1

Working Paper prepared for AUSTRIAN RESEARCH ASSOCIATION, WORKING GROUP ON : “ and Democracy” Research Conference, November 9-10, 2012

Measuring How People View Democracy. Conceptual Problems and Possible Alternatives.1

Norma Osterberg-Kaufmann

Abstract A number of standardized surveys make comparable data available on attitudes of people towards the principles and institutions of democracy and identify a world-wide overwhelming support for democracy. But faced with numerous unexpected findings, the credibility of this predominant support for democracy is questionable. The paper will discuss methodological challenges of survey-research, which are rooted particularly in the phenomena of linguistic and cultural equivalence. Facing the limitations of pure qualitative or quantitative methods, the author will argument for the need for alternatives for surveying attitudes and will introduce Repertory Grid as a mixed- method study. This method has been developed particularly to measure personal constructs of how people view their environment, personal mindsets and opinions. The specific advantage of this method is that the initial data is qualitative, using the language, the words and the associations of the respondents, while the analysis uses statistical methods, which allows a comparison and generalisation of the concepts and meanings.

1. Research Problem Measuring people’s attitudes toward democracy is a key question of current research on democracy as they are crucial for a stable and/ or the chances for . The continuous importance of a deeper understanding of democratization can be illustrated by a number of examples such as the transformation to democracy in the post-socialist Europe, the EU- integration of those countries, democratic or authoritarian stagnation (Merkel 2010), failed external democratization efforts (Grimm 2010) and last but not least the unknown future of the countries of the Arab Spring. Facing the difficulties illustrated by these examples, one fundamental question has to be, whether we are right with the assumption of a universal and homogenous understanding of what democracy means. Or do different cultural and socio- economic preconditions lead to differences in the understanding in different countries and therefore as well to different expectations what democracy and their institutions should achieve. This knowledge would have fundamental consequences for future like the EU-integration of the remaining candidate countries. A broader understanding of these issues

1 Please do not cite, circulate, or copy without authors’ permission. 2 would facilitate the understanding of the evolution of democracy even in well- established , as even there the nature of participation (protests, networks, direct and liquid democracy, etc.) and legitimacy has changed dramatically in the past decades. Therefore it is important to all theorist and practitioners to figure out whether the understanding of and expectations on democracy changed as well and if so, into which direction. A number of standardized surveys make comparable data available on peoples’ attitudes towards the principles and institutions of democracy (Diamond 2008; Welzel/Inglehart 2008). Faced with numerous unexpected findings, such as a more favourable opinion on democracy in Albania and Azerbaijan than in Switzerland or Sweden (Welzel/Inglehart 2008), the credibility of the worldwide overwhelming support for democracy is questionable (Inglehart 2003; Pickel 2006; Dalton/Shin/Jou 2007). Even studies, combining the support for democracy with additional alternative variables such as support for strong leaders or army in governmental accountability, generated contradictory results. These inconsistencies led to the exclusion from various analyzes of certain autocratic ruled countries (Pickel 2006). Considering the number of outliers in various researches, the statement of overwhelming positive attitude towards democracy looses its significance, as it remains valid only for Western Europe and North America. Before analyzing the support of democracy we must clarify whether the meaning of democracy is homogenous (Dalton et al. 2008). Neither by qualitative (Dalton et al. 2008) nor by quantitative research (Inglehart/Welzel 2005; Mattes/Bratton 2007; Welzel/Inglehart 2008) has appropriate explanations for the above-mentioned “unexpected findings” (Welzel/Inglehart 2008) and for the contradictions between empirical observations and theoretical research on democracy. The described problems highlight the methodological challenges for survey- researches, in particular the phenomenon of linguistic and functional equivalence. It becomes an obstacle especially for cross-culture and cross- country studies, as there is probably neither a comparable understanding of the questions or an equal meaning of the used terms or their associations (Lauth/Pickel/Pickel 2009). In addition, because of subjective bias by social 3 desirability and paying lip service, it is doubtful if expressed opinions about social issues are reliable (Pickel 2006). The working paper will review in a state of the art chapter the different approaches on measuring how people view democracy, regarding both quantitative and qualitative methods. By weighting up the respective strengths and weaknesses of both approaches the working paper will suggest a mixed method design to answer the question on people’s attitudes in a more sophisticated manner. The Repertory Grid Interview will be introduced as an appropriate alternative to pure quantitative or pure qualitative methods.

2. State of the Art 2.1 Quantitative: Survey Research The end of the thesis of worldwide overwhelming support of democracy Since Almond and Verba (1963) analyzed the attitudes of citizen in five countries in the 1960s, there is today, thanks to survey research, data available for about 90% of the world’s population. Numerous cross-country and cross-culture surveys2 measure people´s attitudes towards the principles and institutions of democracy. The analyzes thereby make a differentiation between the principal support for democracy as the best form of government3 and the specific support of the democratic political system4 (Pickel 2006). Following the data of survey research, the worldwide support of democracy is overwhelming (Inglehart 2003, Pickel 2006). The support of democracy continuously is around 90% in the majority of countries covered by the World Value Survey over longer survey periods. Following the assumption, that societies expressing themselves supportive to democracy are a democracy (Inglehart 2003), those results are surprising. When looking at the examples mentioned at the beginning that there are higher democracy support rates in Albania than in Sweden for example (Welzel/Inglehart 2008), the

2 World Value Survey, Latinobarometro, Afrobarometer, European values Survey, Eurobarometer, Barometer and Central and Eastern Europe Barometer 3 Principal support is measured by the interviewees opinion if a specific kind of is very good, good, bad or very bad, by the agreement to the statement one should have a democratic political system ort he agreement to the statement that democracy may have its problems but is better than any other kind of government. 4 The specific support of democracy concerns the particular situation of a country, government or current situation. 4 meaningfulness of this support is questionable. Also the high support rates in China should be interpreted with caution, because how people view democracy may differ from the common understanding used in political culture research (Pickel 2006). Political culture research follows the definition of Robert Dahl (1989) who emphasized mainly the representative elements of democracy. The definition focuses on the representatives, which are elected in free and fair and are accountable for all decisions made in a democracy. When adding variables such as the support for anti-democratic alternative, the contradictions are even stronger. A divergent understanding of democracy in contrast to the definition used in political culture research and the heterogeneous understanding from one country to the other become obvious when analyzing the survey results. There are countries equally supporting democracy and anti-democratic alternatives. In pre-Arabian Spring Egypt for example, the involvement of military in government was not seen as contradictory to democracy. In Georgia, people do not see a contradiction between the compliance with democratic principles and having a strong leader in government (WVS 2005/2008). How people view democracy in those countries vary obviously from Dahl´s (1989) definition (see Table 1). Another surprise arises when analyzing people’s opinion about the democratic performance of their in contrast to the quality of democracy measured with indexes like Freedom House. Linked to the question on the satisfaction with democracy in the own country and the evaluation of the current political system, reflecting an abstract support for democratic ideals, is in how far this support also translates into political culture. Interestingly, the data on the question to what extent the own country is perceived as democratic gives a ranking of top 20 of “felt democracy”, composed of countries like Norway, Swiss or Sweden but as well Vietnam or Jordan. The latters has been categorized by Freedom House (2008) as partly free and not free. In addition, the Vietnamese and Jordan people think to a higher degree to live in a well working democracy than the Swedish people think about their country (see Table 2).

5

Table 1: Support of democracy and support of anti-democratic alternatives

„Having a „Having a strong „Having the army democratic leader who does rule" political system“ not have to bother with parliament and elections“ Egypt 98,3 16,0 56,6 Gerorgia 98,1 53,9 14,1 Sweden 97,9 18,0 5,7 Ethiopia 97,7 31,2 24,8 Norway 97,4 15,8 5,4 Indonesia 96,8 23,7 95,5 Andorra 96,6 8,0 2,4 Morocco 96,4 26,7 28,1 Spain 96,2 32,5 11,8 Ghana 96,2 21,9 16,3 Jordan 96,1 18,6 76,0 Switzerland 96 16,4 2,6 Italy 95,4 14,5 6,5 Burkina Faso 95,4 47,9 38,4 Cyprus 95,1 41,9 16,1 Argentina 95 40,9 11,9 Romania 95 78,3 20,4 95 16,9 3,2 New Zealand 94 18,7 4,0 China 93,7 36,1 36,2 Source: World Values Survey 2005

Table 2: Evaluation of the current democracy

„And how democratically is Freedom House Index 2008 this country (own) beeing governed today?“ Ghana 8,5 1/2 Free Norway 8,1 1/1 Free Vietnam 8,0 7/5 Not Free Jordan 7,8 5/4 Partly Free Uruguay 7,7 1/1 Free Switzerland 7,6 1/1 Free Sweden 7,5 1/1 Free Spain 7,4 1/1 Free South Africa 7,4 2/2 Free Canada 7,2 1/1 Free Australia 7,1 1/1 Free Argentina 7,1 2/2 Free Chile 7,0 1/1 Free Thailand 7,0 6/4 Partly Free Malaysia 7,0 4/4 Partly Free Mali 7,0 2/3 Free Japan 6,9 1/2 Free Taiwan 6,9 2/1 Free Finland 6,7 1/1 Free China 6,7 7/6 Not Free Source: World Values Survey 2005, Freedom in the World 2008 6

Similar problems arise when interpreting older data sets5 which led to the exclusion of the contradictory cases from the analyzes. The data of those mostly autocratic ruled countries highly satisfied with their governments democratic performance was simply not interpretable for the researcher (Pickel 2006). As a consequence, there are strong limitations to survey research in and by excluding autocracies from structural analyzes, the researcher is confronted with a significant loss of information. Due to the systematic exclusion of these countries, a meaningful comparison is not possible any more (Pickel 2006). Therefore the statement of a worldwide overwhelming support of democracy cannot be maintained. Facing the huge number of outliers, this thesis can only be kept up for Europe and North America.

Critics and reactions by established measurement of democracy The end of the thesis of worldwide overwhelming support of democracy does not question the cohesion of support of democracy in a society and the democratic constitution of a state, but the postulate of universality of democratic values. The discussed biased results and subsequently the limited meaningfulness of current survey research is based on the assumption of a homogenous understanding of democracy. Literature discusses the effect of social desirability and paying lip service as causes for the limited meaningfulness of these survey research. The assumption is, that this phenomenon appears stronger in autocratic, closed societies than in liberal societies (Pickel 2006). Basing on the hypothesis that certain “mass attitudes are very strongly linked to democracy” (Inglehart/Welzel 2008: 22) Inglehart (2003) and Inglehart/Welzel (2005, 2008) argued for a better selection of indicators in order not to measure lip services but to assess how deeply democracy is rooted in the respective societies. Inglehart and Welzel (2005, 2008) discuss alternative indicators to measure the foundations of democracy in a society such as the extent of trust, political participation and post-materialistic values. They summarize these indicators as self-expression-values, which are closely related to the shape of . The self-expression-values are contrasted by the survival-

5 WVS 1999/2000, Afrobarometer 1999/2000 7 values, which emphasize the opposite (Inglehart/Welzel 2005: 48ff). The authors concluded that self-expression-values are closely linked to democratic institutions, quality of democracy and effective democracy. The self-expression- values, which are not directly linked to democracy, have been much stronger predictors to democracy than the explicit inquiry of democratic support (Inglehart/Welzel 2005: 149ff.). Interestingly the group of countries with weak self-expression and strong survival-values is to a great extent congruent to the group of countries6 Pickel (2006) had to exclude from his analysis. This shows the impact of different norms and values on the understanding of democracy. These findings can be directly related to the even more complex problem of equivalence. Equivalence is always an issue in comparative surveys and becomes an urgent question in cross-country and cross-culture comparisons. This issue reveals the central problem of political culture research. How to operationalize the original research design in an appropriate manner when facing cultural and linguistic differences between the cases. The first point with respect to equivalence in survey research touches upon the “differences in the subjective use of ordinal scales” (Wand 2012: 1). Traditionally, researchers assume “that all respondents interpret subjective scale categories in exactly the same way” (Wand 2012: 1). This practice is questionable, “because respondents may differ in how they interpret scale categories” (Wand 2012: 1). Researchers in survey methodology and cross- culture psychology worked out, that there are countries where individuals tend to select one side of the scale more often than the other. In most cases this phenomenon concerns the side indicating agreement. Other phenomena are countries or cultures producing extreme responds choosing mainly the extreme ends of the scale or only the middle part (Stegemueller 2011: 3). Consequently, even if individuals of different countries or cultures may share preferences or attitudes, they probably answer differently, so the variable lacks equivalence and the data is hardly comparable. A common method to deal with the heterogeneous use of ordinal scales is setting anchoring vignettes (King et al. 2004, King/Wand 2007, Wand 2012). “By

6 Examples for this group of countries are China, Vietnam, Iran, Uganda, Egypt, Nigeria, Botswana, Jordan and Bangladesh. 8 observing how each individual rates a common set of vignettes, a researcher may discern differences in scale use and thereby adjust the meaning of each individual´s self-evaluation” (Wand 2012: 1). Anchoring vignettes have to be added additionally to the initial survey questions, so their application is very costly and not prevailing yet (Lauth et al. 2009a). Anchoring vignettes answer the problem of different understanding and “use of ordinal response categories – such as (1) strongly disagree, (2) disagree, (3) neutral, (4) agree or (5) strongly agree – in different ways” (King/Wand 2007: 46). But the problem of linguistic and cultural equivalence of terms and concept in the questions themselves cannot be solved with anchoring vignettes. In cross- country and cross-culture studies one cannot assume a comparable understanding of the questions, the equal meaning of words or equal associations of the respondents. Particularly affected are subjective or abstract terms and concepts such as democracy. It is uncertain if such concepts will be interpreted in the same way in different countries or different cultures. Therefore it is uncertain if the same instruments effectively measure the same issue in different cases (Lauth et al. 2009a). Well-established solution strategies for linguistic and cultural equivalence are careful translation and back translation, careful wording of the questions, focus groups or cognitive debriefing (King/Wand 2007: 46). The aim of translation and back translation for example is to identify the differences between linguistic and functional equivalence. But this procedure cannot exclude that a different understanding will bias the answer either (Lauth et al. 2009a). Nevertheless researchers ignore the varying understanding in the different research cases and assume big similarity so that the acquired results become meaningful. Back translation does not differentiate between linguistic and cultural equivalence. This procedure has proved to be successful for the Western world. But particularly in studies going beyond Western Europe and North America, the problem cannot be solved in that way. But researchers (Lauth et al. 2009a) currently do not see any better solution and assume that this procedure will even dominate in future. To sum up, no matter the specific research topic, a decisive advantage of quantitative methods is to acquire and to analyze huge amounts of data. 9

Quantitative data thereby meets the requirements of representativity. By standardizing the single interview, the comparison of the data on a macro level becomes possible and allows for generalized results. In addition, data based on standardized interviews is comprehensible and the results are replicable (Jahn 2009, Lauth et al. 2009b, Pickel 2009). But at the same time, quantitative methods imply problems, such as the discussed phenomena of social desirability, paying lip service and problems of equivalence. It is questionable if results of standardized surveys on social facts are substantiated, given the distortions of social desirability and paying lip service (Pickel 2006). As well the problems concerning linguistic and cultural equivalence and particularly their handling in research praxis are serious sources of error, which produces scientific artifacts. The price of standardization and formalization is the loss of context of the single cases with its specific surrounding and latent structures of meaning (Pickel 2009, Pickel/Pickel 2009).

2.2 The qualitative approach Methodologically different from the quantitative approach discussed above, Dalton/Shin/Jou (2008) continued to analyze the substance of the worldwide overwhelming support of democracy. The focus of their research was explicitly the question of how people understand democracy. The authors examined if there is at all an understanding of democracy in African countries and if there is, in which manner. Therefore they analyzed open questions on the meaning of democracy, which have been collected, in addition to the standardized questions, in major cross-national surveys7. Exactly this reference to the single cases is the main premise of qualitative research. To decode the subjective sense of human action, the subject consequently is in the center of research. The best possible openness and little influence by questions and predefined answers, active articulation and formulation of opinions and attitudes with the interviewees own words shall limit the problems of social desirability and paying lip service (Pickel/Pickel

7 Post Communist Citizen Survey 1990, Afrobarometer 2000-2001, East Asia Barometer, Latinobarometro 2001, „Citizens Views of Democracy in Latin America“, „Popular Conceptions of Democracy in Postcommunist Europe“, New Soviet Citizen Project, Asia Foundation (Dalton/Shin/Jou 2008) 10

2009). In addition, the problem of equivalence has a lower impact in open interviews as in the case of conceptual contradictions, the interviewer can ask for clarifications. But even the qualitative interview is not completely free of these limitations. One specific problem of open question interviews are vague answers. Furthermore, the chronological order of questions, often defined when creating an interview guide, may bias the answers. And finally, an additional source of error is the coder who analyzes the open questions. The coder has to identify the meaning of the (probably vague) answers and, in doing so, adds automatically additional variability and subjectivity to the data. Furthermore, qualitative data cannot be compared on a macro level and is restricted to micro level analyzes. Finally, representativity is not given, as it is too costly to collect enough data. Gathering qualitative data as well as analyzing is much more time consuming than quantitative data (Lauth et al. 2009a, Pickel 2009). A specific limitation of the Dalton et al. (2008) data is their varying origin. They analyzed open questions on meaning of democracy, collected via different surveys, asking similar but not identical questions8. The resulting limitations of comparability are enormous. Despite these limitations, Dalton et al. (2008) assume that they achieved comprehensive results. In contemporary research there is a widely shared concern that people in unlikely places have little, vague or a meaningless understanding of democracy. Their support for democracy, so the assumption, is nothing more than the expression of the desire for western standard of living. In addition the rhetoric support for democracy is meaningless facing the usurpation of the term, even by non-democrats. But Dalton et al. (2008) assume that they achieved the surprising result that even people in the unlikely places predominantly have a liberal-democratic definition, emphasizing freedoms and liberties, followed by democratic institutions and processes and social benefits.

8 Postcommunist Citizen (1990) survey for example asked: „There is considerable argument concerning the meaning of democracy. What is your opinion about this question? What is for you the meaning of democracy?“, while the Afrobarometer (2000-2001) asked: „What, if anything, do you understand by the word ´democracy´? What comes to your mind when you hear the word?“ and the Latinobarometro (2001) asked: „To you, what does ´democracy´ mean? What else?“ 11

In light of the above-discussed limitations, doubts on the results are appropriate. The problem of cultural equivalence is, if at all, minimized, but not solved. Furthermore, the method does not allow a differentiation between social desirability and paying lip service and the true norms and values of a society. Even an afghan farmer is able to repeat what he has heard from other people, TV or radio. The will of giving the right answer is deeply rooted in the human psyche. Finally, even the authors (Dalton et al. 2008) of the study are not free of concerns regarding their own results. “It is surprising that a large proportion of the public in developing nations defines democracy in liberal-democratic terms. Even if these are only ´questionnaire democrats´- an interpretation that we reject – it is still remarkable that citizens with limited democratic experience know the ´correct´ answers.” (Dalton et al. 2008: 6f) The existence of questionnaire democrats is an interesting insight indeed, but does not explain how people view democracy, how deep democratic norms and values are rooted in societies and what foundation formal democratic institutions in the respective countries really have.

2.3 Quantitative and Qualitative Methods The different strategies to measure democratic attitudes and how people view democracy reflect the general debate in comparative about pros and cons of qualitative and quantitative methods and the ideal methodological way. But qualitative and quantitative researchers agree that a quantification of the results is necessary to reach the required generalization of statements. Comparable evidence about citizen between macro units is only possible by standardized interviews at all (Lauth et al. 2009a). The résumé in the literature is, that an empirical profound comparative research on values, attitudes and behavior is only possible by standardized interviews (Pickel 2009: 303). The most important purpose of qualitative research on the other hand is to provide a solid base for the design of the quantitative research (Pickel/Pickel 2009). tries to benefit from the potential of both methods and to combine them in mixed-methods-designs. The combination of qualitative and quantitative methods in one study allows a better understanding of the research problem than one of the two methodological approaches on its own could 12 achieve (Pickel 2009). Research designs are possible where the qualitative or quantitative approach will dominate, where both will have equivalent status or multi-level-designs, where the respective approach will be applied in different sequences. Mixed-model-studies are discussed as a special approach of mixed- methods-designs, where already during the different stages of the research process both methods will come into effect. That means, there will not be simply an exchange of the results of parallel-implemented studies, but a mix already within the research phase (research question, data collection, data analyzes) (Pickel 2009: 309). But up to now, this approach has not been implemented in comparative politics or social science at all.

3. Repertory Grid as Alternative Method Mixed-model studies are common in other disciplines since decades already. The psychologist Kelly (1955) developed the Repertory Grid method, as an approach to measure the horizon of experience of persons (Rosenberger/Freitag 2009) in the 1960´s. Repertory Grid is exceptional in its combination of qualitative data collection and quantitative data analyses.

3.1 Theoretical Background Data collection The philosophical basic assumption of the Repertory Grid method is, that people (re)construct reality to be in contact with the world. People anticipate events by individual connections of their own experiences. They evaluate the results of their behavior with the available personal constructs to adapt their behavior according to the environmental necessities (Jankowicz 2004). The Repertory Grid method works with the words of the respondents themselves to measure the subjective constructs of individuals. In the preparation phase of a project, the researcher sets up eight to twelve terms or examples (“elements”), which are related to the designated topic such as “Strong Leader”, “Military Governance”, “Parties”, etc. for the research on “Democracy”. In the theory of Kelly (1955), these “elements” are relevant things, situations or events to the respondent. The selection of the right and most meaningful elements is part of the pre-set research design. The definition of the used 13 elements is the most sensible part of the method, as it impacts the whole interview and the results. Therefore it is important to test the elements carefully and to readjust before starting the field interviews. With the help of personal “constructs” the respondent will compare the “elements” characteristics and will put the “elements” in relation to each other. The “elements” and “constructs” help people to structure the reality. According to Kelly (1955) constructs can be understood as dichotomous dimensions such as “good” opposed to “evil” or “warm and sunny” opposed to “cold and windy” to sort and evaluate the “elements” of an interview into groups on the basis of their similarities. Each “element” will be rated on each “construct” (Jankowicz 2004). By assessing similarities and dissimilarities the respondents “constructs” will sort particular events to particular categories and thereby will create reality to individuals (Elements and Constructs 2011). The Repertory Grid interview will provide the researcher with “a kind of mental map” (Jankowicz 2004: 14) of how people think and see the world. Based on the limited (quantitative) set of elements, respondents provide the researcher with the qualitative (actively formulated) constructs and translate their subjective perception into a quantitative grid (see examples below). There are several substantial questions the Repertory Grid method will enable to answer: 1. How different realities (elements) relates to each other? 2. How people describe (constructs) reality (elements)? 3. Understand, what people mean when talking about specific subjects and 4. Understand the meaning of the respondents words.

Data Analyzes The data collected in the interviews allows various quantitative analyzes, to answer these questions. For the manual analyzes the data will be figured in an elements and constructs grid, which are rating scales “arranged in rows and columns into a table or grid” (Jankovicz 2004: 8).

14

Table 3: Extract from a single grid interview (in german)

Ideale Regierungsform Demokratie Parteien Starker politischer Führer Armee an Regierung beteiligt Wahlen Diktatur Eigenen Land Parlament Rechtsstaat 1 2 3 4 5 6 7 8 9 10 1 2 3 9 7 3 10 2 4 1 Verlässlichkeit Unberechenbarkeit 1 1 3 7 9 3 10 3 2 4 Fair unfair 1 2 5 10 10 1 10 2 4 3 Vom Volk bestimmt Eigeninteressen-geleitet 1 3 4 9 8 5 10 4 4 1 Gleichheit Bevorteilung 4 3 7 9 8 4 10 3 1 9 Kompromiss Schwarz-Weiß 4 1 5 9 9 1 10 2 5 1 Ermöglicht Teilnahme Schließt aus 1 3 2 7 10 1 9 2 1 5 Sie vertreten Sie sind abgehoben 7 2 2 3 9 1 6 4 3 8 Meinungsabhängig beständig 3 1 2 9 9 1 10 3 4 6 Pluralistisch einseitig 1 3 5 10 10 1 10 3 2 1 Gleichberechtigt dominant 1 4 6 10 9 7 10 6 4 2 Sachbezogen subjektiv 1 2 8 10 10 1 10 3 2 1 Fremdinteressiert Eigennutzen getrieben 10 8 2 2 1 3 1 6 6 9 Kämpferisch ausgleichend 1 2 3 7 9 2 10 2 3 2 Offen geschlossen

Within a single grid there are the possibilities to analyze the simple relationships between elements, the simple relationships between constructs, do cluster analyzes and principal component analyzes to get inside the structure of the data (Jankovicz 2004: 94 ff). What is more interesting than a single grid, is analyzing a larger set of interviews. To collect data of many respondents and to analyze more than one grid there are a number of software’s available9. Like in standardized surveys it is possible to transfer the individual data to an aggregated level and allow for a comparison and generalization of the results. In order to get meaningful results, the number of interviews should be three times the number of elements (usually a Grid consists of about eight to twelve elements). To reach representativity it is required to run 30-45 interviews per identified subgroup10 of a sample. As a (non-representative) example and in order to illustrate the process of data collection and analyzes, the following chapter will demonstrate the Repertory Grid method with the help of the software sci:vesco.

9 Gridcore, GridSuite, RepGrid, Idiogrid, sci:vesco for example. 10 This can be categories like male/ female, http://foodiesgarden.net/?p=1501 age >30/ age <30, urban/ countryside, academics/ non-acaemics, main relgions, etc. 15

3.2 Example inquiry: How people view democracy The Repertory Grid Interview The first step of the Repertory-Grid interview is that the respondent decides, which two out of three randomly selected elements are more similar to each other than the third one. As an example, the following elements are given: “elections”, “most democratic country of the world” and “army in governance”. The respondent decided, that “elections” and the “most democratic country of the world” are most similar and differ from “army in governance”. The second step is to define the similarity of these two elements (attribute 1) and the difference to the third element (attribute 2) with his/her own words. In the example used below, the similarity of “elections” and the “most democratic country of the world” are in the respondents view characterized by being “alterable” (attribute 1), while having the “army in government” is characterized by being “rigid” (attribute 2).11 These two attributes form the first individual evaluation construct of the respondent. In the third step the respondent positions all the predefined elements of the interview in her/his own evaluation (rating) construct (the scale between attribute 1 and 2). The rating takes place in a tetralemma field, which means besides the options between attribute 1 and attribute 2 there is as well the options of both or non of the two (Elements and Constructs 2011).

Figure 1: Rating the elements in the tetralemma field

The steps are repeated with different, randomly selected sets of three elements to collect various constructs of the interviewee and the respective distribution of

11 Authors translation of respondents constructs „veränderbar“ vs. „starr“. 16 the elements on these constructs. To receive meaningful results there are at least five interview cycles necessary.

Repertory Grid Data Analyzes The collected data is represented in an elements-and-constructs-matrix (the grid). Via statistical methods coherencies and dependencies of the elements and the constructs become obvious, the data of all respondents are comparable and aggregated to a collective grid. This collective grid allows the analysis of how people define the various elements and the common concept behind the elements – with their own words and attributes. With the help of the principle component analyzes, the numerous dimensions formed by the individual ratings, are reduced to a three-dimensional room, in which all the elements and constructs, and their relations to each other, are represented. To visualize and to understand the relations between elements and constructs and between constructs themselves, principle components analyzes will identify the three most significant components, translate them into a 3- dimensional matrix and transfer the numbers of the matrix (Table 3) into coordinates for each element and each construct. This instrument also allows building groups of elements, which are strongly related to each other. The data of the example inquiry12 illustrates this step. There are three groups of elements in the three-dimensional room, built on the basis of the respondents associations. There is one group containing the elements “”, “strong leader” and “army”, which can be characterized as the group of “autocratic elements” on the one pole of the three-dimensional room. On the other pole there is the biggest group of elements containing the “own country (Germany)”, “parliament”, “ideal form of governance”, “democracy” and the “most democratic country”. This group can be described as the “democratic elements”. Between these two poles is the third group of elements, with “constitutional state”, “elections” and “parties”. These are “institutional elements”. They are situated very closely to the “democratic elements”, but interestingly do not belong to this group in the respondents view.

12 The example inquiry is based on four Repertory Grid Interviews, the author collected as an illustration to describe the method. The data is not representative and therefor cannot provide valid information about what the Germans think about democracy. 17

Figure 2: The data in the three-dimensional room

Related to the elements, all attributes (constructs) the respondents created during the interview are represented. Just the four interviews of the example inquiry generated 40 constructs. To get inside this rich source of information there are statistical tools to help. To group not only the elements but also the constructs according to similar or close meanings for the respondents, cluster analyzes puts the elements and constructs in relation to each other. It visualizes the three groups of elements (autocratic, democratic, institutional) in a different way and tells us more about the attributes the respondents used. The cluster analyzes of the constructs shows for example that respondents use the attributes “human (menschlich)” and “not dangerous (ungefährlich)” in a similar manner. The analyses shows further that those two attributes relate the strongest to the “ideal form of government”. Other examples are the two constructs “diverse (vielfältig)” and “electable (wählbar)”, which are strongly related in the perception of the respondents. Both describe the elements “constitutional state 18

(Rechtstaat)”, “democracy (Demokratie)”, “most democratic country (demokratischste Land)” and the “ideal form of government (beste Regierungsform)” the strongest.

Table 4: Clustering the elements and constructs

19

But what does it mean when respondents characterize the “constitutional state”, “democracy”, the “most democratic country” and the “ideal form of government” as “electable”? Correlation analyzes help to understand the meaning of constructs and elements better, as they indicate coherencies between the constructs or the elements. The correlation of constructs shows for the construct “electable” that it relates strongly to “diverse” in the understanding of respondent 1. As well in the understanding of respondent 3 “electable” is related to “diverse”, even though the relation is not that strong as it is for respondent 1. But the extract of constructs correlations shows as well that by characterizing the “constitutional state”, “democracy”, the “most democratic country” and the “ideal form of government” as “electable”, they have also in mind the attributes “not dangerous”, “long-term thinking”, “multi-facetted” and “diverse”.

Table 5: Correlation table (extract) for constructs (the number in brackets refer to the number of the respondent)

To find out how respondents describe one particular element such as “democracy”, the semantic corridor will highlight the strongest constructs. The better different constructs describe an element, the narrower will be the angle between the construct and the related element. The “semantic corridor” tool will make this coherence visible. Constructs in a narrow corridor of 45° describe a collective semantic cluster to characterize the semantic axes in the room (Elements and Constructs 2011). When looking on the element “democracy” in 20 the sample survey, there are the constructs “long-term thinking”, “diverse”, “alterable” and “social” within the corridor of 45°.

Figure 5: Semantic corridor

According to people´s own understanding of democracy, Repertory Grid furthermore offers the possibility to figure out how democratic people see their own country, via the profiles of elements. It allows the researcher to see which attributes are related to two elements at the same time or which attributes are related to one element but not to another. It helps to answer questions such as what belongs to “democracy” and is missing in the “own country” or what represents the “ideal form of government” but is missing in the “own country”. In the sample, “discussing” is strongly related to “democracy”, but not seen as part of the German political system (“own country”). The “ideal form of government” is characterized as “changeable”, but Germany is not perceived as such. With these profiles, the method offers concrete approaches to improve legitimacy of a political system. Given that people who see “democracy” as the “ideal form of 21 government” characterize “democracy” with the attribute “discussing”, the discussion on different forms of participation gets an additional aspect. Or in the example of the “ideal form of government” being “changeable”, but Germany not, the reform of the German negotiation democracy becomes an issue. Based on a bigger pool of data it is possible to run cluster and group analyzes, looking into the question whether all respondents view democracy in a homogeneous or heterogeneous way. Is there a split between female/ male, Germany (East)/ Germany (West) or younger/ older respondents? Just as in traditional survey research, Repertory Gridding offers the possibility to run various sub-group analyzes, depending on the demographic data collected during the interviews.

4. Conclusion Repertory Grid, as a mixed-model-design, builds on the advantages of qualitative methods during the interview and quantitative methods during the analyzes, while reducing the respective disadvantages. Firstly Repertory Grid weakens the phenomena of social desirability and paying lip-service, as the interview itself is based on the individuals’ value context. It provides an insight into the complexity of the respondents’ entire evaluation system. Secondly the subjective evaluation based on the respondents’ own evaluation/ranking constructs eliminates the effects of linguistic and cultural equivalence within the interview and even allows to understand the different cultural and linguistic meaning of words during the analyzes phase. Finally, the qualitative data of the Repertory Grid interview is being standardized so that the individual data can be aggregated on a macro level, allowing for comparisons on a country/group level as well as generalization. The Repertory Grid Method allows looking into the meaning of abstract concepts, people’s norms, values and attitudes. What does Democracy mean for individuals, groups, cultures and countries? What are the individual and subjective levers of legitimacy of political institutions? If we understand how people subjectively define democracy, we can not only compare between groups, but also interpret their support in a more meaningful way. 22

As a next step, the author of this paper suggests to test the use of the Repertory Grid Method within the field research on Democracy. This first phase will allow identifying and if necessary adjust the set of elements that allow understanding the meaning of Democracy and testing the explanatory power on an aggregated macro level, using the group analyzes. By focusing on a specific demographic cluster in the targeted regions/countries, the number of interviews can be limited.13 Therefore one possible sample could be the comparison of how one specific social group (ex. People with an academic degree between 30-40) view democracy in the former western and eastern part of Germany and/or other EU countries. The basis for the case selection should be a strong research background in order to be able to illustrate the contrast between traditional survey based results and Repertory-Grid based analyzes and to show the benefits of reviewing the old questions of political culture research. An interesting aspect of this first phase is the analyzes if there are different cultural patterns between the two parts of Germany or even between different EU- countries? Are these patterns congruent to the former two German states or did the east/west pattern disappear during the last decades. It will be an indicator for the political culture in the unified Germany and the European Union and make the link to the legitimacy of the countries and European Unions democratic institutions. The second, much more comprehensive, phase of the project will select the cases on the basis of most different case design. The variance of systems compliance, as described in the beginning of the paper, will help to identify the cases. With the use of Repertory Grid interviews and the elements tested in the first phase, the attitudes towards democracy will be assessed. In a second step, the question will be answered if the different answering patterns in existing research can be explained by the different understanding of democracy. Furthermore a cluster analysis will allow building a typology of understanding of democracy. This typology will be tested with the different variables of the respective theories of research on democracies and the research on legitimacy of rule. On the basis of this foundation a new evaluation and interpretation of the existing survey data

13 In order to get meaningful results, the number of interviews should be 3 times the number of elements (usually a Grid consists of about 8-12 elements). 23 will be possible. It will also be possible to re-design traditional, quantitative survey approaches on the background of this differentiated understanding what democracy means in the context of the targeted survey population. The use of the Repertory-Grid method, to answer the question in an international comparison how people understand democracy, can lead to an entirely new approach and knowledge for research on democracy and for all practitioners in the promotion of democracy and foreign aid. By knowing what people are really talking about, even though using international standardized terms and concepts, will help to find the appropriate instruments to push reforms and programs in the right direction. In general, the Repertory Grid method can bring the concept of diversity to scientists and practitioners and will help to understand that the same instruments do not automatically lead to the same results in every case, even if the reform efforts are extensive and have shown their effect in other contexts.

References

Almond, Gabriel/ Verba, Sidney, 1963: The Civic Culture. Political Attitudes and Democracy in Five Nations, Princeton.

Dahl, Robert, 1989: Democracy and Its Critics, London/New Haven.

Dalton, Russell J.; Shin, Doh C.; Jou, Willy, 2008: How People Understand Democracy, in: Diamond, Larry; Plattner, Marc F. (Hrsg.): How People View Democracy. Baltimore. 1-15.

Diamond, Larry, 2008: Introduction, in: Diamond, Larry; Plattner, Marc F. (Hrsg.): How People View Democracy. Maltimore. ix-xxvii.

Freedom House, 2008: Freedom in the World, in: http://www.freedomhouse.org/template.cfm?page=414&year=2008 (13.12.2010)

Grimm, Sonja, 2010: Erzwungene Demokratie. Politische Neuordnung nach militärischer Intervention unter externer Aufsicht. Baden-Baden.

Inglehart, Ronald, 2003: How Solid is Mass Support for Democracy - and How Can We Measure It?, in: Political Science and Politics 36, 1. 51-57.

Inglehart, Ronald; Welzel, Christian, 2005: Modernization, Cultural Change and Democracy:The Human Development Sequence. New York.

24

Jahn, Detlef, 2009: Die Aggregatdatenanalyse in der vergleichenden Politikwissenschaft, in: Pickel, Susanne/ Pickel, Gert/ Lauth, Hans-Joachim/ Jahn, Detlef (Hrsg.): Methoden der vergleichenden Politikwissenschaft. Neue Entwicklungen und Anwendungen. Wiesbaden, 173- 196.

Jankovicz, Devi, 2004: The Easy Guide to Repertory Grids. Chichester.

Kelly, Georg A., 1955: The Psychology of Personal Constructs. New York.

King, Gary; Wand, Jonathan, 2007: Comparing Incomparable Survey Responses: Evaluating and Selecting Anchoring Vignettes, in: Political Analysis 15. 46-66.

Lauth, Hans-Joachim; Pickel, Gert; Pickel, Susanne, 2009: Methoden der vergleichenden Politikwissenschaft. Wiesbaden.

Mattes, Robert; Bratton, Michael, 2007: Learning about Democracy in Africa: Awareness, Performance and Experience, in: American Journal of Political Science 51, 1. 192-217.

Merkel, Wolfgang, 2010: Are returning? Revisiting the 'democratic rollback' hypothesis, in: Contemporary Politics 16, 1.17–31.

Pickel, Gert, 2006: Die Einschätzung der Demokratie im internationalen Vergleich, in: Pickel, Gert; Pickel, Susanne (Hrsg.): Demokratisierung im internationalen Vergleich. Neue Erkenntnisse und Perspektiven. Wiesbaden. 267-300.

Pickel, Gert, 2009: Der Einbezug des Individuums in die Länderanalyse – Umfrageforschung und vergleichende Politikwissenschaft, in: Pickel, Susanne; Pickel, Gert; Lauth, Hans-Joachim; Jahn, Detlef (Hrsg.): Methoden der vergleichenden Politikwissenschaft. Neue Entwicklungen und Anwendungen. Wiesbaden, 298-315.

Pickel, Gert; Pickel, Susanne, 2009: Qualitative Interviews als Verfahren des Ländervergleichs, in: Pickel, Susanne; Pickel, Gert; Lauth, Hans-Joachim; Jahn, Detlef (Hrsg.): Methoden der vergleichenden Politikwissenschaft. Neue Entwicklungen und Anwendungen. Wiesbaden, 441-464.

Pickel, Susanne; Pickel, Gert; Lauth, Hans-Joachim; Jahn, Detlef (2009): Differenzierung und Vielfalt der vergleichenden Methoden in den Sozialwissenschaften, in: Pickel, Susanne; Pickel, Gert; Lauth, Hans-Joachim, Jahn, Detflef (Hrsg.): Methoden der vergleichenden Politik- und Sozialwissenschaft. Neue Entwicklungen und Anwendungen. Wiesbaden. 9-26.

Rosenberger, Matthias; Freitag, Matthias, 2009: Repertory Grid, in: Kühl, Stefan; Strodtholz, Petra; Taffertshofer, Andreas (Hrsg.): Handbuch Methoden der Organisationsforschung, Wiesbaden. 477-496.

25

Elements and Constructs (Hrsg.), 2011: sci:vesco. Handbuch zur Version 3.0. Leipzig.

Stegemueller, Daniel, 2011: Apples and Oranges? The Problem of Equivalence in Comparative Research, in: Political Analysis 19. 471-487.

Wand, Jonathan, 2012: Credible Comparisons Using Interpersonally Incomparable Data: Non-parametric scales with anchoring vignettes, in: American Journal of Politics, Early View (Online Version of Record published before inclusion in an issue) 25 Jun 2012.

Welzel, Christian; Inglehart, Ronald, 2008: The Role of Ordinary People in Democratization, in: Diamond, Larry; Plattner, Marc F. (Hrsg.): How People View Democracy. Baltimore. 16-30.

WORLD VALUES SURVEY, 2005 OFFICIAL DATA FILE v.20090901, 2009. World Values Survey Association (www.worldvaluessurvey.org). Aggregate File Producer: ASEP/JDS, Madrid.