level model to account for effects of survey, respondent and page characteristics on page response times. We are planning to finish our analysis by the end of March and write a working

419

paper in April.

420

Sendra, E. (Department of African Languages and Cultures, SOAS, University of London) Illustrating festivals research through the still and moving image: Finding an academic platform for documentary films This paper suggests that there are areas of research in which the methodology of participant observation and interviews can benefit a great deal from the use of visual methods. I will draw on my own PhD research of festivals in Senegal to argue that the adoption of visual methods can, on the one hand, engage with a more ethical research approach; and on the other hand, it allows researchers to better access the official and non-official versions of the research topic. In other words, while festival organisers may speak and think about their festival in a particular way, this may be experienced differently. Visual methods can help find the space between these two versions or visions of the festival. Firstly, the use of still images offers a more emphatic approach, where researchers allow their readers to better imagine the places and subjects that are being discussed. At the same time, it allows researchers to give something back to the people who make their research possible. In this way, they offer a bidirectional ethical approach, towards readers and “researched subjects”. Secondly, the moving image, and particularly, the making of a documentary film on the research topic, can allow researchers to see what the “researched subjects” think of the work, and what they aim to do with it, and compare this to what actually happens on the ground and how those aspirations are manifested and experienced by the people who benefit from that work. This advantage is inevitably and interestingly affected by the fear of “being looked at” and “shown to”. While initially there could be a positive attitude towards being taken a posed photograph, some people find it difficult to talk in front of the camera, or prefer not to. There are a number of ethical considerations to be taken into account to avoid this fear for the camera’s gaze. It is when this ethical approach is achieved that visual methods can benefit research, without losing sight of the difficulties implied in the process. At the same time, the documentary film can complement one’s written academic work and constitute an important educational tool that can, at the same time, be highly beneficial to the “researched subjects” in recognition of the work they do.

Sieber, S., and Schmidt, T. (Deutsche Bundesbank)

The Influence of an Up-Front Experiment on Respondents’ Recording Behaviour in Payment Diaries: Evidence from Germany In this paper, we analyse the recording behaviour of German consumers in a one week diary on their point-of-sales expenditures. We are particularly interested in the effect of a behavioural experiment, eliciting respondents’ risk preferences, on their recording behaviour. In the experiment, run shortly before the consumers start to fill in the diary, the consumers have the choice between receiving a sure payment of 10 euro and participating in a game. If they opt for playing the game they roll a die and either win 20 euro or nothing. We ask whether respondents’ recording behaviour differs depending on whether individuals who do roll the die lose or win. We argue that winners may attach a more positive feeling to the survey than losers and therefore exhibit more commitment to the diary, e.g. by reporting better quality data. Beyond providing evidence on the effect of conducting up-front experiments in representative surveys our results also contribute to the literature on incentives. For participants who roll the die, the experiment can be seen as a tool to randomly assign an incentive to respondents. Our results indicate that the outcome of the game has an impact on the quantity of transactions recorded, but does not affect other aspects of data quality. It also has a negligible impact on substantive measures like the cash share.

Keywords: data quality, risk experiments, payment diary

Simpson, A. (University of Brighton, England) ‘Studying Up’: Themes of and issues of the inverted ethnographic gaze Drawing on past ethnographic research that explored constructions of deviance and financial life in the City of London, this paper presents a methodological evaluation of studying ‘elite’ cultural spaces. The ethnographic study of power and privilege, or what Gusterson (1997) calls ‘studying up’, is of increased importance in a world characterised by issues of capitalist accumulation and widening structural inequality. A critical reflection on the role of ethnography in cutting through veiled institutional arrangements and the restraining force of power is, arguably, more prevalent than ever (Dinerstein, et al., 2014; Ho, 2009). However, shedding new light on questions of power and legitimacy that structure our society raises crucial methodological concerns of access and resource constraints. Throughout my time in the field, these issues, in a very real sense, limited the scale and scope of my three year doctoral project. Initially, the aim of complete participation and the promise of ‘inside knowledge’ was a persistent lure. However, in a cultural and spatial context where information was prevented from leaking out (let alone welcoming critical ‘outsiders’ in), such aims were always to prove, at best, extremely limiting and, at worst, implausible within the scope of study. Focusing on these issues, this paper asks how can we, as ‘outsiders’, adequately overcome structural limitations and the imbalance of power to ethnographically explore the experiences of elite cultural fields? What this paper argues is that it is possible to establish a ‘fragmented ethnography’ that utilises a mixed methods approach to open the particulars of social experience, ameliorating understanding of an otherwise closed, elite network group. Drawing on aspects of Knoblauch’s (2005) ‘focused ethnography’, this paper establishes a compressed and targeted methodological strategy that places the emphasis of inquiry on intensive multi-sensory experiences which speak to cultures of deviance and harm. However, rather than focusing on small elements of one’s own society, as Knoblauch (2005) suggests, this paper refocuses attention to areas of specialised and elite activity that takes place beyond the traditional reach of critical attention. In the context of the study of social deviance in the City of London, this incorporated a three-tiered research process involving non-participant observation, photographic representation and in-depth interviews. Each methodological strategy sought to purposively capture social meanings, representations of experience as well as key

descriptions, explanations and evaluations of cultural life (Brewer, 2000; O'Reilly, 2005). Led by direct interactions in the field, this three tiered methodological process was able to shed light on the symbolic practice of interaction whilst retaining an analytical reflexivity that built on experience, through the simultaneous process of social interaction and integration, to establish a deeper exploration into the presentation and explanation of culture. This fragmented and multi-site methodological strategy succeeded in compressing the time spent in the field, favouring intensity over extensity, to fit into the demands of a three year doctoral project. Without wanting to leave the ethnographic study of the powerful to informed insiders, it was a strategy that continuously probed and challenged issues of access to create a multi-layered framework of experience. The end result is an overarching picture of social life that challenges the structural limits of access and resources, to offer a focused account of experience. Moreover, it offers a cultural exploration of an embedded system of situated action and meaning making of an elite social group. Bibliography Brewer, J., 2000. Ethnography. Buckingham: Open University Press. Dinerstein, A., Schwartz, G. & Taylor, G., 2014. Sociological Imagination as Social Critique: Interrogating the 'Global Economic Crisis'. Sociology, 48(5), pp. 859-868. Gusterson, H., 1997. Studying Up Revisited. PoLAR: Political and Legal Anthropology Review, Issue 114-119. Ho, K., 2009. Liquidated: An Ethnography of Wall Street. Durham: Duke University Press. Knoblauch, H., 2005. Focused Ethnography. Qualitative Social Research, 6(3). O'Reilly, K., 2005. Ethnographic Methods. London: Routledge.

Simpson, M. (North Carolina Central University) Missing Demographic Data in Organizational Surveys: Disguising Hostile Respondents Researchers, surveying personnel within organizations, confront respondent trust. If respondents do not trust the researcher or management, they may not disclose revealing information such as basic demographic data. The more respondents believe that management might identify them and the more negatively they view their organization, the greater the probability they will avoid detection. If they avoid detection by deliberately not answering a demographic question, the error is more than just “Nonignorable Missing Data.” It inherently biases the data as a non response is a function of individuals social statuses. These are Nonrandom Intentional Missing Data “NRIMD.” In this study, we test the hypothesis that respondents with negative views of their unit have a higher probability of not answering a demographic question. The study employs in the United States Department of Defense's Military Equal Opportunity Climate Survey(MEOCS). The Defense Equal Opportunity Management Institute (DEOMI) administered the MEOCS to over 593,640 personnel between 1990 and 1997. The MEOCS comprises twelve scales that assess EO behaviors and attitudes, general organizational climate, and job satisfaction. The hypothesis was strongly supported. Respondents, who failed to answer the demographic items, held significantly more negative views of their unit's EO climate, its general organizational climate, and had lower job satisfaction. Missing demographic data in the MEOCS are not NRMMD. Two troubling questions arise. First, while dropping missing demographic data biases the results, standard missing data corrections, Multiple such Imputations and EM, fail, as well. Also, others, who fear exposure, may merely fabricate a false identity. If these individuals are numerous, the results could be seriously flawed. Possible solutions and strategies to minimize the impact of NRMMDs are discussed.

Singer, H. (Fern Universitaet, Hagen) Simulated Maximum Likelihood for Continuous-Discrete State Space Models using Langevin Importance Sampling Continuous time models are well known in Sociology through the pioneering work of Simon, Coleman, Doreian and Hummon and others. Although they have the theoretical merit in modeling time as a flowing phenomenon, the empirical application is more difficult in comparison to time series models. This is in part due to the difficulty in computing likelihood functions for sampled, discrete time measurements (daily, weekly etc.), as they occur in empirical research. With large sampling intervals, one cannot simply replace differentials by differences, since one then obtains strongly biased estimates of structural parameters. Instead one has to consider the exact transition probabilities between the times of measurement. Even in the linear case, these probabilities are nonlinear functions of the structural parameter matrices with respective identification and embedding problems. For nonlinear systems, additional problems occur due to the impossibility of computing analytical transition probabilities for most models. There are competing numerical methods based on nonlinear filtering, partial differential equations, integral representations, Monte Carlo and Bayesian approaches. We compute the likelihood function of a nonlinear continuous-discrete state space model (continuous time dynamics of latent variables, discrete time noisy measurements) by using a functional integral representation. The unobserved state vectors are integrated out in order to obtain the marginal distribution of the measurements. The infinite-dimensional integration is evaluated by Monte Carlo simulation with importance sampling. Using a Langevin equation with Metropolis mechanism, it is possible to draw random vectors from the exact importance distribution, although the normalization constant (the desired likelihood function) is unknown. We discuss several methods of estimating the importance distribution. Most importantly, we obtain smooth likelihood surfaces which facilitates the usage of quasi Newton algorithms for determinating the ML estimates. The Monte Carlo method is compared with Kalman filtering and analytical approaches using the Fokker-Planck equation.

More generally, one can compute functionals of diffusion processes such a option prices or the Feynman-Kac formula in quantum field theory. Keywords: - Nonlinear continuous-discrete state space model - Simulated Maximum Likelihood -Langevin Importance Sampling

Smietana, M. (University of Cambridge (UK), Sociology Department, Reproductive Sociology Research Group – and University of California Berkeley, Department of Gender & Women’s Studies) Good science, reflexivity and ethnography: Methodological choreography of a sociological sponsored project In this paper I discuss the opportunities and limitations resulting from introducing reflexive and ethnographic elements in the sociological qualitative research study of surrogacy, ‘SurrogARTs,’ sponsored by the European Commission as my postdoctoral award (2015- 2017). The ethnographic elements, such as shadowing and observation of interviewees (Thompson, 2005) and multi-session interviewing (Stacey, 1990), have been subject to necessary innovations due to challenges presented by the nature and context of the research. The idea of the study was to follow several clusters of intended parents, surrogate mothers, egg donors, and clinic or agency practitioners throughout pregnancy, to see their experiences of the process. This, however, brought at least four challenges: (1) the multiplicity of research sites (several clinics in often transnational arrangements), (2) the limited temporal dimension of the study (the pregnancy), (3) human subject protections, imposed by the involved institutional review boards (IRB), (4) and recruitment difficulties (due to stigmatized nature of researched infertility). Therefore three major methodological innovations were assumed throughout the study: (1) use of technology at times and in sites where I couldn’t be present or wasn’t allowed in person as a researcher, in the form of self-recording by the interviewees, and regular e-mail follow-ups; (2) ethnographic vignettes (Thompson, 2005, 2013): limiting observation to public events such as agency anniversaries and conferences, or to selected key private occasions in the surrogacy process, such as the medical ultrasound visit or picking up the baby; (3) involving agencies, clinics, and associations as key informants and advisors to the project, which both enhanced recruitment and made the methodology more inclusive of the researched. These ethnographic elements have brought unique insights, in line with Judith Stacey’s (1990: 33) observation, ‘Time alone with Dotty after we had established a relationship of substantial rapport and trust radically shattered the homey tale of a rescued, way ward Silicon Valley marriage that the couple had presented for scholarly scrutiny.’

At the same time, from the reflexive viewpoint, as a researcher, I faced the challenge of addressing the complexity of the existing power relationships within which my interviewees were immersed, without downplaying their egalitarian aspirations or causing harm to their parenting projects. As pointed out by Judith Stacey (2004), there has been social pressure for non-normative families and the research on them to prove that they are no different from their heterosexual counterparts, which demonstrates that social inclusion limits de novo definitions of family formation to the given gendered, classed, and racialized hierarchies and histories of particular jurisdictions. Keywords: qualitative sociology, ethnography, transnational surrogacy

Smith, T. (NORC/University of Chicago) Using the Total Survey Error Paradigm to Maximize Comparability in Cross-National Surveys Total survey error (TSE) is a very valuable paradigm for describing and improving surveys, but it can be improved. One key limitation is that TSE was formulated to apply to a single, standalone survey. Yet most survey research combines and compares surveys. TSE can be extended to cover these multi-survey utilizations. TSE needs to be thought of as heavily involving the interaction of error components and the concept of comparison error should be used to extend TSE to cover multiple-surveys including trend analysis, comparative studies, and longitudinal panels. This extension of TSE will greatly improve the design of multi-surveys in general and of comparative (i.e. cross-national/cross-cultural) surveys in particular. Likewise, using TSE can greatly advance the analysis of comparative data by using it to assess and adjust for difference in the error structure across surveys. A comprehensive TSE typology should be used whenever comparative studies are designed and also whenever secondary analysis of comparative studies is carried out. In particular strict application of the TSE paradigm can help to achieve the goal of functional equivalence cross-nationally/culturally. Minimizing TSE is an important goal in survey research in general and is especially valuable for comparative survey research and the TSE paradigm should be used as both an applied application and a research agenda to achieve that goal. Extensive examples from the ISSP and ESS will be used to demonstrate this approach.

Smuda, R. (University of Koblenz Landau, Graduate School “Gender Studies”) Tensions in field access – A reflection Just as writing abstracts for conferences is everyday practice in our ethnographic research project “Gendered University”, so dealing with mechanisms that complicate or even impede field access is also an integral part of our daily work. We are concerned with researching the gender order within the context of higher education against the backdrop of changes in the university with reference to equality during the Excellence Initiative. Thus, gaining access to different areas of the university is a prerequisite for this research project. However, researchers find themselves confronted with an assortment of problems. First of all, these are aspects of practical research. At the same time, examination of these aspects also points to transverse issues. The extent to which these problems are nevertheless mastered and may even harbour opportunities will be discussed in the following. Many challenges related to field access arise particularly due to the time framework regulating the research project. In these situations the researchers have various options for action at their disposal that are accompanied by an upgrading or downgrading of the quality and depth of the data. Weighing up the ways of dealing with the problems of field access can signify a considerable loss of resources. Apart from the resulting demands on an already tightly scheduled project, these decision-making processes influence questions of method that are associated with processes of self-authorisation and also negotiation within the team. (cf. Breidenstein et al., 2013: 50pp.; Dellwing/Prus, 2012: 142pp.; Wolff, 2000: 340pp.). In this context, the challenge is to authorise oneself. With regard to this, the question has to be asked whether this is a matter of research problems that are generally on-going or whether it is something that affects newcomers in science particularly. A transverse problem opens up to researchers who are in a quasi-double role as learners and developers of the method: they are faced with the constant task of striking a balance between doing justice to a tried and tested methodology and emancipating themselves from it to a certain degree in order to be able to shape their own research paths. In summary, it becomes apparent that difficulties in field access require a rethinking of research practice on the one hand, but on the other, are accompanied by a constant confrontation with transverse issues. Even if this leads to the team being challenged in terms of coping with the

accompanying frustration and the tensions during the course of decision making, new opportunities open up precisely as a result of having to cope with these issues. In reacting to adverse field accesses and attempting to get to the bottom of these, researchers are guided by the field, which indeed leads us to methodical challenges, but at the same time enables us to focus our research in order to make it more field-specific, thus generating more precise results. Our contribution is concerned with including the experiences we have gained to this effect during our research project in a scientific exchange of ideas.

Keywords: ethnography, field access challenges, young scientists

Literature Breidenstein,G./Hirschauer,S./Kalthoff,H./Nieswand,B. (2013): Ethnographie. Die Praxis der Feldforschung. Konstanz/München: UVK. Dellwing,M./Prus,R.(2012): Einführung in die interaktionistische Ethnographie. Soziologie im Außendienst. Wiesbaden: VS. Wolff, S.(2000): Wege ins Feld und ihre Varianten. In: Flick, U./von Kardorff, E./Steinke, I.: Qualitative Forschung. Ein Handbuch. Reinbek/Hamburg: Rowohlt.

Snijders, T. and Ortmann, M (TS: University of Groningen and University of Oxford; MO: University of Konstanz) Networks in Social Contexts: the Settings Model The ‘Settings Model’ is a newly developed stochastic actor-oriented model for network dynamics. The data structure includes, next to the network by itself, a representation of the social context in which the network develops. Settings are regarded as meeting opportunities, structuring the creation of new ties. A setting is defined as a graph, usually non-directed, and rules for tie change depend on the collection of settings. Some settings can be given exogenously (e.g., classrooms, departments), others endogenously (e.g., extended local network neighbourhoods). The Settings Model is conceptually as well as computationally more attractive for representing network dynamics in larger networks. This development is meant to extend the applicability of the stochastic actor-oriented model to larger networks. The definition of the Settings Model is discussed and methods for parameter estimation are presented. Some examples are presented: illustrating the difference between the regular actor-oriented model and the Settings Model; and illustrating the use of different specifications of the settings.

Spencer, N. (University of Hertfordshire) Giving researchers confidence to use quantitative methods: overcoming the fear of choosing the wrong technique A good course on Research Methods will include details of the strengths and weaknesses of qualitative, quantitative and mixed methods of analysis. It will also emphasise that a “horses for courses” approach should be taken when deciding upon research methodology, selecting the method(s) that are best suited to addressing the research questions in the context of the availability of resources and information/data. However, it is frequently the case that in practice researchers will have an innate preference for qualitative or quantitative methods, based upon their own personal strengths and weaknesses. Research aims, objectives and hypotheses may be bent in such a way that particular methodologies can be used. In quantitative analysis, there are a large number of techniques that are available for the researcher to use and the ability to have methods that cope with many scenarios is a strength. However, this multiplicity can also create confusion and deter researchers from using quantitative approaches even when they may be able to meaningfully add to the quality of the research. The fear of using inappropriate methods and having their lack of confidence and expertise exposed by peer review of journal articles, etc. is understandable. This paper provides a resource to help researchers choose appropriate methods of quantitative analysis and thus give them confidence that they can undertake studies of this type. Initially it uses flow charts to guide the researcher towards an appropriate family of statistical procedures based upon the research hypotheses being addressed. It argues that flow charts such as these should be an essential part of any research methods training (even if there is no intention to teach the statistical methods themselves) because they build confidence in those who may be fearful of such approaches that using quantitative methods is an attainable goal. Flow charts such as those described above can be found in a number of text books but to date, none properly tackle the thorny question of whether a dataset is best suited for parametric or nonparametric analysis – an important issue in deciding on the appropriate statistical technique in a number of circumstances. Researchers are pointed towards a range of graphical and numerical summaries that can be created to assist in making a decision about whether their data are “Normal enough” but there are no definitive guidelines and only vague, inconsistent

and sometimes unhelpful rules of thumb. The researcher is thus left to exercise his or her professional judgement in an ad hoc manner and this erodes confidence. This paper addresses this issue with the use of further flow charts that distil the work undertaken by statisticians since the 1920s into clear guidance to the researcher as to the most appropriate statistical technique to use. Key words: Flow charts; methods of quantitative analysis; normality

Springer, S. (European Union Agency for Fundamental Rights) Cognitive testing using different interview modes in different cultural contexts – the role of age In the last years the use of different survey interview modes, or their combination, have been considered promising when faced with problems such as increasing interview costs, decreasing response rates and non-response bias. While mode effects have been the subject of extensive research mainly at the national level, little is known about the importance of these effects in cross-national surveys. In particular age might have a different impact on the accuracy and reliability of answers in function of the interview mode and the cultural setting. The European Union Agency for Fundamental Rights (FRA) is currently developing the design for a general population survey on fundamental rights, which will cover all 28 EU Member States. The survey will ask questions about cognitively challenging subjects, such as rights awareness and experiences of fundamental rights ‘in practice’ in different areas of everyday life. Searching for optimal use of limited resources while ensuring the high quality of data, the FRA has opted for cognitive testing that includes the assessment of impact of the different interview modes. Using the same questions, an equal number or respondents has been interviewed either face-to-face, by telephone or they have been asked to complete a self- administrated web-based questionnaire. The objective was to find out if certain questions or types of questions are problematic when using a particular interview mode, taking into account the age, gender and education of the participant. To ensure that in a cross-national context questions are understood correctly, that responses are accurate and reliable, and response categories reflect the experiences of the general population, the cognitive testing applied the same structured survey questionnaire in six EU Member States, followed by cognitive interviews. This approach will allow for the build-up of knowledge on mode effects in a cross-national context. This presentation will show preliminary findings from this cognitive pre-test, concentrating on the importance of age in relation to mode effects in cross-cultural surveys. Keywords: Mode effects, age, cross-national

Stadtfeld, C. (ETH Zurich) Relational Events: Comparing Actor-oriented and Tie-oriented Models The advent of electronic communication, social media, and human sensor technologies have brought about a wealth of dynamic, time-stamped social interaction data that are often easily accessible to social scientists. The Stochastic Actor-Oriented Model (SAOMs) is the most widely used approach for the study of social networks through time. To date, however, SAOMs are mainly applied to the analysis of network panel data. In the recent years new methods have been developed to study sequences of social interaction data without aggregating those into data panels. The most prominent approach is the Relational Event Model (REM). Besides different software implementations that are tailored to specific types of data, the SAOM and the REM differ in their unit of analysis: the SAOM is actor-oriented and the REM is tie-oriented. In this talk, we discuss how an actor-oriented perspective can be taken for the study of time- stamped social interaction data. We present a new software package that allows fitting both actor-oriented models for time-stamped network data and the tie-oriented relational event model and compare results obtained from both methods. We argue that the choice between a tie-oriented or an actor-oriented approach should be guided by theoretical considerations.

Stähli, M., and colleagues of the ‘International Surveys’ team of FORS. (FORS Swiss Centre of Expertise in the Social Sciences)

Threats to comparability: beyond translation and wording When analysts of comparative data find differences they cannot interpret, they may ask themselves if methodological artifacts create these differences. They first would look at differential wording and translation of the incriminated question. There however might be other sources of methodological determinants of the answers, more difficult to control for the analysts, such as sampling and selection effects, mode and context effects. And these effects exist even when the results are reasonably interpretable. In this paper we tackle the question of how far such elements can shape the answers, as compared to wording and translations. For this paper we focus on two simple and widely used concepts: interest in politics and satisfaction with democracy. We explore different sources of data, at national and international level, essentially from the European Social Survey as this survey has highly harmonized methodological specifications and strict quality control procedures. The central aim is to observe what matters more for comparability: an identical wording (respectively ‘good translation’), an identical question format, or other surveys design aspects (the way respondents are sampled and recruited, the data collection mode, the question context, etc.). By crossing the results from different sources, varying the dimension of their differences such as survey, mode, sampling, countries, wording, question format, language, position in questionnaire, etc., we gain a feeling of the respective importance of these different dimensions. An additional test-retest experiment, combining question versions from several surveys in a single questionnaire (Swiss ESS), allows further identifying and disentangling the different sources of the differences. Our results lead us to the conclusion that more research is needed to explore some unmeasured dimensions such as the way the survey is presented, the style and content of previous questions, as they seem to play an often underestimated role. With this contribution, we aim at widening the discussion about comparability beyond wording and translations to additional survey design features. We also explicitly adopt a perspective of constructed reality, thinking that survey measurements are the product of specific survey tools and conditions, and not mere measurements of a reality with potential errors. In this sense we are not literally looking for errors, but trying to understand the process of answer production.

Stamm, I., Eklund, L. and Liebermann, W. (University of California Berkeley, Department of Sociology, Stockholm University, Department of Sociology, University of California Berkeley, Department of Architecture) Crowdsourcing as research method: hype, hopes, and hazards. Crowdsourcing is often hyped as a key element to handling the big data revolution and thereby create new forms of collective knowledge production. This paper aims to demystify the promises of crowdsourcing as research method by drawing attention to the ways in which crowdsourcing affects meanings of scientific knowledge production. Our analysis is informed by the proceedings of a symposium on crowdsourcing in the humanities and social science held at UC Berkeley in 2015, in conjunction with content analysis of ten websites of crowdsourcing service providers, and a critical reflection on the discourse of digital methods. We show that the translation of crowdsourcing from a business process to research method is not trivial. This sociotechnical practice is distinct from other methods as it conveys epistemological assumptions underlying other kinds of methods. While there is no doubt that crowdsourcing can be useful as a research method, the assessment of the produced knowledge lacks traditional research standards. Crowdsourcing rests upon an unpredictable composition of the crowd, malleability of tasks, and in-built ambiguity of platforms, leaving room to draw contradictory fantasies about the crowd. The challenge is to create a fit between the assumed characteristics and capabilities of crowd-taskers and the assigned task. Consequently, the image we have of the crowd and its capacities, and the ideas that researchers have about what good research is together must act as an implicit compass requiring much greater reflexivity than necessary for other kinds of methods. Hence, we find that crowdsourcing does not per se affect meanings of scientific knowledge production. Rather, the applicable image of the crowd shapes the methodological design in all its detail. On a pragmatic note, clarifying how researchers understand and draw on the platforms, crowd-taskers, and tasks has to be the starting point for any crowdsourcing project. These reflections help researcher make decisions about the design of the crowdsourcing process and, eventually, to assess the validity of data and interpretations. Ultimately, we conclude that crowdsourcing does not fit within a single epistemological paradigm, which complicates the assessment of the quality and value of the knowledge produced. Keywords: Crowdsourcing, research method, digital method, methodology, big data, collective knowledge

Stathopoulou, T., and Papageorgiou, H. (National Centre for Social Research-EKKE. Greece, Institute For Language And Speech Processing- Athena Research Centre. Greece) Radicalized Orthodoxy? Civic discontent and religious identification in times of crisis. The economic recession which began in 2008 has triggered a transatlantic wave of protest against monetary policies and austerity measures. At a European level, the effects of crisis have been particularly evident in a significant decrease in trust for politicians and political parties and in the rise of the Far Right. In Greece, civic discontent has manifested itself in large, often violent, protest events, an increasing rate of non-participation in national elections, and in votes for the radical Right. The increasing migrant flows over recent decades, and the recent unprecedented refugee crisis in particular, have been at the epicenter of contentious collective action and political controversy. In this context, the anti-immigrant discourse and practices of the Far Right party are presented as counteracting responses to the transformation of Greece from an Orthodox majority to a diverse religious minority country, thus fuelling public disputes over religious freedoms and national identity. The argument is taking shape along the lines of an exclusionist view of Orthodox identity, thus preserving the strong link between religious and secular nationalism. This paper will examine the potentially significant role of religion in political radicalization and mobilization in the Greek case, as well as the effect of religious identification on “the politics of blame”. Data from various sources will be used to analyse religiously-motivated hate speech and verbal aggressiveness: newspaper articles, online religious sources and social media (Twitter) for the period 2008-2015. Based on a computational social science approach employing automated content analysis with Natural Language Processing Tools (NLP), this work will provide new insights into how radicalizing discourse emerges and evolves. An NLP pipeline unique for the Greek language will be used which comprises a series of processing units based on both machine-learning algorithms and rule-based approaches. Keywords: radicalization, religion, social media

Steele, J., and Ferrer, E. A dynamic systems approach to relationship affect over time: An application of the Exact Discrete Model. In Psychology, the use of continuous time modeling is rare, however, many of the theories in Psychology are dynamic in nature and are often conceptualized as unfolding in continuous time. This makes many areas of Psychology well equipped, if not ideally suited, for continuous time modeling methods. A major obstacle for many practitioners is an intuitive explanation of how discrete time measures, like those common in Psychological research, can be connected to the continuous time processes that are theorized to generate the observable data. Continuous time processes are specified using differential equations, wherein change is the outcome of interest. When the underlying process is believed to be smooth and error free, the system can be expressed using ordinary differential equations, ODEs. However, if the underlying process is also susceptible to stochastic influences, this changes the deterministic system from an ODE to a stochastic differential equation or SDE. The process level stochastic influences do not follow the general rules of differential calculus and require specialized methods to express them sufficiently in continuous time. Once the continuous time portion of the system is specified, there remains a necessary step of connecting it to the available discrete time measures. In this work we use the Exact Discrete Model developed by Bergstrom (1986) to fit daily diary data from romantic couples to a continuous time process. Specifically, we utilize an SDE specification of a continuous time model for dyadic interactions developed by Felmlee and Greenberg (1999). After exploring purely deterministic approaches for fitting our data, we simulate and confirm the SDE specification using a combination of PROC MODEL form the SAS/ETS library as well as the LSDE package written in SAS/IML by Singer (1990). Results from the empirical data are then discussed along with some of the key milestones involved in our progression toward use of the EDM.

Stein, P. and Bekalarczyk, D. (University Duisburg-Essen) Longitudinal modeling of the future development of occupational status in the third generation of migrants in the context of demographic transition by means of a dynamic micro simulation The subject of our project (funded by the German Research Foundation DFG) is to model the future development (30-40 years) of the occupational status – as the central dimension of assimilation – of migrants, who live in Germany. The goal is to answer the question, how assimilation patterns will develop in the context of demographic transition in Germany. We expect that this transition will cause compositional effects, because it changes the ethnic structure of the migrants living in Germany as well as the proportion of all migrants among the whole population. This perspective contributes to the current discussion in the migration research, which is focused on the complex causal mechanisms of assimilation on the individual and institutional level. For theoretical reasons, this project focuses on the third generation of migrants. It is currently not possible to analyze the occupational status in the third generation directly, because most of their members are too young. However, it is possible to forecast the development of occupational status in this generation reliably, because data exists for the most important mechanisms: education (results of in-between stages) and social origin. The methodical goal is to demonstrate that prognosis or simulation methods are appropriate for many sociological research questions (Stein & Bekalarczyk 2016). These methods even have advantages compared to standard multivariate statistical methods, which base on the assumption of linearity or other known mathematical functions. These assumptions are often too simple to reproduce the dynamics of real processes. However, the dynamic can freely “flower out” in a simulation, because there is no need to predetermine the mathematical form of a process. Despite of these advantages these methods are rarely used in sociology. Sociologists seem to have some reservations, which can be easily disproved (Gilbert & Troitzsch 2005): Sociologists know about the poor predictability of the most empirical regression models and about the complexity of processes they deal with. Thus, prognosis methods have a bad reputation, associating with them audacious attempts to predict unpredictable complex processes. But prognosis methods are rather to be understood as

tools with which one can test the outcomes of different (competing) hypothesized scenarios about the future developments, holding factors, which are not of interest, constant. So it is rather to see as a quasi-experimental design. Another unfounded fear is that simulation methods work with artificial data and thus are not able to deal with complex information coming from big empirical data. This is not true. In the methods we use (see below), it is possible to apply broad empirical information. The chosen method is the dynamic micro simulation. In contrast to a simple trend projection, this technique allows to model competitive hypotheses about assimilation mechanisms on individual level and to connect these mechanisms with demographic-relevant processes (child-bearing, aging, dying, migrating). It is also possible to implement cross-level hypotheses including feedbacks from macro-level to the micro-level. The extrapolation is of stochastic nature. This means that random experiments yield the simulated values. Thus the exact result is not predetermined by the settings of the model. This copes with the fact that models always just imperfectly fit with reality and thus makes the results more “realistic”, allowing to quantify the uncertainty of the results. Parameters for the prognosis model will be estimated by longitudinal analyses with empirical data (German Socio-Economic Panel and microcensus). Data from a current microcensus shall act as empirical start data for the simulation. For the longitudinal analyses we use elaborate methods which consider the structure and the possibilities of panel data (such as random effect and growth curve models). References: Gilbert, N. / Troitzsch, K.-G. (2005): Simulation for the Social Scientist. 2nd edition. Open University Press. Stein, P. / Bekalarczyk, D. (2016): Zur Prognose beruflicher Positionierung von Migranten der dritten Generation, in Bachleitner, R. / Weichbold, M. / Pausch, M. (eds.): Wissenschaftstheoretische und methodologische Problemlagen empirischer Voraussagen und statistischer Vorhersagen. Grundlagen und Kontroversen erfahrungswissenschaftlicher Prognoseverfahren, VS-Verlag für Sozialwissenschaften, pp. 171-205

Stevenson, A. (Manchester Metropolitan University) How We Know Each Other: Exploring friendships through filmed friendship ethnography This paper, which includes film excerpts, discusses an ethnographic research project that incorporates film to explore the practices that cement bonds of friendship amongst four friendship pairs. The research explores, using interviews and primarily ethnographic film, practices of friendship. This allows the research to go beyond purely verbal reports and explore friendships as emplaced practices. In exploring the practices and spaces that bind these friendship pairs, it was decided to conduct the research predominantly amongst friends, with participants with whom the researcher is already acquainted to varying degrees of intimacy. This paper reports on research that employs friendship ethnography with ethnographic film to explore friendship itself, and the practices that bind it. The resulting documentary film features friendship pairs involved in dynamic practices such as power yoga, Scottish country dancing, car-sharing and campaigning against food poverty. Arguably, situating the research amongst friends and treating participants according to the ethics of friendship draws us closer to an understanding of the nature of these bonds. Through the use of filmed friendship ethnography, the author argues that friendships are practice-based, suffuse, routinized and emplaced phenomena. The presentation will be illustrated with excerpts from the author’s recent ethnographic film, How We Know Each Other, around which the paper is based. Keywords: Film, friendship, ethnography

Stolz, J. (University of Lausanne) A theory of social games as a metatheory for mixed methods research This paper argues that mixed methods research is currently hindered by a lack not of designs, but of metatheory. Due to a lack of proper metatheory, it is not clear just how and why different methods should be mixed and integrated. I suggest that a new theory of "social games" that draws on a range of different theoretical classics (especially Boudon, Bourdieu, Goffman, Giddens, Wittgenstein) may constitute such a metatheory. Social games are real entities that incorporate players and their attributes, resources, rules, representations, and context. They have micro- and macro properties that may be measured and analyzed with quantitative methods (e.g. distributions of player-attributes, game-attributes); they also have also very distinctive properties that need qualitative methods in order to understand their inner workings and processes: rules that may be reconstructed in textual form, situations-in- the game that have to be "understood", cultural representations that involve different layers of meanings. The metatheory of social games explains why mono-method research often lacks proper explanatory power and helps focusing mixed methods research. I give two examples that show how a metatheory of social games may be used in order to conduct fruitful mixed methods research: an analysis of the reasons why individuals were saved on the Titanic and an analysis of the dynamics of bureaucracy by Peter Blau. Keywords: Mixed Methods, Theory, Social Games

Stoop, I., Denies, K., Douhou, S., Halbherr, V., Kappelhof, J. and Loosveldt, G. (The Netherlands Institute for Social Research/SCP) Minimising unnecessary variation and optimising quality and comparability in the European Social Survey The European Social Survey was conceived in the 1990s by social and political scientists deploring the absence of good-quality comparative data about attitudes, opinions and values. The mission of the biennial ESS is to achieve and spread higher standards of rigour in cross- national research in the social sciences, and to provide data that allows the monitoring of variations over time and between countries. The ESS intends to do this by:  Aiming for high quality, both by implementing national and international best practices and by continuous improvement of methods applied. Key elements are strict probability sampling, face-to-face interviewing, pursuing high response rates, and a thorough questionnaire development and translation process.  Pursuing optimal comparability across countries, through adherence to a uniform set of principles and procedures set out in the project specifications, taking into account national and cultural contexts, and minimising unnecessary variations across countries wherever possible.  Focusing on consistency in methodology and content over time, and thus making it possible to study longitudinal trends. Since the first round in 2002 researchers have become aware of restrictions, trade-offs and dilemmas in aiming for high quality, optimal comparability and consistency. There are areas in which the ESS does well, areas in which the ESS could do better, and areas in which the three aims are hard to conciliate. The ESS is based on input harmonisation, ensuring the fielding of identical questionnaires in each country. Questionnaires are cognitively tested in several cultures and languages. A range of guidelines, tools and assessment measures have been developed to produce translations that have an equivalent meaning in each language. Obtaining functional equivalence is far from easy, however. Questionnaires must be relevant in each of the participating countries, even though health care and educational systems, welfare provision and employment patterns differ. Cultural differences can result in different uses of answer scales, and different

rates of Don’t Knows. The large number of countries makes it impossible to conduct cognitive interviews in every country. Equivalence tests after data collection show that differences in the meaning of concepts and questions between countries and languages cannot be excluded. Secondly, the interview should be conducted in the same way in each country. Some countries use PAPI, however, whereas others use CAPI. One consequence is that the collection of time stamp data will not be possible in some countries, and hence that quality monitoring cannot take place at the same level. In addition, a central monitoring or information about the fieldwork process is not available in real time. Another complication is that the position (free-lance, employees) and remuneration (amount; per interview, per sample unit, per hour) of interviewers differs across countries. Comparability is pursued through standardised training and briefing. These are run by local organisations, however, which may have an impact on comparability, and consistency when fieldwork is not always conducted by the same agency in a country. Thirdly, with regard to sampling and respondent recruitment achieving comparability is also a challenge. There are strict rules in these areas (probability sampling, no substitution, no proxy interviewing, rules for the number and timing of calls), and expert advice is given on the sampling design, fieldwork procedures and response enhancement. Still, there are cross- national differences in the availability of sampling frames, the presence of minority language groups, the experience of interviewers with probability sampling, the likelihood that people will be found at home, and their willingness to participate in a survey. Detailed and complete documentation of fieldwork through contact forms can help to improve quality and comparability, or at least to monitor and assess differences in fieldwork procedures, but this cannot ensure comparability. Finally, there is a need to improve and adapt to changing circumstances. Rephrasing questions that did not work well, using incentives to enhance response rates, standardise interviewer briefing across countries: every measure to improve data quality in between rounds may reduce consistency over time. A lack of comparability seriously diminishes the value and usefulness of cross-national surveys. Input harmonisation is one means to pursue optimal comparability, but it can never ensure this. A decentralised organisation of fieldwork has advantages, but may make central monitoring and control more difficult. Optimal comparability should be pursued as one of the

quality criteria of survey quality. The first steps should be to pursue high quality and to minimise unnecessary variation. The second step should be to fully document the entire survey process to assess to what extent the aims have been met.

Strey, J. P. (Centre for Gender Studies, SOAS, University of London) “You’re a white chick, what are you researching?”: Reflections on using creative methods to counter researcher fatigue “You’re a white chick, what are you researching?” This question highlights my lived experience while conducting my PhD fieldwork on queer women in Bombay and Bangalore, India. The relationship between my positionality as a white American and my purpose made me just one of many ‘Western researchers’ doing projects on the queer community in India. My initial attempts at contact were met with hesitancy, scepticism, and even derision, if there was even a response at all. This refusal is, I believe, due to ‘researcher fatigue’ where the groups and individuals are wary of researchers, especially Western researchers, for a variety of reasons such as lack of follow through, ignorance of issues, and misrepresentation of subjects. In the end, many of my participants told me they only agreed to take part in my research because I was ‘doing things differently.’ That difference is most notable in my use of reflexive photography to visualize and contextualize the issues my participants spoke of. This project involved asking individuals to engage in at least two weeks of reflexive photography in which they thought about how their gender and/or sexuality was lived in their everyday lives. Then, they were asked to take pictures of those moments in which their gender and/or sexuality was most present. Finally, an exit interview was conducted in which the participant described not only why the photo was taken but also what relationship the image had to their lived experience of their gender and/or sexuality. The photos are meant to be seen alongside quotes from the exit interviews, creating both a visual as well as textual experience for the audience to highlight the lived reality of these participants. In this paper, I discuss how using creative, visual methods came to counteract the researcher fatigue felt by many in Bombay and Bangalore, becoming not only a source of visual representation of my participants but also the foundation of my fieldwork. I will discuss my reflexive photography project, including both ethical and practical considerations. Further, I will focus on participant response and involvement highlighting tips and advice on using this method. Lastly, I will end with a discussion of my future goals for the project including how to present these images, putting on exhibitions, and using this work in my PhD thesis. Keywords: Queer, methodology, photography

Struminskaya, B. (GESIS - Leibniz Institute for the Social Sciences) Systematic review and a meta-analysis of studies on panel conditioning Panel conditioning is defined as a learning effect of changing attitudes, behavior or the way they are reported as a result of participation in previous panel waves. Conditioning in panel surveys poses a potential threat to data quality as it can result in measurement error if panel respondents provide different answers than would the persons who were not previously interviewed. An unsystematic look at the available studies on panel conditioning suggests an incoherent picture: in some studies, panel conditioning is not found, in others, it is limited to certain question types. Sometimes panel conditioning appears advantageous for data quality, other times it is disadvantageous. The reason for these seemingly contradictory findings about the direction and magnitude of panel conditioning is in our opinion primarily due to the lack of systematic syntheses that would take the inferential quality of past studies into account. Since survey data is used, the studies on panel conditioning are designed as field experiments (e.g., panel members are randomly assigned to questionnaire content for varying number of waves), quasi-experiments (e.g., panel responses are compared with responses from a fresh cross-section) or correlational studies (e.g., comparing groups with various tenure). In field experiments and quasi-experiments, panel conditioning can be confounded with prior group differences and systematic panel attrition. Therefore, researchers employ matching techniques and other statistical approaches controlling for ex-ante differences and attrition effects into account. The aim of our study is to systematically identify and review existing empirical evidence on panel conditioning, combining quantitative synthesis in the form of a meta-analysis and a qualitative synthesis for studies that do not report indicators that can be converted into effect sizes. First, we identified relevant studies using the following study eligibility criteria: studies that report empirical evidence, contain a description of the sample used for analysis, are administered throughout at least two interviews, and use quantitative methods. Next, we conducted a literature search by using bibliographic databases such as ISI Web of Knowledge, ScienceDirect, Directory of Open Access Journals, EBSCOhost, Emerald, Ingenta select and Google Scholar. We used the following keywords and keywords combinations: panel, conditioning and survey; panel bias, conditioning bias, time in sample, rotation group bias, re-

interview effect, respondent conditioning, mere measurement, and measurement reactivity. After applying our eligibility criteria and additionally excluding studies with self-prediction, studies that concentrate solely on recall, clinical trials, and studies with a focus on induced deliberation, about a hundred studies were identified that comprised the analysis set. Second, we classified the studies according to the type of panel conditioning (advantageous vs. disadvantageous), their designs (experimental vs. correlational), and the availability of the true value (true value known, direction of effect specified, direction unspecified). We further include the following variables that can moderate the strength of panel conditioning effects into our qualitative synthesis: interview mode, survey frequency, amount of previous exposure, question difficulty, question sensitivity, and question type. In the next step, we identify studies that report measures on the basis of which effect sizes can be calculated and carry out quantitative meta-analysis including these studies. Finally, we assess the extent of the possible selection bias that can arise from excluding studies that do not report the measures necessary to carry out quantitative meta-analysis. We conclude with the discussion on the usefulness of alternative calculations of effect sizes (e.g., proposed for panel conditioning by Cantor 2008) and implications for reporting results in future panel conditioning studies.

Szabó, J. (Corvinus University of Budapest) Youngsters' leisure time and social stratification of cultural consumption at an Eastern European festival The paper presents youngsters’ leisure time preferences and the social stratification of cultural consumption in Romania. Festivals provide a great opportunity to get to know the culture of those young people who play an important role in the scenario of leisure time, especially a segment of the middle class, more open and more active than young people in general - who are leaders of opinion among youth, provide models, representing youth of leisure. We present the factors that characterize leisure time patterns, cultural consumption and communication habits of young people in Romania. Since festivals play an important role in the life of youngsters, the festivals are prime places to study the prolongation of the “youth- period”. The empirical research and the data analysis was conducted in the context of Jürgen Zinnecker’s youth epoch change theory which offers the necessary framework for analyzing the intersection between global challenges that young people are facing and their local experiences shaped by the fall of communism and the regime change. In our main hypothesis we test the validity of` `omnivore and univore` arguments discussed by Chan and Goldthorpe (2005,2006,2007) During our analysis we rely on survey data collected at Peninsula Festival in Târgu Mureş between 2006-2013. The used methodology had two components. One of the components was a one page contact form which, among others, helped us gather information about the demographic characteristics of the participants. In parallel – the second component – was a longer questionnaire comprising of several topic-blocks. (The fist component helps us check and correct the data obtained from the second component) Both data-gathering techniques were administered in the whole duration of the festival. The interviewed participants were selected based on stratified random sampling, both territorial and time-wise. The analysis shows that young people who attend youth festivals, particularly the Peninsula Festival in Târgu Mureş are mostly 18-29 years old, they are coming from urban areas, mainly from municipalities, they are students or with higher education studies. Thus, their material situation is better than the national average. The results show that this type of cultural

consumption, of participation at youth festivals, is accessible to young people from urban middle class in Romania.

Tayactac, J B., Gerardo A S., and Guerrero, L. Quality Control Challenges to Modernized Scientific Social Weather Reporting in the Philippines Conducting scientific polling given an archipelagic geographical set-up like the Philippines is indeed very challenging. Over the years, local survey research institutions like Social Weather Stations (SWS, www.sws.org.ph) have developed and successfully implemented procedures for conducting scientific polling in the country. Identifying nationally representative respondents typically involves multi-stage sampling with probability proportional to population size to select provinces, cities/municipalities and barangays or villages, and within sample barangays systematically select dwelling units and randomly select a qualified respondent within the sampled household.

The most common interview mode in the Philippines is the face-to-face paper and pencil interviews (PAPI). Field interviewers receive standard training on fieldwork procedures and questionnaire implementation, and this approach has ensured accurate collection of quality data, as proven by the surveys’ success in predicting various election outcomes that led to the wide acceptance of opinion polling in the country.

Given the increasing demand to produce results at a much faster pace, research institutions like SWS have begun testing and/or using computer assisted personal interviews (CAPI). Programs associated with CAPI typically permit the conduct of interviews whether or not the computer is connected to an internet. And when internet connection becomes available, responses stored in the computer are immediately transmitted to a secure server with processed data readily accessible to the researcher.

While CAPI significantly shortens the time from data collection to data processing, this platform has certainly skipped many quality control procedures commonly applied in PAPI. In PAPI, quality control measures employed by supervisors during fieldwork include review of completed questionnaires, direct observation of interviews, spot checks (surprise checks done

by supervisors while the interviewers are still conducting interviews in the area) and back checks (checks done by supervisors after the interviewers left the area). In using CAPI, there are no more hard copies of questionnaires for supervisors to review and edit, and data obtained rely only to those collected by the CAPI interviewer. Spot checks and back checks as quality control procedures can still be done by supervisors, but again they do not have copies of completed questionnaires to verify answers obtained from respondents.

The use of CAPI also skips quality control procedures in data processing typically applied in PAPI. With PAPI, completed interviews undergo office editing, coding and encoding, validation of encoded data or the use of double encoding, and data reconciliation. With CAPI, there are no more sheets to edit and compare encoded data because the responses obtained from field are automatically encoded when transmitted to a server.

While programs installed in CAPI typically have range of acceptable answers, skipping, branching and filtering questions, rotation and randomization of questions and response options, and even added features like GPS tracking and use of photos, sound and videos, the general observation during the SWS March 2015 CAPI experiment is that the accuracy and quality of data collected rely primarily on the hands of the interviewer conducting the CAPI. SWS has not yet been able to resolve the quality control issues in using CAPI. Keywords: quality control, face-to-face interviews, computer assisted interviews, Philippines

Teti, A. and Gross, C. (Institute of Medical Sociology, Charité – Universitätsmedizin Berlin; Institute of Sociology Leibniz University Hannover) Consistency of Vignette Judgments among Older Respondents Background: The factorial survey (FS) method is increasingly used in the social sciences. It is particularly suitable for studying events or decision situations that are difficult to assess empirically. Previous FS studies have drawn either on homogeneous samples of young, well- educated university students or on general population samples; very few older respondents have been included. This article evaluates whether the FS method is suitable for studying decision situations in gerontological research. Methods: This paper draws on data from the HOME study, which investigated residential mobility in old age. Specifically, n = 103 respondents aged between 55 and 90 years were asked to make hypothetical relocation decisions. The consistency of these responses was assessed as a function of respondents’ age, gender, immigration background, education, household income, employment status, and intention to move. Results: No significant differences in response consistency were attributable to age, gender, education, or immigration background. Response consistency increased with income level, although employed participants gave less consistent responses than did participants who were not in paid employment. The personal relevance of the survey topic was also positively related to response consistency. Conclusion: The main finding is that the FS method can, under certain methodological conditions, be used among older adults. Very old adults, who are otherwise often excluded from complex assessments, can thus indeed be surveyed by means of the FS method, especially when the topic of the survey is relevant to their daily lives. Keywords: Factorial Survey, Elderly, Respondents' effects

Thomas, R K., Barlas, F M., Buttermore, N R., and Smyth, J D. (GfK Custom Research; University of Nebraska) Acquiescence Bias vs. Salience in Response: An Experimental Test Elements in Yes-No Grids are endorsed at higher rates than when they occur in a Multiple Response Format (i.e., select all). Thomas and Barlas (2015) presented evidence that this may be due to increased consideration of each element. While asking about a set of elements, only the most accessible or important will be selected in the Multiple Response Format, asking for a response for each element to be considered with the Yes-No Grid will increase the access of less proximal memories (the salience hypothesis). Alternatively, Callegaro et al. (2015) proposed that acquiescence bias most likely explained heightened endorsement. Acquiescence bias results from socialization that encourages people to be agreeable, which can lead to a greater tendency to endorse ‘agree’ with an agree-disagree response format or select ‘yes’ in a yes-no choice format. This study tested the divergent predictions of the salience and acquiescence bias hypotheses. We had 1,629 KnowledgePanel (probability-based online panel) respondents in our study with a 2 X 5 factorial design – respondents were randomly assigned to a rating dimension (either a descriptive or agreement dimension) and to one of five response formats. Each participant rated four brands with which they had some familiarity along six different attributes (e.g., ‘Is distinctive’, ‘is expensive’, etc.). Those assigned the descriptive dimension were asked if the attributes described each brand, with response formats being either: 1) a multiple response format, 2) dichotomous format (‘Describes, Does not describe’ or ‘Yes-No’), or 3) trichotomous format (‘Describes very well; Describes somewhat; Does not describe’; or ‘Yes, very well; Yes, somewhat; No, does not’). The agreement dimension asked whether the respondent agreed that the attribute described the brand, and had comparable response formats to the descriptive dimension (e.g., ‘Agree, Do not agree’; ‘Yes, agree, No, do not agree’, ‘Agree strongly, Agree somewhat, Do not agree’, etc.). Our expectations were that if acquiescence bias was correct, responses with a ‘Yes’ or an ‘Agree’ in them would be associated with higher levels of endorsement. We did not find any support for that. We discuss the results in terms of the salience hypothesis whereby increased attention increases the search effort somewhat and more distantly related elements are endorsed (whether more weakly associated with the attitude target, or more distant events or less extreme events).

Keywords: Acquiescence bias, Check all format

Thomsen, U. (Institute for Employment Research, Germany) Filling up the gap: Imputation of missing Working Time Information in the German Employment Data The German Federal Employment Agency (BA) stores a large variety of administrative data covering various aspects of the German Labour Market. The data are generated as a by- product of different administrative processes and institutions, reaching from Social Security to Unemployment Insurance or Unemployment Benefit II. These administrative data build the data basis for the official statistics of the statistics authority of the BA, but also for the generation of research data. They are used by the Institute for Employment Research (IAB) to create a wide-ranging micro data base for socio- scientific labour market research. One important part in that data are employment data. They are derived from employer’s mandatory notifications to the statutory health and the pension insurance. The notifications comprise central information of the employment, e.g. beginning and ending date, remuneration, occupational classification, industry classification of the permanent establishment and the job activity profile (Taetigkeitsschluessel). In 2011 the collection method for the variables comprising the job activity profile was changed. As one result of that change the share of missing information on the job activity profile in the employment data increased temporarily to roughly 25 percent. Since wages are meaningful only together with working time information in most applications, these missings reduce the practical value of the data severely. The change of the data collection method also revealed systematic reporting errors in the working time information (employers miss to update the working time information after changes of the working time in continuing employment relationships). Understanding these errors is a crucial prerequisite and ingredient for scientific analyses that are based on these data. We assess the consequences of these errors on wage analyses and develop a simple model to impute the missings. To this aim, we compare the predictive power of Logit, linear probability models, Categorization Trees and nonparametric probit models. Considering the huge size of the data, the validation of the selected model poses a great challenge. For example, leave- one-out cross validation is computationally infeasible for the entire data set. Nevertheless

data quality has to be assured for the imputed information. Some of the tests performed to this end are presented and should be discussed. The resulting imputed information are integrated and marked in the research data. The data are available in the Sample-of-Integrated-Labour-Market-Biographies. These data are subject to the comparatively strict German data protection legislation and might not be passed to third parties except for certain research purpose. Data access is provided by the Research Data Centre (FDZ) of the German Federal Employment Agency at the Institute for Employment Research. The data are suitable for analyses in different disciplines like economics (e.g. econometric analyses), sociology (e.g. research on employment biographies, cohort analysis) or regional science. The FDZ provides access to the data base for third parties.

Thomas, R., Barlas F M., and Buttermore, N M. (GfK Custom Research) Effects of Scale Type and Response Style Differences across Cultures and Countries Response categories may be used differently as a result of ethnic background or country of residence (Baumgartner and Steenkamp, 2001; Chen, Lee, and Stevenson, 1995; Hofstede, Steenkamp, and Wedel, 1999). Many researchers believe that respondents from some countries are less likely to use extreme response categories while those from other countries are more likely to use them. When making comparisons between countries, so that we do not confound culture/country with other factors and before we can attribute differences due to culture and not other factors, we must ensure that: 1. The comparisons do not also involve comparisons between modes (cf. Dillman & Christian, 2005) 2. The countries are equated for all known variables 3. The differences in familiarity or knowledge of the topic are equated. 4. Scale polarity is the same across the countries – bipolar scales are sometimes inappropriately rendered as unipolar scales (and vice versa) in translations. In this study, we compared different scale variants (e.g. unipolar versus bipolar) and extent of semantic anchoring (fully anchored scales give a semantic label for each response while end anchored scales provide only the extremes of the scale) in leading to differences between countries. Method As part of an international web-based survey, this experiment had 36,938 respondents from 6 different countries (Belgium, France, Germany, Italy, Spain, UK) complete a section on attitudes and behaviors concerning a number of different activities. Respondents were randomly assigned to 1 of 4 scale conditions to rate in terms of their liking for the activity: Fully anchored Unipolar (Do not like at all-Like very much) Fully anchored Bipolar (Dislike very much-Like very much) End anchored Unipolar End anchored Bipolar To examine the effects of respondent load on response styles, respondents were randomly assigned to evaluate either 5 or 10 topics.

We found significant differences in endorsement proportions for the response categories as a function of scale type. The fully anchored unipolar scale showed a lower endorsement of the highest response categories across countries. There were mean differences in the evaluations of the issues as a function of country. However, controlling for familiarity with the topic and demographic factors, we found that these differences between countries was eliminated or reduced for most of the activities we examined. We found significant differences in response style proportions (extreme versus middling response styles) as a result of country of residence. However, response style differences between countries were more likely to occur for the end-anchored scales and not for the fully anchored scales. We discuss the findings in terms of factors that need to be controlled for before concluding that response styles are real phenomena.

Thomas, R K., Barlas F M., and Buttermore, N M. Toward Better Grid Formats for Online and Mobile Surveys In self-administered surveys, grids have evolved to efficiently measure multiple concepts using the same graded response format across a number of different targets, with responses typically in columns and the rating targets to evaluate in rows. Generally, grids allow us to more efficiently assess multiple targets compared to presenting each target is presented on separate screens. The response format that is chosen can vary in terms of semantic labeling - every response can have an associated semantic label (fully labeled) or only the end categories have a semantic label (end labeled). The response format could represent a bipolar scale (ranging from a concept and its antithesis, e.g. like – dislike) or a bipolar scale (with a concept and its absence, e.g. like – do not like). Finally, the response format can vary in terms of the number of response categories used. We summarize 4 studies comparing the use of end-anchored and fully-anchored categories with both unipolar and bipolar scales, holding the number of categories constant. The unipolar fully labeled variant generally appeared to be better as a measure across topics, with better differentiation and validity. Since online surveys are increasingly being taken on mobile devices, the utility of a grid is becoming affected by screen size – grids with more responses are much harder to implement on smartphones. In a series of 2 additional studies we examined how reducing the number of grid response categories affects the measurement of a variety of concepts. We found that response formats with fewer response categories take less time to complete, are easier to complete on mobile devices, and can show as much validity as formats with more response categories, especially when using the unipolar variant. We examine the ability of the scales to detect smaller differences between rating targets and found some improved differentiation with scales with fewer responses rather than those with more responses.

Toepoel, V. (Utrecht University) Adapting Surveys to the Mobile World: Data Chunking in the Dutch Probability-Based LISS Panel Mobile phones are replacing key tasks formally done on PC and laptop. Mobile phones are commonly used for short messaging.We should therefore 1) move to mobile or mobile- friendly surveys; 2) find ways to shorten questionnaires (most surveys are too long for mobiles). A way to reduce survey length is data chunking: offering questionnaires in smaller pieces. Data chunking a.k.a. modular survey design is a way to cut-down long survey questionnaires. While data chunking is not new to the survey world (see Johnson, Kelly & Stevens, 2011), there exists no systematic research in how data chunking is related to Total Survey Error and mobile surveys. There are several ways to do this: ‘across respondent’ modularisation, whereby different respondents take each piece, and ‘within respondent’ modularisation, whereby the same respondent is permitted to take pieces of a survey at different times. A longitudinal survey in the probability-based LISS Panel of CentERdata is divided in several experimental groups to investigate data chunking within respondents: panellists who own a mobile phone with Internet connection are randomly assigned to a) normal survey length, b) survey cut in 5 pieces, c) survey cut in 10 pieces. In addition, we experiment with (push) notifications via mail and SMS. We investigate the number of complete and incomplete responses and look at indicators for data quality (straightlining, primacy, survey length, satisfaction with the survey etc.). In addition, we use multiple imputation to investigate if the imputation could increase data quality to a level equal to the full survey. We benchmark our data with other panel variables, including previous waves of the core questionnaire we used. This paper is highly relevant in times were response rates are falling and survey length should be kept to a minimum. Keywords: data chunking, mobile surveys, panel surveys

Tome, R S., Roberts, C., and Joye, D. (University of Lausanne) The Impact of Mode Effects on Measures of Subjective Well-being The past decades have seen growing interest in alternatives to traditional measures of economic prosperity and societal progress. Of these, indicators of subjective wellbeing (SWB) in different parts of the world are particularly popular. Much of the produced research is based on survey data at a time in which traditional survey modes are insufficient to maintain data quality and costs. One of the most widespread alternatives to overcome difficulties is to use various modes of data collection but as a consequence, doubts have been raised related to the comparability of data collected through different modes. This applies to when data comes from a mixed mode survey, but also it is also a problem when analysing data from different countries whose data have been collected through different modes. In our previous study, using data from a mode comparison experiment, we compared survey estimates on the topic of wellbeing by mode (telephone, web and paper) and found measurement and selection differences between the modes of data collection, measurement effects were specifically found on widely used measures such as happiness and job satisfaction. Motivated by this finding, in this study we use a multivariate model to assess variation between the different modes of data collection and analyse the relationship between subjective wellbeing variables and their predictors. In particular, we use confirmatory factor analysis to check the construct validity of two measurement models, one for personal subjective wellbeing and one for job satisfaction. In addition, we implement multigroup structural equation modeling to understand the relationship between the variables and measurement errors. Results indicate overall consistency in both personal wellbeing and job satisfaction for the different modes of data collection, but we do find some variation for some individual measures. In addition, we find that measurement effects on some individual variables included in the analysis do not have a significant impact on the statistical relationship between variables of interest. Keywords: well-being, mode effects, construct validity

Tormos, R., Rudnev, M., and Bartolomé, E. (University of Barcelona) Justifiability of Euthanasia in OECD Countries: Patterns of Change and Age-Period- Cohort Analysis We study the justifiability of euthanasia from a comparative and longitudinal perspective, using EVS/WVS data from 30 OECD countries from 1981 to 2009. Accounts of attitudes toward euthanasia commonly have been longitudinal (Cohen et al., 2006a), comparative (Cohen et al., 2006b, 2014), or focused on individual-level explanations of single-country cases (e.g., van der Maas, 1996; Kemmelmeier et al., 2002; Burdette, Hill, and Moulton, 2005). This research contributes to this literature by combining these approaches within a single framework emphasizing the potential effects of age, period, and cohort on attitude changes. We study age–period–cohort effects on the justifiability of euthanasia across nations. We first descriptively explore period changes, intercohort differences, and intracohort changes across countries and over time. Second, we concurrently model age, period, and cohort effects by applying two novel extensions of multilevel analysis: 1) the dynamic comparative multilevel approach (Fairbrother, 2014; Tormos, 2013) and 2) a comparative version of cross-classified random effects models (Hayward and Krause, 2015). Our main gain is to understand the nature of the changes that occurred over this period of time (1981–2009). We find a clear increase in tolerance of euthanasia across affluent countries. First, the change occurred between generations. This is in line with modernization theory, which holds that formative experiences shape the views of cohorts and that social change occurs progressively due to generational replacement (e.g., Inglehart and Welzel, 2005). However, the degree of change that we observe in the data is unlikely to have come from cohort replacement processes alone. We identify clear trends related to linear period effects which are similar across countries when controlling for age and cohort effects. These trends might be connected to the overall processes of human development and emancipation (Inglehart and Welzel, 2005; Welzel, 2013). Aging, inversely, is demonstrated to have a negative effect on the justifiability of euthanasia. Our results point to the presence of cohort, age, and period effects that produce intracohort changes.

Tranmer, M., Marcum, C., Morton, F., Croft, D., and de Kort, S. (School of Social & Political Sciences, University of Glasgow (Presenter), Chris Marcum - National Human Genome Research Institute, National Institutes of Health, MD, F. Blake Morton - Franklin & Marshall College, PA, Darren P. Croft - Psychology, University of Exeter, Selvino R. de Kort - Manchester Metropolitan University) Using the Relational Event Model (REM) to investigate the temporal dynamics of social networks. Social interaction event data are often aggregated into a single network within a particular time frame - e.g. a network of interaction events over the past month. Analysis of the resulting network can provide a useful insight into the overall extent of interaction in that time frame. However, through aggregation, information is lost about the order in which the interaction events occurred, and hence the sequences of actions of these events over time. Many research hypotheses relate directly to the sequence of events, rather than to their overall volume or presence. Here, we demonstrate how the temporal dynamics of social interaction sequences can be quantified from disaggregated event data using the relational event model (REM). We first outline the REM, explaining how it can be used to model sequences of events unfolding in a network. We then discuss a case study on the European jackdaw, Corvus monedula, in which temporal patterns of persistence and reciprocity of action are of interest, and present and discuss the results of a REM analysis of these data. One of the strengths of a REM analysis is its ability to take into account different ways in which data are collected. We provide details of how the models may be fitted in the R statistical software environment, and outline some recent extensions to the REM framework.

Trübner, M. (University of Bonn, Department of Sociology) Header Images in Web Surveys – Curse or Blessing? Visual elements such as header images are typically included in web surveys to make surveys more appealing for respondents. However, design might also induce a systematic bias in response behavior. In order to examine potential attention-seeking effects as well as context effects of header images on different devices, a web survey experiment based on a representative sample of 1,298 students was conducted at the University of Bonn. Respondents were either confronted with a picture of an auditorium with students sitting in a lecture, a picture of leisure activities on campus, or no picture, respectively. To control for position effects, pictures were placed either on the right or left top side of the questionnaire. As both image conditions are university-related, it is assumed that they lead to an assimilation effect that manifests in higher estimates of time spent at the university as compared to the control group, however, at a different level. Further, it is hypothesized that respondents who are confronted with the images are more aware of their time spent at university, which induces a contrast effect in the evaluation of their satisfaction with time spent for other activities. It is hypothesized that computers and laptops are more prone to these effects compared to smaller mobile devices like tablets and smartphones. Keywords: web survey, context effects, device effects

Tuvikene, T., Neves Alves, S., and Hilbrandt, H. (Tallinn University) Fast comparisons for urban studies? Writing across Bafatá, Berlin and Tallinn Urban comparisons have often implied large research projects with systematic detailed investigations of similarities and differences, or, alternatively, in the recent surge of postcolonial comparative urbanism they have meant in-depth case studies. However, in that way they have remained either unattainable for less-resourced researchers or too singular in their comparative framework. This paper advocates for an alternative approach in which multiple case-specific explorations carried out by individual researchers are brought together in one singular comparative research project. We term this comparative strategy a multi-sited individualising comparison. Responding to the urge of postcolonial research to combine the insights of north and south, we developed writing across three widely diverging contexts of Bafatá (Guinea-Bissau), Berlin (Germany) and Tallinn (Estonia). Bringing together cases and the literature, this strategy allows for comparison of cities as distinct as ours, and generates conceptual insights that challenge some of the taken for granted assumptions of incommensurability between cities. We also discuss the ways in which the proposed approach allows to bring together different empirical cases such as water governing in Bafatá, housing in Berlin and car parking in Tallinn. In our approach, the comparison is done through ‘statements’ that should hold across diverse sites. That is, the formation and clarification of statements occurs by bouncing a statement back and forth across research sites. This paper thus works to advance more applicable approaches of comparisons that can be carried out by just bringing together different researchers in conversations that, moreover, could be carried out over distance without the necessity to go in-depth into each other’s research projects.

Turhan, Z. (Goldsmiths, University of London) Conceptual framework and reflexivity for the examination of the engagement process in a sensitive qualitative research The purpose of this research is to examine the perspectives of perpetrators on their engagement and the programme providers’ experiences and involvement with perpetrators during the domestic violence intervention process. The core objective is based on the premise that perpetrators’ perspectives influence their participation, engagement and the completion rates of programmes. This paper starts with an overview of the conceptual framework which provides philosophical assumptions and guides the methodological selection. Then, the paper briefly discusses the theoretical basis and strengths of choosing the qualitative research approach and the reasons for choosing grounded theory and feminist research approaches. This paper also clarifies the benefits of the conceptual framework and analytical literature review in the examination of participants’ process of engagement in a sensitive topic. However, methodological concerns are also likely to be raised in this sensitive research including the need for a long term relationship with participants. Although this research does not aim to detail individuals’ biography or long term life history, the examination of perpetrators’ engagement in three stages in an intervention is a long enough process because participants will be male perpetrators. Furthermore, potential difficulties include the insufficient amount of time to follow up perpetrators’ engagement in an intervention process; their unwillingness to participate in research; the frequency of drop outs in the programmes; and the experiences of denial and minimisations of their violent behaviour. Within these obstacles in researching their unique perspectives and experiences of engagement, this paper examines how the grounded theory approach may help to analyse this complex process of engagement in an intervention programme. In presenting the research approach, the paper also discusses the potential strategies that may increase individuals’ participation, build rapport and reduce power relations for the successful collection of data and analysis by providing reflexivity.

Ueblacker, J. (Research Institute for Societal Development, Research Area “Integrative Urban Development”; University of Cologne, Institute for Sociology and Social Psychology) Mixed Methods-Mixed Research Synthesis in Urban Sociology: The Example of Gentrification Research in Germany The socio-spatial foundation is the defining characteristic of urban sociology. Theories and models are not only describing the city as whole but also areas of smaller scale such as neighbourhoods. Most of the questions posed in neighbourhood research can only be answered using individual level longitudinal data, which in Germany is only scarcely available, if at all. A gap between theoretical models of neighbourhood change and the possibilities to adequately test these models exists. An example is the research on gentrification. It mostly consists of single case analysis of neighbourhoods. These apply a wide range of different quantitative and qualitative methods, often within the same study. The methodological challenge in synthesizing this research therefore lies in the variety of data types. These range from official register data, neighbourhood surveys, qualitative interviews, structured observations to ethnographic approaches. Asking for the contribution of an until now unassessed part of research I conduct a systematic review of the grey literature on gentrification in Germany since 1980. The search strategy contains comprehensive library search, search in online-publishing portals for students’ thesis, snowball search via retrieved bibliographies and contacting about 40 scientific peers. The selection is done via a key word list and analysis of tables of content of the retrieved studies. According to Sandelowski et al. (2012) a synthesis by configuration is applied. It uses a flexible conceptual framework that is partly derived from the already known research in this area. For the actual process of extracting and synthesizing information the method of meta-study by Paterson et al. (2001) was chosen. It allows to distinguish between the synthesis of theory, methods and data. The empirical findings of each study are first extracted to the according categories in the conceptual framework. Second, a cross-study comparison is conducted within

the categories of that framework. A similar approach is suggested by Miles and Huberman (1994). At the moment information from 71 studies was extracted and partly synthesized. The contribution of this research is twofold. On the methodological side, it gives an example for the application of a mixed methods-mixed research synthesis, a field that still is in its beginnings. Furthermore, it is exclusively focussing on the grey literature, a branch of academic output that on the one hand suffers from great quality fluctuations and on the other hand promises to contain innovative insights that usually would not be appreciated. On the side of urban sociology, it opens the doors for further application of systematic reviews by giving a first example with the synthesis of a methodological heterogeneous research branch, that seems to be stuck in its case logic without being able to further develop its theories on neighbourhood change. Keywords: Mixed Methods-Mixed Research Synthesis, Gentrification, Urban Sociology Literature Miles MB, Huberman M. Qualitative data analysis: an expanded sourcebook. London: Sage, 1994 Paterson BL, Thorne SE, Canam C, Jillings C. Meta-study of qualitative health research. Thousand Oaks, California: Sage, 2001 Sandelowski, Margarete; Voils, Corrine I.; Leeman, Jennifer; Crandell, Jamie L. (2012): Mapping the Mixed Methods-Mixed Research Synthesis Terrain. In: Journal of mixed methods research 6 (4), S. 317–331

Unfried, M., and Buder, F. (GfK Verein) Increasing Respondents’ Engagement with Incentivized and Implicit Interviews: Learnings from Behavioral Economics Latent variables like trust or brand image are widely used in explaining consumer behavior. As these constructs are not directly observable, multi-item scales (or grid questions), sets of consecutive text statements using the same response scale, are a common instrument. However, the use of grid questions is controversially discussed especially in terms of respondent engagement. This contribution evaluates how simple behavioral economics games can produce similar information about subjects’ attitude and provides insights into how these games can engage respondents. The data for this contribution were obtained from a two-part survey study originally designed to analyze perception and behavior of Europeans about their neighbor countries. To this end, 1015 participants from five major European countries participated in the study (about 200 each). The sample was representative in terms of age, gender, education and territorial distribution. Participants had to play two simple economic games with monetary incentives: a version of an honesty game and a real effort task. Subsequently to each game, the participants assessed the average behavior of the other participants per country to measure the participants’ attitudes towards people from all countries. The third part of the study was a standard questionnaires consisting of several multi-item scales designed as question grids about the perception of behavioral and cultural aspects of all countries included and the respondents experience with the other countries. In the volunteering game, the respondents had to count a certain symbol in a table. For each table counted correctly respondents earned EUR 0.5 for a charity organization. After playing the game themselves, the respondents had to assess how many tables the participants of each country counted correctly on average. With the assessments, respondents were able to earn money for themselves (at most EUR 1.50) depending on the accuracy of their assessment. In the honesty game, participants were asked to toss a coin and report the result. If they reported tails they received EUR 1.00, if they reported head they received nothing.

Participants of a certain country are supposed to behave dishonest on average if more than 50% report tails. Subsequently, the respondents again assessed which percentage of participants in each country reported tails. The closer their assessment was to the actual percentage the higher their payoff (EUR 1.50 at most). The data were used to compare games and scales. Firstly, we checked if grid questions and games yield similar results. Secondly, data from the grid questions were analyzed with respect to indications of lower respondent engagement. Grid questions and games seem to yield comparable results on average. Games, however, seem to have three advantages: i) they seem to be more precise as assessments show stronger discrimination between countries than questions, ii) they don’t lead to decreasing engagement over time, and iii) payment can be designed incentive compatible. For questions, it is only possible to reward respondents’ participation, not thoughtful responses. In the applied games, respondents’ answers determine their individual payoff motivating them to give substantial responses. Keywords: behavioral economics, incentive compatibility, respondent engagement

Unger, S. (Institute for Employment Reserach (IAB) Measurement invariance between employed and unemployed persons - An example using a multi-item health measure Surveys often rely on subjective measures for various concepts such as wellbeing, health or happiness. For these measures it is important that different groups of the population understand the measured concepts and items in the same way. Otherwise conclusions drawn from analyses may be misleading. The SF-12 is a widely used scale to measure health-related quality of life. It consists of 12 questions covering both mental and physical health and health restrictions. It has been used by former research to assess the effect of unemployment on individual health. However, there is some reason to question the validity of the SF-12 for the comparison of employed and unemployed respondents because of some critical questions which refer to work or other regular activities. Employed persons who feel restricted at work due to their health may rate their health lower than unemployed persons of the same health status. While there is research on the intercultural equivalence of the SF-12 (Schulz, 2012), there exists no previous research testing if the SF-12 is equivalent across groups with different employment status. I assess, whether the SF-12 is suited to analyse health effects of (un- )employment using the large German Panel Study Labour Markets and Social Security (PASS) which consists of a benefit recipient sample and a sample from the general population. The household panel has been conducted since 2006 and covers over 8,000 respondents each year. The overrepresentation of benefit recipients who are most often unemployed, allows detailed comparisons between employed and unemployed persons. Assessing measurement invariance between employed and unemployed persons is of utmost importance because it is a prerequisite for drawing correct (causal) conclusions. I use the CAF-framework in order to tackle equivalence of health measures for employed and unemployed persons.

Vacchiano M., Martì J. and Yepes, L. (Universitat Autònoma de Barcelona) The use of social networks in the age of uncertainty. Young adults’ strategies along the job search. In the last two decades, several authors have shown how global changes had a significant impact on life and work experiences, especially for those individuals whose social integration is still ongoing: the young adults (Beck, 1992; Giddens, 1999; Furlong, 2012). The effects of the economic crisis, framed in a more liberalized and flexible capitalism, have generated labour pathways composed by episodes of unemployment, precariousness and occupational segregation for this social group, converting the job search in a process characterized by a broader range of uncertainties compared with previous generations. Nowadays, individuals born between the end of the seventies and the early nineties, that is, the generation also known as Millenials, Digital Natives or Generation Y, are facing their transition to work responding to this increasing uncertainty through a variegate number of individual adaptations and strategies. We claim that for this cohort the use of kinships, close friends, colleagues and in general, personal networks, represent an informal channel of access to labour market that fits with the unpredictable course of labour pathways, nowadays less standardized, with rapid shifts between jobs and education experience. Although the role of social capital in the labour market has been widely investigated in the last decades, we believe that the new context of economic and social transformation, coupled with the economic downturn, has shaped an unprecedented scenario which force young adults to plan quick strategies to face the increasing uncertainty, making the characteristics of social networks – informality, rapidity, closeness – more visited and central along the job search process (Requena, 1991; Lin, 2011) Considering this assumption, the aim of this presentation is to shows empirical data collected in order to analyse the specificity of these strategies among young adults in Catalonia. Particularly, we focus on the use of social networks along the job search, analysing how frequently young adults use contacts as method of job search comparing with other methods. Plus, we also investigate differences in social capital mobilization strategies, related with

gender, education, economic status, (use of kinships, workmate, schoolmate, neighbours, acquaintances), focusing in the mutual interaction between a certain use of personal networks and the characteristics of the personal labour and educational pathway. Therefore, we plan to explain the process of data collection, underlined advantages and difficulties derived by the use of several methods; precisely, we use both Name Generator to collect personal networks and a life-grid drawn by the interviews to collect the educational and labour pathway.

Valentova , M. (LISER Luxembourg) Did the escalation of the crisis of 2008 affect the perception of immigration-related threats? A natural experiment. The paper applies a natural experiment approach comparing observations of the European Value Survey data prior and after the escalation of the financial crisis in 2008 to disentangle the short-term causal effect of the crisis on the relationship between social vulnerability and threat perceptions on residents of Luxembourg (both Luxembourg as well as foreign- nationals). Propensity score matching is used to ensure the balanced composition of the control and treated group with respect to the observed covariates. The results show that vulnerable foreign-nationals in Luxembourg demonstrated higher perception of job related threats after the escalation of the financial crisis than their pre-crisis counterparts. Among Luxembourg nationals, no significant effect of the escalation of the crisis was found. Key words: perception of immigration-related threats, financial crisis, social vulnerability, quasi-natural experiment

Valér, V. (Babes-Bolyai University Cluj, Romania) The Quality of Online Survey Data Using Prize Draws Online research may contain several methodologically-induced effects. The present paper analyses the representativeness of the self-selection of respondents as compared to the composition of the population, as far as substantive and methodological aspects is concerned. We will analyse the question from two perspectives. The first one is the formal structural question of representativeness, i.e. if the composition of the sample reflects the social- demographic composition of the population. The other question is that – although survey data contain a mixture of substantive and methodologically-induced variations – to what extent the online self-selection of respondents distorts substantive variation as far as the essential problems raised by our research are concerned, especially if we have applied prize draws in order to motivate answering (by ulterior prize draws). Our research question specifically focuses on the question if the different motivations behind online answering may exert certain unexpected effects on the composition of the sample which represents a significant change in methodological variations as compared to variations in real, substantive answers among the population. We consider that this question is important because based on the already known social and demographic composition of the population we may not always see a distortion of the sample, but the ability to motivate respondents may have certain aspects related to personality, or perhaps financial implications such as those triggered by offering prizes. For our empirical analysis we have used the population of a Romanian online research among students from Cluj-Napoca. The sampling process was carried out in June 2015, with 3,732 valid cases, representing the valid student answers. The research was carried out by Babes- Bolyai University (Sociology and Social Work Faculty) and Mathias Corvinus Collegium. The sample contains approximately 10 percent of the student population, according to a representative distribution by faculties and ethnicity. During the preparation of our research, we have made use of internal electronic registries of the universities for data collection purposes. Based on mailing lists, theoretically more than 90% of the entire student population was requested to fill out the online questionnaire, which basically required ten minutes. In order to motivate the students, we drew different prizes and gifts.

The research questions focused on: to what extent the choice for a certain programme is being shaped by social background and ethnicity; is there such a tendency that in their choices students of lower social background are guided towards degrees easily marketable on the labour market, as a more cost efficient solution? The motivations behind vocational choices reflect educational aspirations, but we analyse whether these are based on professional interests or rather on cost-benefit calculations. We have compared the representativeness of the online sample with the results of an offline sampling carried out 4 months later on the same population. We have analysed the substantive and methodological effects based on the variation of answers given to some questions from both data collections. The results are edifying, although we would need to compare more similar online-offline researches in order to draw some final conclusions. Keywords: surveys, online, students, Romania, methodological variations

van Egmond, M., and van de Vaart, W. (University of Humanistic studies, Utrecht) Data quality in asynchronous online focus groups: the directive versus the non-directive moderator style As an alternative to traditional face-to-face focus groups [TFG’s], online focus groups [OFG’s] have been introduced as a qualitative data collection method about 15 years ago. Where TFG’s are less suitable for discussing sensitive topics, it is observed that the anonymity in OFG’s invokes more self-disclosure. Anonymity of OFG’s also enhances more equal opportunities to participate in group interaction. Participants will be less hindered by physical characteristics such as age, ethnicity, appearance and socioeconomic status, which may inhibit exchange in TFG’s. Other advantages of OFG’s include: access to otherwise hard-to- reach groups, a wide geographical spread, possibility for respondents to reflect and nuance and lower costs. While synchronous OFG’s take place real-time in a pre-set timeframe, asynchronous OFG’s are conducted over a longer time period and participants can join in their own chosen time and place. Given their broad range of applications we focus on asynchronous OFG’s. Specific for asynchronous OFG’s participants is that they can reflect on their ideas before they react. The same applies to the moderator, who can reflect on previous questions and reactions before deciding how to intervene. This more delayed interaction could also lead to a more cognitive, reasoned and less impulsive, emotional exchange and a different kind of group process. However hardly any research exists into characteristics of OFG’s and how they relate to data quality. Since the success a focus group largely depends on the role that is assigned to the moderator, we designed a field experiment that focuses on two different moderator styles: the directive versus the non-directive style. The ‘directive’ condition is characterized by a largely structured interview format (including interviewer instructions) and an active, emotion regulating moderator role. In the ‘non-directive’ condition the interview format is less structured and the moderator has a more distant, process-regulating role. The field experiment forms part of a larger study into consequences of welfare policy changes in the Netherlands for the daily work processes of social workers. We selected participants from six teams of social workers from six Dutch cities and created two focus groups, one for

each condition, matching participants on relevant characteristics. One of the researchers was trained as a moderator to perform as such in each condition. In this way the two conditions will vary on aspects like stimulation of information sharing, using closed and open questions, paraphrasing, inviting participants to speak and use of meta-language. The contributions of focus group members in both conditions will be analyzed in terms of data and process quality and compared. Indicators of data quality include amount of information gendered, the richness of information, and the amount of sensitive information. Indicators of (possibly) related process characteristics include aspects like the equality of member participation, the frequency of interactions, the group atmosphere and changes of such indicators over time. The outcomes will be related to theoretical notions on group interaction and online data collection. Possible consequences for the research practice will be discussed. Keywords: online focus groups, data quality, moderator styles

van de Vaart, W. (University of Humanistic Studies, Utrecht, Netherlands) Exploring flexible data collection methods for vulnerable, older populations Many social surveys include substantial numbers of people in complex personal, social, organizational or societal situations. Standardized data collection methods often are not fully qualified to obtain the required information from these hard to examine populations. Communicative, cognitive and/or motivational disabilities may make it hard to use standardized data collection methods in an adequate way. This study explores tailored data collection procedures for one such population: older people receiving long-term care. With that it address the fact that the own perceptions and experiences of older people receiving long-term care (LTC) remain highly under-researched due to lack of proper methods. Substantially the study focuses on of ‘meaning in life’ (MiL). Especially for elderly people receiving LTC, MiL is a salient issue. They are in their final years of life, have lost much of their independency and have to deal with physical, mental and social adversities. While research indicates that MiL correlates with better mental and bodily health, it is largely neglected in measurements of wellbeing and quality of life. The current research project involves collaboration with the Dutch Expertise Network Meaning in Life and Elderly. The project seeks to contribute to the further development of a national quality framework for Dutch LTC organizations on ‘coping with existential questions in long-term care for older people’. This paper encompasses qualitative research after indicators for a) ‘the quality of coping with existential questions’ as observed in LTC organizations, and b) the impact of ‘coping with existential questions’ on MiL of older clients. An inductive, explorative approach was used to find indicators that are based on experiences of older clients and health staff. A case study was designed concerning two purposefully selected LTC organizations. Data were gathered through documents, in-depth interviews with older residents and focus group interviews with staff from multidisciplinary teams and clients/family councils. In order to explore best ways of data gathering on such elusive themes in a vulnerable population, we designed a flexible interviewing method and combined several questioning procedures. Interviews were tailored in an interactive way, MiL/coping questions referred to organization activities and events, questions on ‘quality’ and ‘impact’ were based on concrete experiences, and evaluations could be provided by marks and/or open replies.

The study resulted in substantial themes that could well be connected to theoretical dimensions of MiL (like connectedness, goal directedness, transcendence) and can be specified into indicators for ‘quality of coping with existential questions’ and ‘impact on MiL of older clients. Furthermore, the study demonstrated that data gathering on these themes was in particular effective if a flexible and interactive approach was combined with a structured and focused interview guide. These conclusions are derived from an explorative, descriptive case study only and thus are meant as a first step to develop tailor-made methods that fit vulnerable older people. It will be discussed how - and under what conditions - such methods can be suitable for or ‘translated’ to general survey settings. Keywords: data collection, flexible methods, older population

van Houten, G., Schwarzer, S. and Simmons, K. (Pew Research Center) Using Timestamps for the Evaluation of Data Quality The use of new technologies for face-to-face surveys, such as devices for computer-assisted personal interviewing (CAPI), present many new possibilities to verify and ensure the quality of data collected. While the collection of the data comes at low cost, the costs for additional programming and the analysis of such measures should not exceed what is practical. One of the most common paradata is the measurement of time throughout the survey. The overall length of an interview has traditionally been used to verify an interview (ISO Standard 20252). The introduction of CAPI allows researchers to measure the length of sections in addition to the overall length, which could be seen as a valuable alternative for verifying an interview. Section-level timings can identify long or short sections within an interview and pinpoint interviewers that produce unusually long or short sections times. These unusual section times may indicate difficult sections, or could identify protocol violations committed by the interviewer. But multiple factors might contribute to suspicious-looking patterns in data and these detection methods should not be used in isolation. Consequently, falsified interviews should show other possibly suspicious characteristics. In our paper we will use data from surveys conducted by the Pew Research Center to investigate how to efficiently use time stamps to identify potentially fraudulent interviews. We will evaluate the relationship between time measures throughout the survey and other factors that are used to determine the quality of survey data, such as straightlining, the frequency of non-response answers, as well as uniformity and variance of responses. And since the duration of an interview can also be affected by the respondent’s ability to understand and answer the questions, we will also analyze the relationship between section timings and different respondent and interviewer characteristics. We will conclude by providing recommendations on how timestamps can be used effectively to evaluate the quality of survey data.

van Vlimmeren, E., Moors, G., and Gelissen, J. (Tilburg University) Developing a diagnostic tool for detecting response styles in survey research Previous research identifies response styles as an important source of systematic error in cross-cultural research. We contribute to this line of research by developing a diagnostic tool to detect response styles in large-scale questionnaires that can also be used when established methods are not applicable. Besides that, we explore different strategies on how to control for these response styles in cross-cultural data analysis once they are detected. The diagnostic tool we develop solves some of the issues we observed with the current methods of detecting response styles. One of such methods is the sum-score approach used by e.g. Greenleaf (1992) to count extreme responding and Baumgartner & Steenkamp (2001) to define an index of acquiescence. The procedure involves selecting a set of items with low inter-item correlations so that summing the number of times a respondent has chosen a particular response -- extreme responding or acquiescence respectively -- does not reflect content. A drawback is that this 'what you count is what you measure' procedure does not allow to evaluate whether it significantly contributes to explaining the data. The second method of detecting response style behavior circumvents this because it uses latent variable modeling to measure latent content and latent style simultaneously (Billiet and McClendon, 2000; and Moors, 2003). Comparisons of models with and without response styles allows to research whether response style contribute in explaining the responses given in the questionnaire. The drawback in this approach is that the detection of response styles is limited to the items used in the analysis. Furthermore additional constraints on the scales used might be necessary. For instance: to detect acquiescence the item-set should include indicative as well as contra-indicative items of the same concept. The method we propose includes the following features: (a) we use two sets of items with low inter-correlations selected from different items sets across the questionnaire; and (b) we use a latent class approach that allows identifying systematic preferences of particular response categories. We prefer the use of two sets of items over one set to reduce the chance of unanticipated content related covariance between items. This is made possible by adopting a latent class analysis in which we research whether respondents have chosen in a systematic way a particular response category across the items in the analysis. We simultaneously model response preferences in the two sets of items. Each item in the first set has a paired item in

the second set with which it correlates. By including direct effects between paired items we eliminate plausible "traces" of unanticipated content variance. The effectiveness of the above model is confirmed using data of the European Values Study (EVS) 2008. Three common response style patterns emerge from the data; extreme response style, mild responding and acquiescence. Dis-acquiescence was not observed as a systematic response behavior. Besides that, a considerable group of respondents showed no particular response style pattern at all. Considerable variation in response style behavior across countries is observed thus suggesting that country differences in observed item scores is potentially partially spurious on response style variation. To research this potential spuriousness in cross-cultural differences in item scores we followed two strategies that could be used by applied researchers. They involve the use of the saved posterior response probabilities from our model -- which will be made available to researchers on their request. In the first strategy, these posterior response style probabilities are entered as control variables in common regression models to obtain response style corrected scores. For the second strategy, response style adjusted weights are computed that equalize the country level response styles to that of the complete European sample. In order to evaluate the quality of these strategies we compare their outcomes with a benchmark model that involves extending the latent class response style measurement model with the item we want to correct, by defining a SEM-like equivalent. The comparisons to the benchmark revealed comparable results, so both adding the response style probabilities to regression models, as well as using response style weight adjusted scores work adequately in correcting for response styles in comparing item scores across countries.

Vandenplas, C., and Loosveldt, G. (KU Leuven) Monitoring the evolution of the fieldwork power: Illustration based on the seventh round of the European Social Survey Over the last twenty years, response propensities of possible participants have been decreasing. In an attempt to counterbalance this phenomenon, fieldwork efforts have been expanded by increasing the budget allocated to the fieldwork and by improving fieldwork strategies. Developing fieldwork strategies (such as responsive or adaptive designs) relies on the ability of monitoring the fieldwork and the evolution of its quality indicators based on paradata (e.g., response rates, contact rates, cooperation rates or representativity). Monitoring these indicators can better inform fieldwork management if they come paired with a benchmark. This benchmark relies on empirical information collected in the first phase of the fieldwork or, for repeated or longitudinal surveys, in previous rounds or waves. In this article, we suggest monitoring the evolution of the fieldwork power, which is defined as the production of the fieldwork unit per time unit. We operationalize the fieldwork power as the weekly number of completed interviews and of contacts, and the ratio of the number of completed interviews divided by the number of contact attempts and by the number of refusals. We use a repeated measurement multi-level model with surveys in the previous rounds of the European Social Survey as macro level and the weekly fieldwork power as repeated measurements to create benchmarks. For each country participating to the seventh round of the European Social Survey, we suggest considering three sets of surveys: all surveys from the previous rounds, surveys from the previous rounds displaying similar characteristics and surveys from previous rounds in the same country. We also monitor effort and data quality metrics. Belgium and the Czech Republic in round 7 are used as illustrations. The results show how problems in the fieldwork evolution can be detected by monitoring the fieldwork power and by comparing it to the benchmarks. The analysis also proves helpful as post-survey fieldwork evaluation and links effort, productivity and data quality. Keywords: fieldwork monitoring, European Social Survey, production metrics, paradata

Vidovićová, L., and Petrová Kafková, M. (Masaryk University) Who should interview older respondents about ageing-related topics? The effect of interviewer and interviewee characteristics on the subjective indicator of interview quality Increasing interest in the issue of ageing and old age is accompanied by a debate on the possible methodological pitfalls that might influence research into this specific, albeit very heterogeneous population. As some authors claim, the survey interview is a task-oriented structured conversation with scientific purpose, however, many of the norms of daily interaction still apply, and the age based social norms and chronological age related expectations may interfere with the data collection process. In this paper, we worked with five representative surveys conducted in the Czech Republic between 2007 and 2014. We followed the age and gender of the interviewer and age and gender of the respondent and their dyads in order to test whether and how age and gender influence a) the recorded difficulty in carrying out a particular interview; 2) selected subjective variables, such as subjective age or morale; and 3) whether differences in recorded difficulty of the interview vary according to the general topic of the survey. Only fully completed interviews were tested. We found that male interviewers reported a higher degree of difficulty in completing interviews, especially when being younger, interviewing female respondents, and when dealing with complex issues, such as age discrimination. Also, with respect to the subjective age indicator, specific age-gender dyads solicited slightly different results. Further, a respondent´s education and feeling of well-being and/or morale (measured by Philadelphia Geriatric Centre Morale Scale) was found to have an influence on interviewer’s subjective feeling of how difficult an interview was to complete. The obvious expectation that the research among the respondents in their fourth age is more difficult than with the more active “third agers” is supported by empirical evidence, pointing to the importance of interviewer characteristics. In the discussion, we present arguments for improving interviewee training in order to increase the quality and reliability of the data collection in the older population.

The preparation of the paper was supported by the Czech Scientific Agency grant No. 15- 03156S “The Fourth age: the identity of disability during the period of active ageing”. Keywords: interviewer effect; age/gender relations; subjective measures

von Scheve, C. (Freie Universität Berlin) Core relational themes as a method for reconstructing emotion from text Text can be emotionally captivating and moving, even if it does not contain a single emotion word, such as anger or joy. Some of the emotion-inducing qualities of text are found at the structural level (e.g., rhythm, meter), in phonetics (e.g., presence of phonemes), or the use of figurative language (e.g., rhetoric, metaphors). In sociological analysis, however, we are often confronted with ad-hoc narratives produced by interviewees or speakers in public discourse that lack a finely crafted aesthetic dimension. Reconstructing a speaker’s likely emotional experience from text as well as assessing a text’s capacity to induce specific emotions in recipients is thus methodologically challenging. However, many research designs in sociology and other social science disciplines require that we assess how a respondent may have felt in a specific situation or estimate how an audience might react to a specific narrative or speech act. In this contribution, I suggest that “appraisal theories” of emotion in conjunction with hermeneutic analysis provide valuable insights for this methodological task. In a nutshell, appraisal theories assume that emotions result from interpretations of events regarding a person’s goals, concerns, and normative standards. For example, if I witness my favorite team scoring a goal, I’m happy. The person next to me, who supports the other team, witnesses the same event and experiences sadness. Richard Lazarus in his 1991 Emotion and Adaptation developed the appraisal theoretic concept of “core relational themes” (CRT) that reflect the basic meaning structure of discrete emotions. To use the above examples, the CRT of happiness is “making progress towards a goal” and the CRT of sadness is “experiencing a loss”. I propose that to reconstruct the emotional experience of a speaker or to assess the emotional impact of a text on a specific audience, one can identify these core relational themes and hence “reverse engineer” the underlying emotion. Hermeneutics as a basis of qualitative inquiry generally aims at reconstructing meaning from text, often to understand a specific course of social action. Interpretation presupposes that at least some “common ground” and “knowledge” be shared between an interpreter and the interpreted. Given that a text contains, for example, autobiographical or other self-related narrative that reveals a speaker’s goals, beliefs and desires, and that the researcher can reasonably assume to share stocks of institutionalized knowledge with the speaker, it should

be possible to identify CRTs and related emotions. Likewise, it should be possible to estimate likely emotional reactions of a given audience to which we can attribute certain beliefs and desires. This procedure of course does not provide insights into “actual” emotions, because it relies on respondents’ subjective accounts of events or on attributions of how an audience will interpret a speech act. However, it allows to establish the emotional “frames” and “repertoires” along which individuals and audiences alike make sense of the world. Keywords: Emotion, qualitative methods, core relational themes

Vogl, S. and Fuchs, M. (University of Vienna; TU Darmstadt) Item-Nonresponse in surveys with children and adolescence. Children are increasingly seen as competent informants on their activities, views and needs. Different abilities of children can lead to inappropriate (formally or regarding the content) answers or to the misunderstanding of proceedings during an interview process. To identify such problems in a survey we use item-nonresponse as an indicator. We refer to the satisficing concept (Krosnick 1991) and focus on task difficulty and respondents’ abilities to explain or predict a poor question outcome, namely item-nonresponse. More specifically, we investigate the effect of question and person characteristics (and their interaction) on item nonresponse for written self-administered questionnaires in paper and pencil classroom setting. The study is based on 2 large scale surveys conducted in Germany in 2004 and 2010 with about 6,000 students in total from the age of 9 to 17 years from different school types. In the analyses we employ hierarchical regression taking account of item and person characteristics. Results indicate that younger respondents are especially prone to item- nonresponse. Also, we identify several question characteristics that pose particular problems to young respondents. Keywords: Item-Nonresponse; item characteristics; children

Wan, A. Model Averaging in Sociological Research Model selection, which affects all stages of inference, is of special relevance to empirical sociological research that makes use of statistics. Typically, a sociologist would select a single model from the pool of all possible models, proceed as if the selected model were known beforehand, and neglect the uncertainty in the model selection step. Recent statistical literature has advocated model averaging as an alternative to model selection. Model averaging assigns weights to different models. These weights are then used to produce average estimates of the unknown parameters. There is ample evidence to indicate that model averaging frequently produces more efficient estimators than model selection. Model averaging also guards against the selection of very poor models. The purpose of this paper is to introduce model averaging to the sociological research community through the ordered probit and nested logit models that are widely used in empirical sociological research as a platform. We examine a range of averaging and selection schemes for these models. Our results show that although neither averaging nor selection is a uniformly better strategy than the other, selection results in the poorest estimates far more frequently than averaging does, and more often than not, averaging yields superior estimates to selection. Among the averaging methods considered, the one based on a smoothed version of the Bayesian Information criterion frequently produces the most accurate estimates. The proposed methodology is applied to two real data sets related to sociology.

Wagner M., and Bergmann, M. (Munich Center for the Economics of Aging, Max Planck Institute for Social Law and Social Policy) Cultural response styles in the evaluation of people´s social contacts in 11 European countries Over the last decades research has repeatedly found a clear North-South divide between European countries regarding loneliness: People from Northern European countries report to be less lonely than people from Southern European countries. This result is surprising – as social contacts are considered as main factor for loneliness and in Northern European countries more people live alone and families bonds tend to be “weaker”. The aim of this study is to explain this finding by analyzing the cultural patterns in the evaluations of social contacts and loneliness. A first step takes a closer look at the relation between social contacts and loneliness: Loneliness is not defined by the number of social contacts per se. By definition, it rather conveys a subjective evaluation of people´s contacts as being satisfying or unsatisfying. So loneliness can be considered as the degree of satisfaction with the existing social contacts both in terms of quantity and quality. However, norms and values may determine the frequency and quality of social contacts that are considered as appropriate in a given society. It is argued that people in societies with mainly individualistic values are less prone to loneliness than their counterparts’ societies where collective values are dominant. This is because the optimal set of relationships is larger and harder to obtain in collectivistic societies, especially for older people. Furthermore, these country differences can be biased by the way people from different cultural backgrounds express or admit negative feelings. Admitting loneliness might be a social stigma in some countries but not in others. Against this background we will analyze country differences by using anchoring vignettes. In general, anchoring vignettes are applied to correct for biases in response behavior. Instead of correcting for these differences we take the vignettes to measure the scope of the cultural response styles of loneliness on the country level. We use data from the Survey of Health, Ageing and Retirement in Europe (SHARE), a cross-national panel survey for people aged 50 years and older. In 2007, vignettes from 7500 respondents from 11 European countries (Sweden, Denmark, Netherlands, Germany, Belgium, France, Italy, Spain, Greece, the Czech Republic, and Poland) were collected. Respondents were asked to rate their grade of

satisfaction with their social contacts. Afterwards they are asked to rate the satisfaction of two hypothetical persons concerning their hypothetical networks. First results show that people in Northern (and Western) countries, such as the Netherlands, Denmark and Sweden, rate the social contacts of the hypothetical persons more positively than respondents from Southern (and Eastern) countries, such as Poland, Italy and Greece. Based on this descriptive finding we assess factors that may explain these differences, e.g. people´s own assessment of social contacts, demographic variables, or country-level variables. Moreover, we discuss if country differences in loneliness can be interpreted as unsolicited method effects or as substantive cultural differences.

Wagner, C., du Toit, J. and Venter, S. (Department of Psychology, University of Pretoria, South Africa and Department of Town & Regional Planning, University of Pretoria, South Africa) "I actually realised how much we know”: Developing workplace research skills for psychology students through interdisciplinary near-peer mentoring A coursework Master’s degree is required to become a psychologist in South Africa. One of the specialisation routes is ‘Research Psychology’ that trains students in a broad range of skills in the social sciences and to be experts in psychological research methods. Experiential learning is often touted as an effective way of imparting research skills. This suggests that master’s students undergoing training to become research psychologists should be exposed to actual research, including the supervision and managing of research teams and projects of an increasingly interdisciplinary nature. The Departments of Psychology and Town and Regional Planning at the University of Pretoria, South Africa, developed a one-on-one near- peer mentoring model in which master’s students in research psychology mentored undergraduate students in urban planning who were in the process of writing their final-year research reports on an environmental psychology topic. The model was intended to give psychology students experience in mentoring and general research supervision as a forerunner to managing research teams and projects in practice. A focus group discussion with the six psychology students showed that they experienced their mentoring relationship as both rewarding and challenging. They felt that they developed additional skills such as communicating basic research methods and report editing. Recommendations for using near- peer mentoring for teaching workplace research skills are indicated. Key words: interdisciplinary, near-peer mentoring, research skills

Wang, C. (Research Center for Humanities and Social Science, Academia Sinica; Chinese Culture University)

Factors Influencing Non-Response in Surveying Taiwan’s Bureaucrats Conducting government bureaucrat survey research has been a common practice in the field of public administration. Different from general survey, government bureaucrats are more sensitive to politics and easily hesitate to express their opinions in interviewers. When implementing questionnaires survey, it is quite usual for researchers to encounter the nonresponse feedback which inevitably has some statistical impacts when analyzing and interpreting the results. On the basis of this perception, this paper takes the survey of 2008 Taiwan Government Bureaucrat as an example to explore the non-response-“unit” and “item” -in surveing bureaucrats. Based on the findings, this study finds the relationship between respondent gender, public service, and age is statistic specific to the nonresponse of government bureaucrats. Suggestions, in empirical and theoretical terms, were discussed.

Keywords: Government Bureaucrat Survey, unit-nonresponse, item-nonresponse, sensitive questions, paradata

Warner, U. and Hoffmeyer-Zlotnik, J. H.P. (Institute of Political Science, Justus Liebig University of Giessen, Germany) International Harmonized Labor Status Concepts and Measures for the Analysis of Social Structures across countries and cultures: A questionnaire module for comparative social surveys This paper presents a questionnaire module to measure harmonized labor status across countries and cultures. We present our own instrument to measure occupation and labor status as social demographic background variables in cross-national comparative surveys. The labor status of survey respondents is an important socio-economic and demographic background variable necessary for the analyses of social class, social stratification, social mobility, and the social structure. Respondent’s labor market situation is an important indicator to explain social inequalities and the social closeness or openness of modern societies. In the social sciences, the labor status variables are used to derive socio-economic status and/or occupational prestige. This paper describes labor status as an indicator for class, social status and as an indicator of the respondent's position in the life-cycle. We identify variables necessary to measure occupation, employment and the labor force concept from ILO. Finally, we summarize the findings from a pilot survey. The aims of the proposed instrument for the measurement of socio-economic status and personal labor market situation are threefold: a) to collect all the information needed to determine respondents' socio-economic status, b) to capture respondents' personal labor market situation as precisely and comprehensively as possible with the help of a small number of questions, and c) to ensure comparability with the reference data of the national statistical offices by incorporating the ILO's labor force concept. First, we determine whether the respondent is employed and working, and, if yes, how many hours a week. Then, the labor status of those respondents who are not working or who are marginally employed is ascertained using relatively detailed categories. If the target person is unemployed, seeking employment, or undergoing retraining, they are asked if they would be available for work in the short term. The main focus of the fourth step is the professional status of the respondent. Those target persons who are no longer in employment are asked whether they worked full- or part-time in the past.

In the case of respondents who are currently or were previously in full- or part-time employment, a description of their main job is obtained in order to be able to code it into ISCO-88 or ISCO-08. This step entails finding out whether the respondent is the main earner in the household. If this is not the case, the occupation, occupational category, and supervisory responsibilities of the main earner must be ascertained. This additional focus on a second person in the household – the main contributor to the household income – is necessary in order to ascertain the household's socio-economic status because the person with the highest status also influences the status of the other household members. Finally respondents are asked the question that is central to the ILO labor force concept: "Did you do any work for pay, profit or family gain for at least one hour during the past week?" If the answer is "no", they are asked to give the reason for this inactivity, even if it was only temporary. During the interview we collect all information necessary to compute the most important indicators for the comparisons of social structures (SIOPS, ISEI, ISCO, EGP, ICSE, etc).

Weidenhaus, G. (Institut für Soziologie TU Berlin) Methodological Approaches to Space From an epistemological point of view it is quite difficult to develop an appropriate concept of space. The reason is, that the word “space” is used in several different senses. According to Jürgen Habermas, we can differentiate between talking about the physical world, the social world and the inner world of subjects – the psychological world. The problem with the word “space” is, that it is used with references to all these worlds. First of all, I will develop three concepts of space from an epistemological perspective. Spatial descriptions can refer to the physical world („the distance between point A and point B is four kilometers“) or to the social world („Germany is a nation state“) or to the inner world of subjects (“this crack in the rock seems unbelievably deep to me”). Underlying these spatial descriptions are very different conceptions of space. I want to argue that these three different concepts of space are common in social science, but mostly without reflecting the crucial differences between these approaches. There are different rules to evaluate the adequacy of those descriptions. Physical descriptions of space must be verified in the outer world, so they have to be measured. Social descriptions of space require the knowledge and the recognition of the meaningful social institution (such as nation states or living rooms), so they have to be reconstructed and on the third, one has to believe psychological descriptions of space. We got three different concepts of space that – epistemological speaking – are incommensurable. Within a second step, I like to show that all three concepts are relevant from the perspective of social science. Depending on the cognitive interest and the research question we should use a specific concept of space. Social scientists who explore mobility, multilocality or functional spatial differentiation usually use a concept of physical space. The pivotal category is distance. Within this kind of research, space is an independent variable. With the concept of physical space social scientist do their research in space. Quiet different to the use of physical spatial references, when Social scientist focus on a concept of social space. They investigate how people orientate in the world by meaningful social constitutions of space (for example: Do people constitute homes and how?). The analysis of psychological space is still rare in the social sciences. With a concept of psychological space one can address questions

concerning the immediate prelingual perception of space. It seems to me that this perspective is not irrelevant especially for architecture and planning science. Within the third step, I try to show that using different kinds of concepts leads to different kind of data and different methods that can be used for spatial analysis. Objectified distance data are necessary to describe physical space or we can use a wide range of data (qualitative or quantitative) to explore and reconstruct relevant social spaces and their historical development.

Keywords: Spatial Analysis, Physical Space, Social Space

Weis, C. (De Montfort University) Wearing out my welcome? Negotiating authority and access when recruiting participants with the help of gatekeepers In this presentation, I discuss the methodological and ethical choices and struggles that I encountered when recruiting participants - with the help of gatekeepers - during my ethnographic research on commercial gestational surrogacy in Russia. I grapple with the following: Who has and who should have authority to grant or restrict the researcher access potential participants? When does authority become illegitimate or even abusive? When enmeshed in asymmetrical power dynamics, how to make ethical choices consistent with the feminist research ethos to be transparent about research steps, to empower research participants and to challenge existing power relationships – yet without putting research participants or one’s own research access at risk? I elaborate on these questions via examples from research. 1) One IV specialist: He offered to function as a gatekeeper by informing surrogacy workers about my research during their appointments (embryo transfer) and proposing their participation. His efforts helped me gain access to these women, yet at the same time, I suspected that they might experience his suggestions as coercion, given his high status and authority. 2) Surrogacy agency: access was granted to conduct interviews and participant observation with surrogacy workers on condition that the surrogacy workers agree. Over time, the agency’s welcome wore out, while their surrogacy workers continued to welcome my research. Should I accept the agency’s desire to terminate my research with their surrogacy workers, or accede to the wishes of the surrogacy workers and my own research needs and continue working with them? 3) A surrogacy worker whom I recruited independently via online recruitment strategies agreed to participate in my research. During the course of data collection, she signed up with an agency that rejected my research, leaving me with the ethical dilemma: should I continue working with her, or not? 4) The general director of an agency offered to hire me as their international representative and contact person for international client parents. In turn I would have gained access to the

inner workings of the agency and to surrogacy workers who might otherwise have declined to collaborate. Over time I discovered various degrees of agency misconduct. As agencies grew aware of my increasing knowledge, my welcome wore out. This knowledge confronted me with another ethical dilemma: Do I share or withhold that information from the surrogacy workers? The ethical dilemma I faced hinged on timing: the surrogacy workers I had met through agencies had already signed contracts with them. Thus a warning would come too late and not all agency surrogacy workers had bad experiences. What was the right thing to do? I contend that it is the responsibility of feminist researchers to disobey the powerful and their intention to maintain a status quo when that perpetuates incapacitation and control of the subaltern, and to work towards the cause of empowerment. Yet, in so doing, we need to be careful to fully protect participants’ right to anonymity and confidentiality. Keywords: recruitment, gatekeepers, hospital setting, ethical choices, gestational surrogacy, Russia

Weller, A. (TU Chemnitz) The visual as modus operandi The following paper wants to outline the implications of working on a visual mode using the example of a research project. Pictures show us a unique perspective on the world and expose all of this in a simultaneously way. To analyze visual data you have to deal with this simultaneousness, which is one of the main characteristics of pictures. Susanne Langer (1965) emphasizes the simultaneity as integral function of the visual in her concept of presentative symbolism. The crucial task is how to pay attention to all the details and also be aware of the whole thing. In the research project ‘Wissensbilder’ I reached and analyzed a corpus of nearly 200 drawings and paintings, depicted by pupils of the 7th, 9th and 11th class of a saxonian Gymnasium. They were drawn regarding two themes, which were given to the pupils beforehand. For the analysis a contrastive method was applied. In a process of circular 'Bildauslegung' (corporal and cognitive interpretation of all the pictures in the corpus) and 'Bildanschauung' (the deeper interpretation of themes and subthemes recognized in the process of 'Bildauslegung') the eyes have to deal with a corpus of pictures and hence a comparative visual impression. In the first phase of analysis the range of graphical answers was identified involving four experts for a first coding of the visual data. This first external corpus-based 'Bildauslegung' was used to extend the researchers point of view and to meet the challenge of working with the ambiguity of visual data. Heinrich Wölfflin (1948 [1915]) contrasts the visual impression in order to discover the quality of expression. He uses the method of parallel projection to compare pictures with each other and expand the horizon of interpretation. Therefore the comparison enables the researcher to see the essential of the antagonys with a great power (Wölfflin 1948 [1915]: 54). Thus the pictures were analyzed as a corpus to broaden the horizon of interpretation. Analyzing multiple pictures and operating with a whole collection of pictures, recombining and regrouping them becomes the primary mode of analysis (Müller 2011). The comparison of pictures not only presents the similarities, but rather the differences in the particular depictions. Goffman emphasizes that a certain set of pictures allows the recipient to see not only patterns, but also different settings in just one frame (Goffman 1981: 109). For the analysis I used 'Bildtafeln' (roughly picture tableaus, Warburg 2000) to generate insights and

document the process of analysis, i.e. save the comparisons. The 'Bildtafeln' especially support the visual mode of the analysis because they require a seeing based on comparison that allows to envision the tertium comparationis. The identifying of the tertium is the goal of the comparison that leads the researcher to new aspects of seeing and new possibilities of contrasting (Müller 2011). With the tableaus you are able to visualize themes and compositions without deconstructing the picture in single visual fragments. They allow you to concentrate on the pictures as holistic artefacts without ignoring single details, that show themselves in the simultanous analysis of pictures. The tableaus furthermore allow the researcher to present results in intuitive, easy to understand and vivid way. When it comes to reach a wide publicity the Web 2.0 provides great opportunities for sharing visual data. For documenting the research process I shared and published different tableau versions on the Weblog of my research project (www.projekt- wissensbilder.de) and promoted the results via Facebook and Twitter. Additionally this helps to formulate the results in a more comprehensible way and to connect and get feedback from the scientific community. Furthermore I work with visualizations to display and condense the results of the particular study (schematic drawings), to visualize theoretical constructs and summarize presentations on conferences (sketchnotes), and to enrich my own presentations (drawings). New media offers a platform for sharing not only the results, but also the visualized theories and thoughts and in order to strengthen the visual as modus operandi in research.

Reference: Langer, Susanne (1992 [1965]): Philosophie auf neuem Wege. Das Symbol im Denken, im Ritus und in der Kunst. Frankfurt am Main: Fischer. Goffman, Erving (1981): Geschlecht und Werbung. Frankfurt: Suhrkamp. Müller, Michael R. (2012): Figurative Hermeneutik: Zur methodologischen Konzeption einer Wissenssoziologie des Bildes. In: Sozialer Sinn: Zeitschrift für hermeneutische Sozialforschung, Jg. 13, Heft 1, S. 129-161. Warburg, Aby (2000): Der Bilderatlas Mnemosyne. Berlin: Akademie Verlag. Wölfflin, Heinrich (1929 [1915]): Kunstgeschichtliche Grundbegriffe. Das Problem der Stilentwicklung in der neueren Kunst, München (7. Auflage).

Wengrzik, J. (GESIS - Leibniz-Institute for the Social Sciences)

Web Surveys versus Other Survey Modes: An Updated and Extended Meta-Analysis Comparing Response Rates Relevance & Research Question: The aim of this meta-analysis is to update and extend a previous meta-analysis by Lozar Manfreda et. al. (2008), synthesizing experiments published between 1998 and 2005 on response rate differences between Web and other survey modes. Based on 45 experimental studies identified, Web surveys yielded 11% lower response rates on average compared to all other modes. Furthermore, substantial variability around this average effect size was discovered, which could have been partially explained by the sample recruitment base, the solicitation mode and the number of contacts. Other factors as the type of mode web was compared to, type of target population, sponsorship, publication year or incentives showed no significant influence in this previous study. However, the original study by Lozar Manfreda et. al. (2008) faced a statistical power problem due to small numbers of observations for certain moderator levels. Since 2005, numerous studies carrying information about response rates differences between Web and other survey modes have been published. Consequently, we seek to address the following research questions in our update: How robust are the previous findings over time? Do web surveys still yield lower response rates, the same, or even larger response rates? If we still detect any effects, which factors influence response rate differences? Methods & Data: To identify studies published after 2005, we used the same eligibility criteria employed by Lozar Manfreda et. al. (2008) and have identified 63 new experimental comparisons, extracted from 43 papers. In total this meta-analysis refers to more than 100 experimental studies. Results: The overall findings indicate a remarkably robust response rate difference over time (13% lower response rates for Web surveys on average). Moreover, because of the increased number of primary studies, we were able to get more accurate estimates for all moderator variables and extended our analysis to additional moderators. A significant influence showed the type of mode web was compared to, the sample recruitment strategy, the target population, the type of sponsorship, if the study used incentives, in which region the study

was conducted, if a prenotification was used, the number of contact attempts and the number of follow-ups. Added Value: Practical implications and avenues for future research are being discussed.

Weßling, K. (University of Tübingen, Germany) Spatial dependencies and heterogeneity in individual returns to education Educational attainment is among the most important individual-level determinants of earned income and occupational position. It is therefore no surprise that returns to education have been a frequently analyzed topic in empirical sociological and economic research. Beyond individual-level factors returns to education are likely to depend on location. Local or regional contexts are unequal in terms of socio-economic conditions. These differing conditions can be assumed to influence individual outcomes such as returns to education. However, these contexts do not necessarily follow administrative boundaries. There may also be disproportionate similarity and mutual influences between proximate units of observation, contradicting the standard statistical assumption of independent cases. Hence, addressing the question of how regional or local context conditions contribute towards explaining disparities in returns to education faces a number of methodological challenges. We analyze economic returns to education with a specific focus on spatial dependencies and heterogeneity. Our contribution addresses two central methodological issues: (1) How should contextual settings of returns to education be operationalized in terms of their spatial definition? (2) What type of statistical models should be applied to analyze spatial effects in returns to education?

We base our analyses upon data from the German Socio-Economic Panel Study (GSOEP). The GSOEP is a representative longitudinal study of private households. It provides information on individuals’ educational careers as well as on their occupational status and income situation. Beyond that, individuals can be located within geographical units so that the spatial variability in returns to education can be estimated. Geographical information is available on different levels of aggregation (e.g., ZIP code areas, municipalities or districts). We make use of this variability and aggregate the units flexibly. We account for spatial clustering and assess the relevance of spatial dependencies with different methodological approaches: There are several approved techniques for including

information on clustering such as calculating clustered standard errors. Multi-level techniques go a step further and allow assessing the impact of variables on different analytical levels. Such models, however, still assume that the local and regional mechanisms work within the boundaries of the considered local or regional units of aggregation; they neglect the potential effects of surrounding units. Therefore, we compare these models with geographical weighting regression techniques that take non-stationarity and spatial proximities within the sample into account. We derive local standard errors and assess whether effects of explanatory factors and the explanatory power of the model vary geographically. Spatial regression models require the definition of relations of proximity and distance among the observations, and the specification of these may be another factor influencing the results. We therefore vary relevant parameters in order to localize contextual settings that impact economic returns to education and to capture spatial dependencies among individuals.

Wetzels, M. (Technical University Berlin) The problems of practicing video graphic methods at the example of religious and sportive mass events In this proposal I want to offer a discussion about the issues that social scientist may have to face by practicing video graphic methods in the public society at the case of mass events. Despite the fact that there are several items of technical issues (what sort of camera can be used in a certain field?) and situational effects and problems at the time scientific people are in ‘their fields’ (Am I at a stadium or a public viewing event?), the double aspect of being a private and a scientific person can be a crucial influence for the work of social scientist. Questions like: “What do I do if someone is exposing me in front of everyone? Am I even allowed to record this event? What should I do if people get angry at me?” may appear. This has to be taken seriously and discussed, if we want to develop video graphic methods further, especially in (re-)designing theoretical approaches. Focussing on three aspects (frames of an event, interaction with people, the double “role” of the scientist) I want to show how these examples can create an uncertainty for “going into the field”, even and especially if you’re a “native” in “your” field. To substantiate my point, I want to show some empirical data (video material) we’ve made in our project by visiting religious and sportive mass event in Germany and linking this to the aspects I want to discuss.

Wijnant, A. and Elshout, S. (CentERdata, Tilburg University) Mobile devices in web panels: a case study of how to upgrade an existing panel to be ready for mobile device respondents People are increasingly accessing the internet with touchscreen mobile devices, such as smartphones and tablets. Webpanel statistics show that these devices are increasingly being used to complete questionnaires. The unique characteristics of mobile devices require a different type of user interface, which the standard web interview systems do not offer. In this presentation we show what was necessary to change in an existing web panel environment and how we implemented these changes. Items like questionnaire design, invitation mode, user interfaces and data traffic optimization are part of the operation to create a panel that is ready for the future.

Wolf, C. and Roßteutscher, S. (GESIS - Leibniz-Institute for the Social Sciences, Germany) Religiosity and political attitudes in Europe, 1981 to 2008 Research question Traditionally religion and politics had been closely connected in Europe as signified -- among other things -- by the significance of Christian parties in Europe. These parties usually hold conservative, family oriented values. As the significance of religiosity is declining the question arises how this affects the relation between religiosity and political attitudes. Hypotheses One could expect that secularization increases the cleavage between secular and religious persons leading to a polarization of political attitudes between these groups. This is a finding Chaves and others have reported for the USA (God gap). Alternatively, however, one could assume that the connection of religiosity and political attitudes loosens because religion becomes privatized and is less and less connected with public life; a process sometime described as individualization or de-structuration. Proponents of this hypothesis point to the declining significance of religion for voting which can be observed throughout Europe. Data and Method We investigate these hypotheses on the basis of data from the European Values Study (EVS) which collected four waves of interview data between 1981 and 2008. We include 26 countries in our analysis that participated in at least three of the four rounds of EVS. Religiosity is measured using a broad set of 10 indicators. As indicators of political attitudes we use ideological orientation, political participation and trust in institutions. For each country and each observation year we compute the correlation between religiosity and these political indicators and then follow the trajectories of these correlation coefficients. Findings Our findings show that Europe has experienced a development that is very different from what is reported for the USA. In contrast to polarization we find very clear signs for a declining importance of religiosity for political attitudes. A finding that itself can be read as another dimension of the secularization process Europe is experiencing.

Wulfe, M. (The World Bank Group) Using the World Bank's Survey Solutions to Maintain Data Quality during Fieldwork: Examples in South Africa and Tonga

Survey Solutions is the Computer-Assisted Personal Interview (CAPI) technology developed by the World Bank. It assists governments, statistical offices and non-governmental organizations in conducting complex surveys with dynamic structures using tablet devices. The software can be tailored to the needs of the clients, allowing them to successfully complete simple and more sophisticated projects, from basic evaluation questionnaires to complex multistage panel surveys, all with minimal requirements for programming skills. The software is offered free of charge; its development is co-financed by the World Bank, the Bill and Melinda Gates Foundation, and the Food and Agriculture Organization of the United Nations. Surveys are conducted on low-cost Android tablets. Survey Solutions has been successfully used in more than 100 surveys worldwide. This presentation will demonstrate how Survey Solutions produces high quality data by monitoring the data as it is collected in the field. With a user-friendly online interface that requires no programming or customization, it generates multiple reports for fieldwork quality control. The reports include cumulative fieldwork progress for data in each stage of processing; geographic visualization of cases on map reports; and enabling survey management to review data immediately after transmission. Data collection in Tonga and South Africa will be used as examples to demonstrate the use of the software in very large and very small surveys. This presentation will introduce each of the components of the software and how they are used to maintain the highest level of data quality during fieldwork. Keywords: World Bank, Survey Solutions, CAPI, tablets, Android

Yam, S. - C. J. (Chinese University of Hong Kong) Econopolitpsychoanthroculturalsociological Reality: Hidden Discourses of the Early 21st- Century Social Sciences The rise of interdisciplinary mixed methods has provided cooperative opportunities among social science disciplines, but this also brings conflicting aims and interests into the choice of research method with no simple way to resolution. In this paper I attempt to untangle these conflicts at the source by studying the motive problem of methodological conflict/cooperation - the way that disciplines 'introduce' themselves and talk about others. I apply (critical) discourse analysis to early 21st-century textbooks from the six social science disciplines – economics, political science, psychology, anthropology, cultural studies, and sociology – to investigate the discursive resources that disciplines use to frame themselves and interdisciplinary relationships. Using Yam’s (2015a) Central Dogma of Social Science as the framework, my focus is on how textbooks approach the scope/determinacy dilemma with their latent disciplinary discourses. I reveal how low-scope disciplines expand their apparent scope of mathematical models with a discursive process that I refer to as ‘selective radiation’. Mid-scope, quantitative/qualitative disciplines limit their scope through a guided focus on the experimental method/diversity. High-scope disciplines ignore determinacy concerns by conversing with fluid concepts. Finally, disciplines can also position as a synthesis and fragment into fields with multiple and conflicting yardsticks. The results suggest that disciplines spend more effort elaborating scope than determinacy, and a key starting point for interdisciplinary cooperation is reflexivity/self-criticism.

Keywords; Philosophy of social science; interdisciplinary cooperation; Central Dogma of Social Science

Zaccaria, D., Sala, E. and Guaita, A. (Golgi Cenci Foundation, Abbiategrasso-Milan, Italy; Department of Sociology and Social Research, University of Milan Bicocca, Italy) Non-response, attrition and age in a cohort study of older people in Italy: The case of InveCe.Ab study 1. Background and research aims Wave 1 non-response may have detrimental effects on the quality of longitudinal data. For longitudinal health surveys of older people, little is known on the mechanisms that lead respondents to participate in surveys, especially to the first wave of data collection. The few papers that investigated the determinants of wave 1 non-response found that older people and less educated respondents are less likely to respond. However, other characteristics, such as sex or partners’decision to take part, may also be associated to survey participation. In addition, research has never explored the extent to which male response pattern differs from female one. This is a key point as contact strategies may be tailored according to respondents’ gender. This paper explores the mechanisms that lead to Wave 1 nonresponse in a longitudinal survey of older people. Using a detailed set of socio-demographic characteristics, we (i) investigate associations between respondents’ characteristics (especially age) and the propensity to participate and (ii) explore whether there are any gender differences in response patterns. We also (iii) explore the reasons for not responding. 2. Data InveCe.Ab study is a biannual cohort study with a focus on mental health conditions of the older people born between 1935 and 1939 and living in Abbiategrasso, a city of 32,000 inhabitants near Milan (Italy). Wave 1 was carried out in 2010 and involved 1,321 respondents (wave 1 RR is 80.4). The data collection involved both an evaluation of respondents’ medical and neuropsychological conditions and the administration of a face to face questionnaire collecting information on housing conditions, social relations, etc. The list of eligible respondents was drawn from the registries of the local Municipality; these registries provide detailed information on sample members’ socio-demographic characteristics.

3. Methods We use a combination of bivariate and multivariate methods. We use bivariate analysis in the preliminary stage of the data analysis; we shall then apply regression-based techniques (i. e., logistic regression) to evaluate the effect of a set of correlates (age, sex, household effect, etc.) on response at Wave 1. 4. Preliminary results on wave 1 non-response Preliminary results from the bivariate analysis have shown that Wave 1 response is not dependent on respondents’ year of birth. However, it is positively associated to education (t(df 326)=5,3, p=0,000) and sex (chi2(df 2)=17,9, p=0,000). Men are found to be more likely than women to participate in Wave 1. Preliminary results from logistic regression have shown that older people who live with a partner (eligible for the study) are four times more likely to participate to the first Wave of data collection. When exploring the reasons for non-response we find that 63% and 25% of eligible respondents do not take part as they are not interested or report health problems, respectively. Keywords: Aging population, Non-response, Reasons for non-participation

Zajc, N. (Statistic Office of Republic of Slovenia) Current experience with mixed mode including web mode in opinion surveys in official statistic A survey as a method of data collection is used in many and different areas. Due to the increasing diversity of data collection methods, SURS is confronted with the problem of which method to choose in order to gain the best data. On the other hand, because of the flood of marketing surveys, a feeling is being spread among the people that surveys are irrelevant or they represent some burden for the respondents. In this case, collateral damage is produced for organizations professionally engaged in conducting surveys, including the national statistical offices (NSI). The Statistical Office of the Republic of Slovenia (SURS) conducts numerous surveys of business entities; they are carried out through postal questionnaires, telephone (CATI) and web data collection methods. SURS conducts surveys of individuals and households which are also the subject of this article. For surveys of individuals and households SURS primarily uses the telephone and field data collection methods in longitudinal surveys; however, a mixed mode of data collection is also used. SURS is facing an increasing unit non-response rate and an increasing coverage error, because people do not want to participate or their telephone numbers are not listed, which will be shown in the following chapters. One approach to solving the problem of rising non-response rates and coverage errors is to use combined methods of surveying (mixed mode). Such an approach combines the advantages of the different methods and at the same time reduces the disadvantages of these methods. We can decide to use mixed mode methods in all phases of the survey process. However, researchers’ unreasonably deciding to use mixed modes also gives rise to negative effects on the quality of data and survey costs. In 2014, SURS carried out a pilot consumer opinion survey, where we implemented the second method of data collection (web). We wanted to reduce the non-response rate and the coverage error. Paper presents the research design of this and its results. We also present the results of the first few months of 2016 when we started using the combined method in this survey. Keywords: web survey, mixed mode, official statistics

Zangger, C. (University of Bern, Switzerland) Bringing space into the equation. Using spatial econometrics to untangle neighbourhood effects on educational outcomes Although there is increasing evidence for spatial externalities of one’s place of residence on various forms of life chances such as individual health, labor market involvement or educational achievement (for reviews see Jencks & Mayer 1990; Dietz 2002; Sampson et al. 2002; Galster 2012), the identification of the corresponding mediating social mechanisms is part of an ongoing debate. The well-known methodological challenges—namely the assessment of the spatial scale of “neighborhood”, the presence of unobserved (self-) selection processes and the reflection problem of endogenous effects (Manski 1993; Lupton & Kneale 2012)—hinder the evaluation of how, when, and for whom different social contexts matter (Sharkey & Faber 2014). There is, however, an additional methodological issue that— although associated with the mentioned challenges—has not yet been addressed in the literature on neighborhood effects. While most approaches stress the crucial importance of the social interdependence of individual action within a given context, the methods used to evaluate these theories assume just the opposite—namely independent and identically distributed observations (Beck et al. 2006). We address this mismatch between the theoretical and the methodological framework in the study of neighborhood effects by using spatial econometric techniques that allow the modelling of spatial and social multiplier effects at the individual level (LeSage & Pace 2009; Vega & Elhorst 2013; Elhorst 2014). Using data from Switzerland, we analyze how the social composition of a neighborhood affects children’s educational achievement. On the one hand, it is demonstrated in which ways the incorporation of social interdependencies at the individual level can be used to reproduce and understand the results from contextual models that make use of aggregated data on given social entities. Hence, this approach allows us to adequately incorporate people’s everyday perceptions of “neighborhood” into our models. On the other hand, we present spatial Durbin models as an adequate way to directly assess the suggested social mechanisms (e.g., the positive impact of the interaction with high status neighbors). In this regard, we finally illustrate how such models can be used to incorporate spatial externalities into a rational action model of educational inequalities (Breen &

Goldthorpe 1997). Thus, spatial econometrics bear the potential greatly enhancing our understanding of the mediating social mechanisms of neighborhood effects.

References: Beck, Nathaniel, Kristian Skrede Gleditsch, and Kyle Beardsley. 2006. “Space Is More than Geography: Using Spatial Econometrics in the Study of Political Economy.” International Studies Quarterly 50 (1): 27–44. Breen, Richard, and John H. Goldthorpe. 1997. “Explaining Educational Differentials: Towards a Formal Rational Action Theory.” Rationality and Society 9 (3): 275–305. Dietz, Robert D. 2002. “The Estimation of Neighborhood Effects in the Social Sciences: An Interdisciplinary Approach.” Social Science Research 31 (4): 539–75. Elhorst, J. Paul. 2014. Spatial Econometrics. Form Cross-Sectional to Spatial Panels. Springer Briefs in Regional Science. Heidelberg: Springer. Galster, George C. 2012. “The Mechanism(s) of Neighbourhood Effects: Theory, Evidence, and Policy Implications.” In Neighbourhood Effects Research: New Perspectives, edited by Maarten van Haam, David Manley, Nick Bailey, Ludi Simpson, and Duncan Maclennan, 23 – 56. New York: Springer. Jencks, Christopher, and Susan E. Mayer. 1990. “The Social Consequences of Growing Up in a Poor Neighborhood.” In Inner-City Poverty in the United States, edited by Laurence E. Lynn and Michael G. H. McGeary, 111–86. Washington D.C.: National Academy Press. LeSage, James P., and R. Kelley Pace. 2009. Introduction to Spatial Econometrics. Boca Raton: Taylor & Francis. Lupton, Ruth, and Dylan Kneale. 2012. “Theorising and Measuring Place in Neighbourhood Effect Research: The Example of Teenage Parenthood in England.” In Neighbourhood Effects Research: New Perspectives, edited by Maarten van Haam, David Manley, Nick Bailey, Ludi Simpson, and Duncan Maclennan, 121–45. New York u.a.: Springer. Manski, Charles F. 1993. “Identification of Endogenous Social Effects: The Reflection Problem.” Review of Economic Studies 60 (3): 531–42. Sampson, Robert J., Jeffrey D. Morenoff, and Thomas Gannon-Rowley. 2002. “Assessing ‘Neighborhood Effects’: Social Processes and New Directions in Research.” Annual Review of Sociology 28: 443–78. Sharkey, Patrick, and Jacob W Faber. 2014. “Where, When, Why, and for Whom Do Residential Contexts Matter? Moving Away from the Dichotomous Understanding of Neighborhood Effects.” Annual Review of Sociology 40: 559–79. Vega, Solmaria Halleck, and J. Paul Elhorst. 2013. “On Spatial Econometric Models, Spillover Effects, and W.” In 53rd ERSA Congress. Palermo, Italy.

Zifonun, D. and Naglo, K. (Philipps-Universität Marburg) Core Activity, Event and Crisis: Ethnographies in Football Worlds This paper argues that (1) the study of present-day society benefits from an ethnographic approach to core activities, events and crisis and that (2) existential commitment to social worlds is a prerequisite for successful ethnography, even though the study of core activities, events and crisis requires only short-term field visits. The theoretical starting point for the argument is in the sociology of social worlds. It conceives of social structure as being segmented into separate spheres of action and perception with their own special stock of knowledge and patterns of social relations. Rather than viewing society as granting full membership to a community with its ‘whole way of life’ or as being stratified into classes with their respective consciousness, it pictures society as a set of part- time worlds with part-time memberships. Sociological ethnographies that take a social worlds perspective are concerned with social relations that emerge in social worlds. On the one hand, this makes it necessary for ethnographers to immerse in these relations, acquire the status of a participant in these relations and become existentially committed. While this is in line with classic ethnographic work and requires a long-term participation in the social world under study, the part-time character of social worlds already on a first level limits the time spend in the field for observant participation. On the other hand, the peculiar structure of social relations and encounters in social worlds comes into play. When arguing from a ‘community’ or ‘class’ perspective, social relations are conceptualized as fluid, poly-thematic and expansive and as an ongoing stream of relatively unquestioned routine everyday-life encounters, interrupted by ceremonies and rituals. From a ‘social worlds’ perspective, however, social relations appear to be focused, issue-centered and limited which plays out in encounters that consciously aim at performing and supporting core activities, punctuated by events and crisis. Thus, on a second level, field time can be limited by focusing on participation in (and in-depth analysis of) core activities, events and crisis. This allows for a reconstruction of (1) of the subjective experience of interaction, its conditions and consequences, (2) the order of interaction itself and (3) the meaning system that governs interaction. The argument is developed by using the involvement of the two authors in the world of grass roots football and their participation in core activities, events and crisis in its subworlds. The

approach seeks to locate the grass-roots game in its various manifestations within an analytical framework that will help to facilitate the development of comparative perspectives. Thus football at the non-elite level is very much a glocalized world in which social actors interpret global processes and phenomena according to their particular beliefs and local needs. The grass-roots football experience, on account of its sheer complexity and the fact that it is so often embedded in a sense of local identity, challenges the idea of globalization as a one-dimensional process characterized by standardization. The data used are (1) a football play from the free kick to the goal scored that displays the team’s style and conveys the club’s identity (as an example for a core activity); (2) the so called Ehrenrunde (Lap of Honour) organized by the DFB (German Football Association), through the introduction of which the DFB is offering 63 selected amateur clubs the opportunity to present the “2014 FIFA World Cup Winners Trophy” for one day on their premises (event); and (3) a team meeting following a violent incident that threatened to delegitimize the club and almost lead to its exclusion from league operations (crisis). The paper argues that calling for existential commitment and for limiting attention to core activities, events and crisis is only seemingly contradictory. Rather, two different modes of temporality need to be distinguished: the biographic time of the professional ethnographer and the research time of a specific ethnographic enterprise. The paper shows that by studying core activities, events and crisis field time can be reduced in a controlled manner while this still allows to systematically reconstruct the structure of social relations.

Zimmermann, T. (University of Kassel, Germany) Assessing the causal relationship between students’ aspirations and school success – a comparison of analytical procedures Background: This presentation consists of an analysis of the relationship between students’ aspirations and school success. It aims to test the hypothesis that aspirations have a causal effect on students’ success in school, which is measured by students’ marks in German and math. Different statistical procedures have been utilized to test this hypothesis. These statistical procedures include ordinary least squares (OLS) regression, pooled OLS, and random and fixed effects models. The different effect estimates of students’ aspirations are presented and discussed in the context of assumptions underlying the various methods.

Theory This contribution draws on the Wisconsin Model of Status Attainment (Sewell et al., 1969, 1970). In the context of this model, aspirations can be defined as a “cognitive orientational aspect of goal-directed behavior” (Haller, 1968, p. 484). The model assumes that families’ social background influences students’ achievement in school by affecting their aspirations and behaviors. According to this assumption, students with high educational aspirations and a beneficial learning behavior are more likely to obtain good school results than those with less advantageous attitudes and behaviors.

Methods: In the case of OLS, students’ marks in German and math were regressed on their educational aspirations and other explanatory variables in a cross-sectional sample. Longitudinal data were utilized for pooled OLS regression, and random and fixed effects regression. As in the case of OLS, students’ marks in German and math were regressed on their educational aspirations and other explanatory variables. Finally, the estimates obtained from all the regression types for the explanatory variables of interest were compared with each other.

Data:

The data source consists of the first three waves of the National Educational Panel Study (NEPS), starting cohort, Grade 5. The NEPS sample includes a total of 4,989 students from regular schools. Students attending special schools were excluded due to limited information about their educational aspirations. The data are collected annually and are based on paper and pencil questionnaires from students and computer-assisted telephone interviews with their parents. Students who visited a primary school or orientation stage between 2010 and 2012 were excluded from the sample.

Results and conclusion: The results show that the estimated effect varied between the various regression types. This was particularly true for the time-varying regressors of interest, such as students’ educational aspirations and whether between or within variation was used in the analysis. All regression types overestimated the effects of time-varying regressors in comparison with the fixed effects model. Whereas in the OLS, the pooled OLS, and the random effects model the estimates of some of the time-varying variables contributed significantly to an explanation of the dependent variable (students’ school results), in the case of the fixed effects model, these estimators were no longer significant. The example provides some practical evidence of how easily misleading conclusions derived from statistical analysis can be – conclusions that may influence policy recommendations. Keywords Causal Inference, Social Science Research, Survey Data

Züll, C. and Scholz, E. (GESIS, Leibnitz-Institut für Sozialwissenschaften, Abteilung Survey Design and Methodology, Mannheim, Germany) Effects of Left-right Scale Design in a Cross-national Perspective Left-right self-placement on a uni-dimensional scale is one of the standard questions in many social and political surveys to measure respondents’ ideological orientation in a minimalist way. Although the left-right scale is a standard question, the design of the scale is not standardized across surveys. One aspect of scale design is the offer of a midpoint. This paper is about design effects on central left-right scale placement in a cross-national context. How do respondents answer if there is no true midpoint? Do respondents who want to express a middle position, in the case of a 10-point scale, use scale middle categories to substitute a true midpoint? Offering a midpoint is much debated among political and survey researchers and quite often, a midpoint serves as a hidden “don’t know” or a missing attitude. Does nonresponse increase when non-attitudes cannot be expressed by choosing the neutral midpoint to hide non-response? If middle categories in the 10-point scale work as a substitute for a true midpoint in the 11-point scale, non-response will not differ. In addition, we investigate whether social desirability will affect response behavior, in particular if a scale without a midpoint is asked where respondents who want to place at the scale middle are forced to decide to either go into the left or the right direction of the left-right scale. Does social desirability push into the same or into different directions in cross-national perspective? Our assumption is that the direction will differ according to what is desirable or acceptable, and what is acceptable might differ in cross-national perspective.

Methodological research using a German SOEP experiment found that the 10-point and the 11-point scale are similar in non-response and perform well with regard to reliability but that the 11-point scale seems to work better than the 10-point scale in terms of substantive analyses.

Are these German findings robust across countries? We run a split-half experiment where either a 10-point left-right self-placement scale or an 11-point scale was asked in an experimental online survey fielded in Canada, Denmark, Germany, Hungary, Spain, and the U.S. and asked probe questions on the meaning of left and right. Preliminary results seem to confirm the idea that respondents who are in favor of choosing a scale middle find a virtual

center in the 10-point left-right scale and non-response does not show significant differences between 10-point and 11-point scale. In addition, the idea of a push into the social desirable direction in the 10-point scale seems to be confirmed by our data while the effect is not very large. In summary, our analyses indicate that the left-right scale does not work similarly across countries and design effects are not always clear. These different results across countries may indicate that there are different understandings of the left-right scale in a cross-national perspective which makes its comparative use difficult. However, it seems that the 11-point scale works better due to the social desirability effect in the 10-point scale.

Keywords: response behavior, left-right self-placement, cross-cultural comparison