20141114 D2 5 Citizenintegration Emergent

EMERGENCY MANAGEMENT IN SOCIAL MEDIA GENERATION

Deliverable 2.5 Continuous Citizens and EMS Involvement by Social Media

Christian Reuter1, Marc-André Kaufhold1, Thomas Spielhofer² University of Siegen1, Tavistock Institute²

April 2017

Work Package 2

Project Coordinator Prof. Dr.-Ing. Rainer Koch (University of Paderborn)

7th Framework Programme for Research and Technological Development COOPERATION SEC-2013.6.1-1: The impact of social media in emergencies

D2.5: Continuous citizens and EMS involvement over Social Media, Version V2, PU

Distribution level Public (PU)

Due date 30/04/2016

Sent to coordinator 30/04/2017 (M37)

No. of document D2.5

Title Continuous Citizens and EMS Involvement by Social Media

Status & Version Final

Work Package 2: Impact of Social Media in Emergencies

Related Deliverables D2.2, D3.6, D3.7

Leading Partner University of Siegen

Leading Authors Christian Reuter, University of Siegen Marc-André Kaufhold, University of Siegen Thomas Spielhofer, Tavistock Institute

Contributors Christian Reuter, Thomas Ludwig, Therese Friberg, Sylvia Pratzler-Wanczura, and Alexis Gizikis (section 3.1) Christian Reuter, Thomas Ludwig, Marc-André Kaufhold, and Thomas Spielhofer (section 3.2) Christian Reuter, Christoph Amelunxen, and Matthias Moi (section 3.3) Marc-André Kaufhold and Christian Reuter (section 4.1) Christian Reuter and Thomas Spielhofer (section 4.2) Christian Reuter, Marc-André Kaufhold, Inken Leopold, and Hannah Knipp (section 4.3)

Reviewers Rajendra Akerkar, WNRI

Keywords Social Media, User Involvement, Citizens, Emergency Services

This project has received funding from the European Union’s Seventh Framework Programme for research, technological development and demonstration under grant agreement no 608352.

I D2.5: Continuous citizens and EMS involvement over Social Media, Version V2, PU Table of contents

Glossary ...... III 1 Introduction ...... 1 1.1 Abstract ...... 1 1.2 Purpose of the document ...... 1 1.3 Target audience ...... 1 2 Continuous User Involvement in EmerGent ...... 2 2.1 Theoretical Approaches of Continuous User Involvement ...... 2 2.2 Conducted Methods with Emergency Services ...... 3 2.3 Conducted Methods with Citizens ...... 3 2.4 Conducted Methods with Emergency Services and Citizens ...... 3 3 Continuous Integration of Emergency Services ...... 4 3.1 Qualitative Interview Study: Current and Potential Use of Social Media ...... 4 3.2 Quantitative Online Survey: Attitudes of Emergency Service Staff towards Social Media ...... 4 3.3 Qualitative Evaluation of Alerts and Notifications for Emergency Management . 5 3.4 Workshops wit Emergency Services: EAB Workshop I & II, Final Workshop ...... 5 4 Continuous Integration of Citizen ...... 7 4.1 Qualitative Case Study on Social Media for Volunteer Communities ...... 7 4.2 Quantitative Online Survey on Citizens’ Attitudes towards Social Media ...... 7 4.3 Study on Distribution, Use and Public Views on Crisis Apps ...... 8 5 Continuous Integration of Emergency Services and Citizens ...... 9 5.1 Comparative Case Studies on Flood Events ...... 9 6 Appendix ...... 10 7 Conclusion and Outlook ...... 11 References ...... 12

II D2.5: Continuous citizens and EMS involvement over Social Media, Version V2, PU Glossary

Abbreviation Expression

EmerGent Emergency Management in Social Media Generation

EMC Emergency Management Cycle

ES Emergency Service

EMS Emergency Management Service

FB Facebook

VOST Virtual Operations Support Team

III D2.5: Continuous citizens and EMS involvement over Social Media, Version V2, PU 1 Introduction

1.1 Abstract User involvement is an important component in the development of effective software engineering system. Especially while designing software for safety-critical settings the involvement of end-users is crucial. Within the project, end-users, such as citizens and emergency services, have been involved a lot. This deliverable describes how end-users have been involved. Furthermore, we describe results on citizens’ perception of social media as well as the results on a study on the distribution and user of crisis apps.

1.2 Purpose of the document The task 2.5 will consciously involve citizens and EMS in the impact analysis, the analysis regarding integration potentials of social media into phases of the EMC and the development of software components using a living lab approach as a systematic user co-creation approach integrating research and innovation processes. • Acquisition of users who will participate long-term (the whole duration of the project) in the tasks. • Development of basic infrastructures (e.g. social networking system) to maintain and support the relationship. • Continuous integration of EmerGent interventions (concepts and prototypes). • Improvement of EMS’ work practices and the involvement of citizens in the EMC by involvement of social and  to deliver input for the impact assessment in Task 2.2 and Task 2.3.

1.3 Target audience Researchers and practitioners in the field.

1/13 D2.5: Continuous citizens and EMS involvement over Social Media, Version V2, PU 2 Continuous User Involvement in EmerGent

2.1 Theoretical Approaches of Continuous User Involvement User involvement in a design process is a necessary point to understand and figure out the user needs and desires (Eriksson et al., 2006). Therefore, fieldwork and ethnographical studies are essential either to compare the users’ statements and descriptions of their environment with the real status quo or to gain a deeper understanding of the habits and working routines of the users’ everyday life. It can be assumed that there is a difference between what people say, what people say they do and what they actually do (Randall et al., 2007). The user itself has become the most important component in modern development processes. Getting to know the users thoughts, ideas and knowledges in field studies, the developer profits by the users experiences in order to invent new solutions (Eriksson et al., 2006). As the first ideas about smart homes occurred, the need for a real home environment for getting representative testing results became much more important. The concept Living Labs first was established by William Mitchel at the MIT Media Lab to represent a user-centered methodology for research in the terms of sensing, prototyping validating and refining complex solutions in multiple real-world conditions (Eriksson et al., 2006). Besides the technological advantages of Living Labs like detailed recording and observing methods, allow deep insights into the users attitude towards new technology, which is crucial for the success of ideas based on high technology (Eriksson et al., 2006). The difference between the Living Lab approach and other user-centered methods is the idea of breaking up the trial and error process and adopt a collaborative design process between users and developers working closely together (Eriksson et al., 2006). Especially for a long-time user involvement the concept of Living Labs has been a proven method. Offering the space and the possibilities to create ideas and to experience from the real use context, Living Labs provide an active role to the user and stimulate the ongoing research during a project (Ogonowski et al., 2013). The idea of staying in touch with the participants is a productive way to provide a positive feeling of integration and helps the participants to identify themselves with the course of the project, which is crucial for the success of the Living Lab approach (Ogonowski et al., 2013). At the University of Siegen the PraxLabs approach combines methodologies of Living Labs with Ethnography and Participatory Design approaches. The PraxLabs of Siegen are situated in real households in everyday life contexts consisting of representative end-user groups in a long- term perspective (Müller et al., 2014). This approach provides a very close every-day proximity of the users towards the design process, as they are part of the team of designers and developers. Exploring prototypes, co-creating new ideas with team members, evaluating and experience new solutions or ideas are main opportunities of the PraxLab approach the participants shall find out (Müller et al., 2014). The assured level of detail of a field study in Living Labs or PraxLabs is, because of its richness of empirical material, indispensable for the design process and therefore vaunts an great value of importance to the user involvement of design processes (Randall et al., 2007).

2/13 D2.5: Continuous citizens and EMS involvement over Social Media, Version V2, PU Within EmerGent emergency services and citizens have been involved in a continuous way:

2.2 Conducted Methods with Emergency Services  Qualitative interview study (see section 3.1)  Quantitative online survey (see section 3.2)  Qualitative evaluation of the prototype (see section 3.3)  Workshops with emergency services (see deliverable D2.6+D2.7)

2.3 Conducted Methods with Citizens  Qualitative case study (see section 4.1 and D2.3)  Quantitative online survey (see section 4.2)  Mixed method study on crisis apps (see section 4.3)

2.4 Conducted Methods with Emergency Services and Citizens  Comparative case studies (see deliverable D2.1+D2.3)

3/13 D2.5: Continuous citizens and EMS involvement over Social Media, Version V2, PU 3 Continuous Integration of Emergency Services This section presents different kinds of emergency services’ involvement (qualitative interview study, quantitative online survey, qualitative evaluation, workshops with emergency services) and our findings based on it.

3.1 Qualitative Interview Study: Current and Potential Use of Social Media Christian Reuter, Thomas Ludwig, Therese Friberg, Sylvia Pratzler-Wanczura, and Alexis Gizikis. 2015. Social Media and Emergency Services? Interview Study on Current and Potential Use in 7 European Countries. International Journal of Information Systems for Crisis Response and Management (IJISCRAM) 7, 2. “Social media is much just used for private as well as business purposes, obviously, also during emergencies. Emergency services are often confronted with the amount of information from social media and might consider using them – or not using them. This article highlights the perception of emergency services on social media during emergencies. Within our European research project EMERGENT, we therefore conducted an interview study with emergency service staff (N=11) from seven European countries and eight different cities. Our results highlight the current and potential use of social media, the emergency service’s participation in research on social media as well as current challenges, benefits and future plans. “ (Reuter et al., 2015)

3.2 Quantitative Online Survey: Attitudes of Emergency Service Staff towards Social Media Christian Reuter, Thomas Ludwig, Marc-André Kaufhold, and Thomas Spielhofer. 2016. Emergency Services Attitudes towards Social Media: A Quantitative and Qualitative Survey across Europe. International Journal on Human-Computer Studies (IJHCS) 95: 96–111. “Various studies show that social media is used in emergencies – and that in spite of possible challenges for emergency services, beneficial use cases can be identified. However, relatively little empirical data is available regarding the attitudes of emergency services towards social media, and almost none of a comparative nature. This article summarizes the findings of a survey conducted of the EU project ‘EmerGent’ with 761 emergency service staff across 32 European countries from September to December 2014. The main aims of the survey were to explore the attitudes expressed by emergency service staff towards social media for private and organizational use as well as the levels and main factors influencing the current and likely future use of social media in their organizations. Based on our results, we discuss possible enhancements of the emergency management cycle using social media.” (Reuter, Ludwig, et al., 2016) Highlights: 1. We did a survey with 761 emergency service staff across 32 European countries 2. The majority of emergency services has positive attitudes towards social media 3. It is more used to share information (44%) than to receive messages (19%) 4. A future increase is expected (74%), even more for organizations already using it 5. There is a huge gap between rhetoric (66%) and reality (23%) in social media use

4/13 D2.5: Continuous citizens and EMS involvement over Social Media, Version V2, PU 3.3 Qualitative Evaluation of Alerts and Notifications for Emergency Management Christian Reuter, Christoph Amelunxen, and Matthias Moi. 2016. Semi-Automatic Alerts and Notifications for Emergency Services based on Cross-Platform Social Media Data –Evaluation of a Prototype. In Informatik 2016: von Menschen für Menschen. “The convergence of social networking and mobile media technology is shifting the way how people communicate and gain or share information. People are using social media to a greater extent, also in emergency situations. During disasters throughout the world, such as the 2010 Haiti earthquake, the 2013 European floods, or the terror attacks 2015 in Paris and 2016 in Brussels, this has been illustrated again. Often information about disasters even finds its way faster to social media than it reaches regular news companies and emergency services. However, approaches for processing and analysing the vast quantities of data produced have even more potential. Yet many emergency services still have not found a way to put this potential to an effective use. Within our project EmerGent we are developing a system to process and analyse information from social media particularly tailored for the specific needs of emergency services. The aim is to transform the high volume of noisy data into a low volume of rich content that is useful to emergency personnel. In the first part of this paper we present our approach from a user interface perspective. The second part deals with the evaluation of the approach and the derivation of future potentials of the approach.” (Reuter, Amelunxen, et al., 2016)

3.4 Workshops wit Emergency Services: EAB Workshop I & II, Final Workshop O’Brien, T., Gizikis, A., Brugghemans, B., Spielhofer, T., Moi, M., & Friber. (2016). Deliverable 2.6: Workshops I and II. Gizikis, A., Susaeta, I. G., Spielhofer, T., & Bizjak, G. (2017). Deliverable 2.7: Workshop III. Retrieved from http://www.fp7-emergent.eu/wp- content/uploads/2017/04/20170330_D2.7_Workshop_III_EmerGent_pub.pdf The target audience for members of the EAB are domain experts and key stakeholders such as civil authorities/ES and also the commercial sector, including advisors, consultants and technology providers (O’Brien et al., 2016). While establishing the end-user advisory board it was thought that all types of emergency services i.e. Fire and Rescue, Emergency Medical Service, Police Service should be represented in the EAB, in addition to experts with and without experience in the use of social media in emergencies. “The first EAB workshop focused on the work of the project during months 4 to 7. At this stage, the main focus of the workshop was the requirements definition of EmerGent as a system (related to ‘T3.4 Requirements gathering’) and the end user based view on the potentials of social media for ES and citizens’ involvement in the EMC (related to ‘T3.5 Analysis of potentials of citizens involvement in the EMC’ and ‘T3.6 Analysis of potentials using social media for ES in the EMC’. While planning the first workshop, the first activity that was needed was an introduction to the EmerGent approach, in terms of project objectives and expected outcomes. At the same time, the consortium considered important to achieve a good balance between project presentations and discussions during the workshops. Although the importance of project presentations is not underestimated, the consortium will mostly benefit from comments, feedback, inputs and new ideas that arise during the interactive discussions. A pre-

5/13 D2.5: Continuous citizens and EMS involvement over Social Media, Version V2, PU workshop webinar with the purpose of providing an introduction to EmerGent was organised which then allowed more time for discussion during the workshop.” (O’Brien et al., 2016) “The second EAB workshop focused on the work of the project after the initial set of requirements for the EmerGent ES Interface and APP were defined and while the impact assessment and study of potential were continued during months 9 to 20. The main focus of the workshop were the different models of citizen to authorities communication (related to ‘WP3 in terms of understanding and analysing the different models of C2A communication’), demo and feedback on the ES Interface, including the refinement of requirements (related to ‘T3.4 Requirements gathering’ and ‘WP4, WP5 and WP6 in terms of receiving first feedback on the integrated components’, criteria for Information quality assessment (related to ‘T4.3 Definition and prioritising of information quality criteria and indicators’), and guidelines for ES and citizens (related to ‘T7.3 Guidelines to increase the role of social media in emergencies’).” (O’Brien et al., 2016) “Following the discussions during the 2016 Annual Conference of the ‘Vereinigung für Gefahrstoff- und Brandschutzforschung’ (VGBF), when EmerGent was first presented to this audience, it was agreed to organise the final EmerGent workshop during the 2017 VGBF conference, with title ‘Social Media & GIS in Emergency Response’. The final workshop took place during the 25th annual conference in Salzburg, Austria, between 14th and 15th February 2017. In addition to the conference participants, the consortium invited members of the EAB to the conference. The EAB has provided the opportunity to collaborate with external experts in social media and emergency management and directly involve them in the activities of the project. EAB experts shared their vision and experience with the project team, while their feedback and advice were previously captured in two dedicated end-user advisory board workshops. The information provided by the EAB was used by the project team to drive the project activities over the course of the three-year project. On 14th February, the EmerGent results were presented during the conference and the consortium organised a live simulated demonstration of the ES Interface and the App. On 15th February, the consortium continued the workshop with the EAB experts only and engaged in a discussion about the evaluation of the project results and the next steps.” (Gizikis et al., 2017)

6/13 D2.5: Continuous citizens and EMS involvement over Social Media, Version V2, PU 4 Continuous Integration of Citizen This section presents three kinds of citizens’ involvement (qualitative case study, quantitative online survey, study on warn apps) and our findings based on it.

4.1 Qualitative Case Study on Social Media for Volunteer Communities Marc-André Kaufhold and Christian Reuter. 2016. The Self-Organization of Digital Volunteers across Social Media: The Case of the 2013 European Floods in Germany. Journal of Homeland Security and Emergency Management (HSEM) 13, 1: 137–166. “For almost 15 years, social media have been regularly used during emergencies. One of the most recent, and instructive, examples of its widespread use during a large-scale scenario in Europe were the 2013 European floods. Public reporting during the event indicated, and our analysis confirms, that Twitter, Facebook (FB), Google Maps and other services were frequently used by affected citizen and volunteers to coordinate help activities among themselves. We conducted a qualitative analysis of selected emergent volunteer communities in Germany on FB and Twitter among others, and subsequently conducted interviews with FB group founders and activists. Our aim was to analyze the use of social media during this particular event, especially by digital volunteers. Our study illustrates the relevance of social media for German citizens in cases of disaster, focusing especially on the role of the moderator. Our specific emphasis was the embedding of social media in the organizing work done by said volunteers, emphasizing both the patterns of social media use and the challenges that result. We show that different social media were used in different ways: Twitter was used in the main for status updates while FB-pages were mostly intended to provide an overview. FB-groups also coordinated a multitude of activities.” (Kaufhold & Reuter, 2016)

4.2 Quantitative Online Survey on Citizens’ Attitudes towards Social Media Christian Reuter and Thomas Spielhofer. 2016. Towards Social Resilience: A Quantitative and Qualitative Survey on Citizens’ Perception of Social Media in Emergencies in Europe. Journal Technological Forecasting and Social Change (TFSC) onlinefirs. “Social media is increasingly being used during emergencies. Most available studies are focused on how citizens and/or authorities use these technologies in concrete events. However, larger quantitative studies with significant results on attitudes, needs and future plans of citizens in such events are not available - especially such of a comparative nature related to emergency services. As part of the EU project ‘EmerGent’ this article presents the findings of a survey of 1,034 citizens across 30 European countries conducted between February and June 2015 to explore citizens’ attitudes towards the use of social media for private purposes and in emergency situations. The article briefly compares these findings with a second survey conducted with 761 emergency service staff across 32 European countries from September to December 2014. The aim of the overall study is to discuss citizens’ attitudes towards social media in emergencies in order to derive challenges and opportunities for social resilience.” (Reuter & Spielhofer, 2016) Highlights: 1. We did a survey with 1,034 citizens staff cross 30 European countries 2. Social media is more used to search (43%) than to share information (27%)

7/13 D2.5: Continuous citizens and EMS involvement over Social Media, Version V2, PU 3. Emergency services are expected to monitor social media (69%) 4. Mistrust as the main barrier to using social media in emergencies 5. Very little awareness of Social Media Safety Services and Emergency Apps

4.3 Study on Distribution, Use and Public Views on Crisis Apps Christian Reuter, Marc-André Kaufhold, Inken Leopold, and Hannah Knipp. 2017. Katwarn, NINA or FEMA? Mixed-Method Study on Distribution, Use and Public Views on Crisis Apps. In European Conference on Information Systems (ECIS). “Crises, such as thunderstorms and an increasing number of (recognized) terroristic attacks in 2016, do not only lead to extensive monetary damage, but also threaten human lives and influence citizens’ perceptions of safety. In such situations, the population demands information about the damage and safe behaviour. Although some apps are available to provide this information, the number of users seems relatively low. Focussing on Germany, this study aims to research (1) the distribution of crisis apps in the population, (2) the kinds of crisis apps currently used, (3) as well as needed core functionalities of warning apps. This study analyses crisis apps by investigating their utilisation quantitatively in a snowball-based survey in Europe (n=1,034) and in a representative survey in Germany (n=1,369). Based on this, the German warning apps Katwarn and NINA (Notfall-Informations- und Nachrichten-App) and the US- American FEMA-app (Federal Emergency Management Agency) are evaluated qualitatively (n=22). The results revealed requirements, which informed the implementation of a warning app prototype. The prototype combines the identified advantages of the apps evaluated in the study, containing warnings and recommendations for action and functions to contact friends, helpers and the helpers’ organisations. The contributions of this work are findings on the distribution of warning apps in Europe and Germany (both 16%), empirically based requirements for warning apps, which can be integrated in further developments of existing apps, and a prototype for such an app. “ (Reuter et al., 2017).

8/13 D2.5: Continuous citizens and EMS involvement over Social Media, Version V2, PU 5 Continuous Integration of Emergency Services and Citizens This section presents the combined involvement of emergency services and citizens in terms of a conducted comparative case study and our findings based on it.

5.1 Comparative Case Studies on Flood Events Spielhofer, T., Junge, K., Cullen, J., Drabble, D., Hahne, A. S., Bisjak, G., … Gizikis, A. (1016). Deliverable 2.3: Impact of Social Media for EMS and citizens using EmerGent concepts. “The overall aims of study, which focused on the theme of ‘flooding’, were researching a) the impact of social media on how emergency services (ES) respond and react in an emergency and b) how citizens use social media in an emergency, how they use it to interact with each other and others, including ES, and what impact this has on them (Spielhofer et al., 2016). The pre- study is based on a ‘multiple case study’ approach which allows for exploration of the impact of social media in emergency situations through the use of a ‘replication strategy’, in which successive case examples are selected to explore and confirm or disprove the patterns identified in the initial case examples (Yin, 2013). In a preliminary analysis (Junge et al., 2014), five themes were identified, which formed the specific research questions and study objectives, which are a) patterns of social media usage, b) social media roles, c) quality of information, d) scale of social media information and e) organizational and professional factors. The case study methodology was based on the three interconnected stages of scoping, data collection, analysis and integration. The findings from the different flood case studies (2013 Elbe in Germany, 2015 Tbilisi in Georgia, 2014 Ljubljana in Slovenia, 2010 Wroclaw in Poland, 2014 Western Norway floods, and 2013/14 Wiltshire in UK) were summarized in standardized case study reports and finally analyzed comparatively using an item analysis procedure.” (Kaufhold & Reuter, 2017)

9/13 D2.5: Continuous citizens and EMS involvement over Social Media, Version V2, PU 6 Appendix Appendix I: Christian Reuter, Thomas Ludwig, Therese Friberg, Sylvia Pratzler-Wanczura, and Alexis Gizikis. 2015. Social Media and Emergency Services? Interview Study on Current and Potential Use in 7 European Countries. International Journal of Information Systems for Crisis Response and Management (IJISCRAM) 7, 2. Appendix II: Christian Reuter, Thomas Ludwig, Marc-André Kaufhold, and Thomas Spielhofer. 2016. Emergency Services Attitudes towards Social Media: A Quantitative and Qualitative Survey across Europe. International Journal on Human-Computer Studies (IJHCS) 95: 96–111. Appendix III: Christian Reuter, Christoph Amelunxen, and Matthias Moi. 2016. Semi-Automatic Alerts and Notifications for Emergency Services based on Cross-Platform Social Media Data – Evaluation of a Prototype. In Informatik 2016: von Menschen für Menschen. Appendix IV: Marc-André Kaufhold and Christian Reuter. 2016. The Self-Organization of Digital Volunteers Christian Reuter and Thomas Spielhofer. 2016. Towards Social Resilience: A Quantitative and Qualitative Survey on Citizens’ Perception of Social Media in Emergencies in Europe. Journal Technological Forecasting and Social Change (TFSC) onlinefirs. Appendix V: Christian Reuter and Thomas Spielhofer. 2016. Towards Social Resilience: A Quantitative and Qualitative Survey on Citizens’ Perception of Social Media in Emergencies in Europe. Journal Technological Forecasting and Social Change (TFSC) onlinefirs. Appendix VI: Christian Reuter, Marc-André Kaufhold, Inken Leopold, and Hannah Knipp. 2017. Katwarn, NINA or FEMA? Mixed-Method Study on Distribution, Use and Public Views on Crisis Apps. In European Conference on Information Systems (ECIS).

10/13 D2.5: Continuous citizens and EMS involvement over Social Media, Version V2, PU 7 Conclusion and Outlook This deliverable shows how users have been involved in our project. Based on a variety of methods (qualitative and quantitative) interesting results have been achieved. Further steps for the continuous citizens and EMS involvement by social media are planned in the upcoming project months.  Field trials in Dortmund and Hamburg: Based on the first evaluation (Reuter, Amelunxen, et al., 2016) presented in section 3.3, a second evaluation of the overall EmerGent system is currently running in the format of field trials with Dortmund and Hamburg. This allows EMS to test the EmerGent system within their daily routine and provide continuous feedback on the performance of the system. Also, the EmerGent app is soon provided to selected citizens, organized as Virtual Operation Support Teams (VOST), to allow the evaluation of the citizens’ perspective. At the end of each field trial, semi-structured interviews are planned with the users of the system.  Second round of a qualitative study with emergency services: It is planned to rerun the study “Attitudes of Emergency Service Staff towards Social Media” (Reuter, Ludwig, et al., 2016), which is presented section 3.2, to get additional feedback on the emergency services attitudes towards social media and to track possible intermediate changes on their perception of social media.  Representative studies with citizens in Germany and United Kingdom: There are existing studies on the citizens’ perception on social media (Reuter & Spielhofer, 2016) but, to the best of our knowledge, it seems that none of the existing studies was based on a representative sample. Thus, we are conducting representative studies with citizens in Germany and United Kingdom to get insights into their general and emergency-specific use of social media, and furthermore cover the topic of mobile crisis apps.  Comparative case studies with emergency services: The aim of this case study focuses on researching the impact of social media on how emergency services respond and react in an emergency (Kaufhold & Reuter, 2017). Based on the results deliverable 2.3, the objectives of this study comprise emergency services’ a) tools, platforms or technologies, b) organizational structures and facilitators, c) social media users and uses, d) information validation, e) staff skills and resources, and f) moderation of volunteer communities. It is intended as another ‘multiple case study’ round of successive case examples to explore and confirm or disprove the patterns identified in the initial case examples (Yin, 2013) and, if all or most of the cases provide similar results, to develop a preliminary theory that describes the phenomena (Eisenhardt, 1989). The results will be integrated into deliverable 2.4.

11/13 D2.5: Continuous citizens and EMS involvement over Social Media, Version V2, PU References Eisenhardt, M. K. (1989). Building Theories from Case Study Research. Academy of Management Review, 14(4), 532–550. Eriksson, M., Niitamo, V., Kulkki, S., & Hribernik, K. A. (2006). Living Labs as a Multi-Contextual R & D Methodology. In Proceedings of the International Conference on Concurrent Enterprising. Milan, Italy. Gizikis, A., Susaeta, I. G., Spielhofer, T., & Bizjak, G. (2017). Deliverable 2.7: Workshop III. Retrieved from http://www.fp7-emergent.eu/wp- content/uploads/2017/04/20170330_D2.7_Workshop_III_EmerGent_pub.pdf Junge, K., Spielhofer, T., Gieve, M., Cullen, J., & Drabble, D. (2014). Deliverable 2.2: Impact of Social Media on Emergency Services and Citizens. Paderborn. Retrieved from http://www.fp7-emergent.eu/wp-content/uploads/2014/05/20141219_D2.2_Impact-ofsocial-media-on-ES-and-citizens.pdf Kaufhold, M.-A., & Reuter, C. (2016). The Self-Organization of Digital Volunteers across Social Media: The Case of the 2013 European Floods in Germany. Journal of Homeland Security and Emergency Management (HSEM), 13(1), 137–166. Kaufhold, M.-A., & Reuter, C. (2017). The Impact of Social Media in Emergencies: A Case Study with the Fire Department of Frankfurt. In T. Comes, F. Bénaben, C. Hanachi, & M. Lauras (Eds.), Proceedings of Information Systems for Crisis Response and Management (ISCRAM). Müller, C., Schorch, M., & Wieching, R. (2014). PraxLabs as a Setting for Participatory Technology Research and Design in the Field of HRI and Demography. In Proceedings of the Workshop “Socially Assistive Robots for the Aging Population: Are we trapped in Stereotypes?”, Human Robot Interaction Conference. Bielefeld, Germany: Bielefeld University. O’Brien, T., Gizikis, A., Brugghemans, B., Spielhofer, T., Moi, M., & Friber. (2016). Deliverable 2.6: Workshops I and II. Ogonowski, C., Ley, B., Hess, J., Wan, L., & Wulf, V. (2013). Designing for the Living Room: Long- Term User Involvement in a Living Lab. In Proceedings of the Conference on Human Factors in Computing Systems (CHI) (pp. 1539–1548). Paris, France: ACM. Randall, D., Harper, R. H. R., & Rouncefield, M. (2007). Fieldwork for Design: Theory and Practice. London, United Kingdom: Springer. Reuter, C., Amelunxen, C., & Moi, M. (2016). Semi-Automatic Alerts and Notifications for Emergency Services based on Cross-Platform Social Media Data –Evaluation of a Prototype. In H. C. Mayr & M. Pinzger (Eds.), Informatik 2016: von Menschen für Menschen. Klagenfurt: GI-Edition-Lecture Notes in Informatics (LNI). Reuter, C., Kaufhold, M.-A., Leopold, I., & Knipp, H. (2017). Katwarn, NINA or FEMA? Mixed- Method Study on Distribution, Use and Public Views on Crisis Apps. In European Conference on Information Systems (ECIS). Guimarães, Portugal. Reuter, C., Ludwig, T., Friberg, T., Pratzler-Wanczura, S., & Gizikis, A. (2015). Social Media and Emergency Services? Interview Study on Current and Potential Use in 7 European Countries. International Journal of Information Systems for Crisis Response and Management (IJISCRAM), 7(2), 36–58.

12/13 D2.5: Continuous citizens and EMS involvement over Social Media, Version V2, PU Reuter, C., Ludwig, T., Kaufhold, M.-A., & Spielhofer, T. (2016). Emergency Services Attitudes towards Social Media: A Quantitative and Qualitative Survey across Europe. International Journal on Human-Computer Studies (IJHCS), 95, 96–111. Reuter, C., & Spielhofer, T. (2016). Towards Social Resilience: A Quantitative and Qualitative Survey on Citizens’ Perception of Social Media in Emergencies in Europe. Journal Technological Forecasting and Social Change (TFSC). Spielhofer, T., Junge, K., Cullen, J., Drabble, D., Hahne, A. S., Bisjak, G., … Gizikis, A. (2016). Deliverable 2.3: Impact of Social Media for EMS and citizens using EmerGent concepts. Yin, R. K. (2013). Case Study Research: Design and Methods. California: Sage Publications Ltd.

13/13 36 International Journal of Information Systems for Crisis Response and Management, 7(2), 36-58, April-June 2015

Social Media and Emergency Services? Interview Study on Current and Potential Use in 7 European Countries

Christian Reuter, University of Siegen, Siegen, Germany Thomas Ludwig, University of Siegen, Siegen, Germany Therese Friberg, University of Paderborn, Paderborn, Germany Sylvia Pratzler-Wanczura, Fire Department of Dortmund, Dortmund, Germany Alexis Gizikis, European Emergency Number Association, Brussels, Belgium

ABSTRACT

Social media is much just used for private as well as business purposes, obviously, also during emergencies. Emergency services are often confronted with the amount of information from social media and might consider using them – or not using them. This article highlights the perception of emergency services on social media during emergencies. Within their European research project EMERGENT, the authors therefore conducted an interview study with emergency service staff (N=11) from seven European countries and eight different cities. Their results highlight the current and potential use of social media, the emergency service’s participation in research on social media as well as current challenges, benefits and future plans.

Keywords: Emergency Services, EMERGENT, Europe, Social Media

1. INTRODUCTION AND RELATED WORK

The need of emergency services to employ with social media has risen during the last years, as long as these kinds of media are used more and more – of course also during emergencies. Social media is thereby defined as a “group of Internet-based applications that build on the ideological and technological foundations of Web 2.0, and that allow the creation and exchange of User Generated Content” (Kaplan & Haenlein, 2010).

DOI: 10.4018/IJISCRAM.2015040103

1.1. Previous Cases

Since now, the majority of research focuses on crises and emergencies in the USA and deals with Twitter (Reuter et al., 2012). Table 1 summarizes the literature in a structured way by providing an overview of studies with regard to their reference, the related case or scenario, a brief overview of the scientific contribution and a keyword. The cases are sorted by the year the event took place. They have been identified while searching in Google scholar for the keywords “social media”, “emergency”, “disaster”, “crisis”. However, due to the amount of studies, only the most project-related have been selected to provide an appropriate overview.

1.2. Previous Surveys with Citizens

Beside the several studies about the use of social media with regard to emergencies, just a few surveys on the perception of social media exist: With over 1,000 participants, a comparative study of the Canadian Red Cross (2012) aimed to figure out to what extent Canadian citizens use social media and mobile devices in crisis communication and what they expect from the emergency services both, currently and in the future. The American Red Cross (2012b) also studied citizens’ use of social media during emergencies, with 1,017 online and 1,018 telephone survey respondents. However, these surveys only focus on citizens and not on professional emergency services.

1.3. Previous Surveys with Emergency Services

Another comparative study published by the American National Emergency Management Associa- tion (NEMA) contains the results of a survey conducted in 2012 among members of emergency services from 50 Federal States of the US (San et al., 2013). The survey, which involved 505 respondents, focused on the current degree of use of social media in crisis situations by emergency services and the future development of the organizations in respect of possible use. Additional questions were asked regarding general opinions of social media and the trustworthiness of citizen-generated information. Although the respondents indicated a positive attitude towards social media in general and valued its suitability for information dissemination, 75% mentioned the requirement of verifying citizen-generated content, and otherwise questioned its credibility. However, the main barrier identified, was the lack of personnel, experience and knowledge to take on additional responsibilities. However, they argued that the “largely untapped resource” of digital volunteers could “help to alleviate some personnel issues”. The study revealed that 85% of US authorities already use social media. A further survey of 241 US emergency managers at the county level in 2014 shows that only about half of these agencies use social media (Plotnick et al., 2015). Most of them do not have any formal policies to guide their use. Of those that do have formal policies, about one quarter actually forbid the use of social media. As main barriers for communication from authorities to citizens a lack of staff, guidance and skills have been identified; main barriers for the other way around (from citizens to authorities) are staff, trustworthiness and information overload. The authors conclude that “the agencies and their representatives are not yet ready to embrace social media and use it to its fullest potential. For the most part, current social media use is for dissemination of information, not the collection of it”. Furthermore “in addition to technological advances, policy and management changes are needed as well, to remove the “red tape” (lack of guidelines or even prohibitions against use) that impedes the effective use” of social media (Plotnick et al., 2015). Flizikowski et al. (2014) present a survey within Europe, conducted among citizens (317 respondents) and emergency services (130 respondents plus 33 interviews from Finland, France,

Copyright © 2015, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global is prohibited. 38 International Journal of Information Systems for Crisis Response and Management, 7(2), 36-58, April-June 2015

Table 1. Overview of selected cases in literature

Reference Case Contribution Keyword

(Liu et al., 2008) 2004 Indian Ocean tsunami Photo repository sites were used by citizens to exchange Photo Sharing information.

( M u r p h y & 2005 Hurricane Katrina PeopleFinder and ShelterFinder Finding Jennex, 2006)

(Endsley et al., 2005 Hurricane Katrina, 2010 Indicates that the perceived credibility of Social Media Information 2014) volcano Eyjafjallajökull in information is less than of printed, official online or televised Credibility Iceland news and information from family, relatives or friends.

(Shklovski et al., 2007 Southern California Photo repository sites were used by citizens to exchange Backchannel 2008) wildfires information. communication

(Amanda Lee 2008 hurricanes Gustav and Ike Depicts differences between the use of Twitter in crises and Microblogging Hughes & Palen, the general use. 2009)

(Qu et al., 2009) 2008 Sichuan earthquake Outlines that people gather and synthesize information. Information synthesis

(Sutton, 2010) 2 0 0 8 Te n n e s s e e R i v e r Outlines the phenomena of broadcasting. Broadcasting technological failure

(Heverin & Zach, 2009 attack on four police officers Shows the ability of Twitter to organize and disseminate Types of Tweets 2010) in Lakewood, Washington crisis-related information.

( L a t o n e r o & 2009 Los Angeles Fire Public Information Officers highlight the importance of the Information Shklovski, 2011) Department information evangelist within organizations. Evangelism

(Starbird & Palen, 2009 Oklahoma Fires Highlights the role of retweeting. Collective 2010) Intelligence

(Vieweg et al., 2009 Red River Floods Highlights broadcasting by people on the ground as well as Situational 2010) activities of directing, relaying, synthesizing, and redistributing. Awareness

(Birkbak, 2012) 2010 Bornholm blizzard Shows that the geographical location and self-selection into Emergent Groups groups create different views of a crisis situation

(Starbird & Palen, 2010 Haiti earthquake Was analyzed with the help of translators and reveals the Digital Volunteers 2011) phenomenon of “digital volunteers”.

(Starbird, 2013) 2010 Haiti earthquake Examines collective intelligence as transformations of Collective information through activities. Intelligence

(Reuter et al., 2010 mass panic at the Love Outlines the need for duplex communication. Crisis Management 2012) Parade music festival in Germany, 2010 volcano Eyjafjallajökull in Iceland

(Nagy et al., 2012) 2010 San Bruno Californian gas Analysis for identifying and extracting subjective information Sentiment Analysis explosion and fire disaster by using language processing and linguistic approaches

(Starbird & Palen, 2011 Egyptian uprising Shows how the crowd expresses solidarity and does the work of Information 2012) information processing through recommendation and filtering. c o n t a g i o n a n d diffusion

(Wilensky, 2014) 2011 Great East Japan Earthquake Emphasizes the use of Twitter to provide emotional support Commuters and mentions the problem of widely publishing obsolete or inaccurate information.

(Perng et al., 2012) 2011 Norway attacks The notion of peripheral response has been developed in relation Peripheral response to emergent forms of agile and dialogic emergency response.

(Jennex, 2012) 2011 San Diego / Southwest The availability of Social Media illustrates that “contrary to Availability of Social Blackout expectations, the cell phone system did not have the expected Media availability”.

(St. Denis et al., 2011 Shadow Lake fire Shows the deployment of trusted digital volunteers as a virtual Trusted Volunteers 2012) team to support a incident management team.

(Reuter et al., 2011 Super Outbreak Distinguishes groups of twitterers, such as helpers, reporters, Volunteers 2013) retweeters, and repeaters.

(Wulf et al., 2013) 2011 Tunisian revolution Social media linked the young activists with actors in other Political cities.

continued on following page

Table 1. Continued

Reference Case Contribution Keyword

(Yang et al., 2013) 2012 hurricane Isaac Leads to knowledge which classification algorithms work best F o u r P h a s e s in each phase of emergency. o f E m e r g e n c y Management

( A m a n d a L . 2012 hurricane Sandy Shows that few departments used online channels in their Officials Social Hughes et al., response efforts and that communication differed between fire Media use 2014) and police departments and across media types.

(American Red 2012 Online and Telephone States that 25% of the participants will download an emergency Aid Organization Cross, 2012a) Survey app and 12% of the general public used Social Media in crises.

(Bergstrand et al., 2012 Analyses the Social Media Presents an account type typology containing high-level Emergency Response 2013) use of government authorities organizational accounts, accounts for formal functions and roles, formal personal accounts and affiliated personal accounts.

(Fuchs et al., 2013) 2013 European Flood in Germany Confirms the potential of Twitter as a distributed ‘social sensor’ Visual Analytics but at the same time highlights some caveats in interpreting immediate results.

(Reuter et al., 2013 European Flood in Germany Identifies challenges (1) clarity and representation of relevant Moderators and 2015) content, (2) moderation and autonomous work, (3) feedback Design Challenges and updates in interaction relationships and (4) integration of technologies and interaction types.

(Cobb et al., 2014) 2013 Investigates the current Identifies design implications for integrating the activities of Digital Volunteers tools, work practices and ad distributed volunteers. hoc collaboration of distributed digital volunteer teams

(Gorp, 2014) 2014 Investigation of V&TCs Categorizes Volunteer and Technical Communities into Vo l u n t e e r a n d and aid organizations software platform development communities. Technical Communities (V&TCs)

Portugal, Norway, Ireland, Great Britain and Poland). The study focuses on the identification of user needs concerning crisis management with the support of social media and mobile devices. The main goal of the study was to identify possibilities and challenges of social media integration into crisis response management. Generally the participants had a positive attitude towards social media. During the study, both citizens and emergency services identified the same challenges, such as a lack of knowledge, personnel issues, and uniform terms of use, credibility of citizen-generated content, and accessibility for older generations.

1.4. Research Gap

As seen some of this related work is focusing on citizens’ perception. Other work focuses on the perception of emergency services in the US. Just Flizikowski et al. (2014) focuses on European emergency services. However they study the current state and challenges, but did not explicitly focus on future plans and the activities in research. This gap is going to be addressed within this paper by our interview study. Our research question therefore was:

What is the use of social media by emergency services in Europe?

In order to answer this question the sub questions of (a) current use, (b) potential use, (c) participation in research, (d) benefits, (e) challenges and (f) future plans will be studied.

Copyright © 2015, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global is prohibited. 40 International Journal of Information Systems for Crisis Response and Management, 7(2), 36-58, April-June 2015

2. METHODOLOGY OF DATA COLLECTION

To collect information about the different types of authorities’ users and their motivation and attitudes in the use or potential use of social media in emergencies, a number of interviews with authorities’ professionals were conducted.

2.1. Methodology and Details of Question Asked

It was agreed to focus on qualitative rather than quantitative data, and in this respect, the interviews followed a semi-structured form. The openness provided by this kind of form, in comparison to a strictly form, is deemed to provide the opportunity for more direct communication between the interviewer and the interviewee. This approach allows a more in-depth exploration of the issues being covered, which provides a richer set of results due to ideas and topics brought up during the interview. Additionally, this form allows the interviewer to ask questions in ways more appropriate to each interviewee based on his/her background (see also Baxter & Courage, 2005). Before the conduction of the interviews, all participants were briefed about the general project idea and the purpose of the interviews (including the grant of the EmerGent). We also prepare a guidance document to ensure that interviewers explored the same topics in a similar way. The guidance document contains a framework of questions outlined in Annex A, with free answers or in some cases multiple choice answers or rating scale answers. The interview covered different topics, structured in six sections:

A. Introduction to the organization B. Information about the PSAP and/or emergency dispatching center C. Technical information about the PSAP and/or emergency dispatching center D. The use of social media in emergencies E. Challenges and benefits of the use of social media in emergencies F. Future plans

2.2. Collection of Results

Members of the project EMERGENT conducted interviews during May 2014, in face-to-face or telephone sessions. The results were usually collected offline and were later added by each interviewer to an online tool for consistency and further analysis. The results are provided in the following sections anonymously and neither with an indication of the country nor the interviewee. The sections provide a detailed analysis of the data collected during the interviews and describe the current and potential future use of social media in emergency management, by taking into account participant perceptions and experiences so far. Overviews of the collected responses are presented in tables. Conclusions and important highlights of the analysis of results are provided within each section. The first sections are focused on presenting the profile and operational characteristics of the respondents and their organizations (section 3), their media use for operations and communication (section 4) while the remaining section focus on social media use (section 5).

3. Organization Profile 3.1. Participants (Question A4)

The selection criteria focused on collecting responses from a wide range of authorities and emergency services. Interview participants were selected across the networks of consortium partners and from a variety of Public Safety Answering Points (PSAPs) and emergency dispatch centers. Here, the aim was to cover all types of authorities relevant for the project. While the interviews collected responses from authorities across different countries, the selection criteria were not explicitly based on the geographical location. Similarly, use of social media was not a prerequisite for participating in the study, but on the contrary responses from both authorities using and not using social media were welcome. Due to the nature and wide range of the topics covered, for example operational and technical topics, in some cases more than one person from the same organization contributed to the interview. In total, eleven interviews with 19 participants were conducted. The participants come from seven European countries and eight different cities, including five national capital cities, both represented within the consortium of the project and outside it. The exact locations of the interviews are not provided in this article to ensure the anonymity of the respondents.

3.2. Type and Size of Organization (Question A1)

A wide range of authorities, such as police, fire departments, emergency medical and govern- mental authorities participated in the study, in addition to respondents who were responsible for directly receiving and answering emergency calls (PSAPs). The conducted interviews involve organizations that operate on all administrative levels at local, regional and national level, while most of them operate locally. The case of service providers not directly responsible to handle emergencies, such as a police information service, was also studied as such types of providers are directly related to emergencies, collaborate during emergencies with the emergency services and also use social media in their daily activities.

3.3. PSAP Models (Question A2)

A list of different models (Machado, 2014) was used as the common basis for reporting and comparing the conducted interviews. While this list is not exhaustive and does not cover the entire call handling model or all possible models available worldwide, it was found sufficient for this study, because it helped quickly and easily refer to the major characteristics of operation amongst the respondents. The majority of PSAP model of operation was represented in the study. The participants from PSAPs followed mainly four models:

• Model 1: Emergency Response Organizations handling emergency calls • Model 2: Filtering stage 1 PSAP and resource dispatching stage 2 PSAPs • Model 3: Data gathering by stage 1 PSAP, resource dispatching by stage 2 in an integrated control room. In model 3 all involved emergency services use the same infra-structure (hardware / software), in contrast to model 2, where emergency services use the different infra-structure and interoperability can be a major issue. • Model 4: Emergency Response Organizations independent PSAP

Each model is depicted in the following figures (Figure 1).

Copyright © 2015, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global is prohibited. 42 International Journal of Information Systems for Crisis Response and Management, 7(2), 36-58, April-June 2015

Table 2 provides an overview of the participants’ organizational profile in terms of the PSAP model they follow and their operational level.

3.4. Participant Position and Functionalities (Question A3)

Participants in the interviews came from different professional backgrounds, including operations, communications, technical etc. Since in most cases, more than one participant contributed to each of the eleven cases, 19 professionals with senior positions participated in this study, including:

• Senior specialist in crisis communications • Head of the Communication Centre • Head of the Centre • Commander • PSAP leader

Figure 1. PSAP MODELS (Machado, 2014)

Table 2. Organization profile (Source: May 2014 Survey of authorities)

Case Case Case Case Case Case Case 7 Case Case Case Case 1 2 3 4 5 6 8 9 10 11 PSAP XXXX-X--XXX PSAP Model 1 3 3 4 n/a 2 n/a n/a 2 2 2 Operation loc loc reg loc nat loc, loc, nat loc, loc loc loc level reg reg, nat loc = local; reg = regional, nat = national

• Technical team leader • Head of training department

Their responsibilities in their organizations included dispatch planning, field support, logistics, communication management, team coordination, crisis management, technical administration, exercise preparation etc. This data is provided to help understand the perspectives on the use of social media in relation to the operational/organizational responsibilities. It was important all participants, regardless of their position and background, acknowledged the importance of social media in emergencies, considered it an interesting topic that they like to follow and agreed it cannot be overlooked despite the difficulties currently encountered.

3.5. Risks, Vulnerabilities and Operational Size (Questions A5-A8)

The following table (see Table 3) aims to highlight among others the likelihood and intensity of bilateral communication needs, i.e. from citizens to authorities and vice-versa that results out of the characteristics of population, given risks, hazards and vulnerability as well as the current calls/missions.

Table 3. Characteristics of the analyzed interviews (Source: own elaboration)

Case 1 Case 2 Case 3 Case 4 Case 6 Case 7 Case 8 Case 9 Case 10 Case 11 and 5

Region

Population in 500 – ≤ 500 500 - > 1000 ≤ 500 ≤ 500 ≤ 500 > 1000 ≤ 500 500 - thousands 1000 1000 1000

Risks, Hazards and Vulnerability

High population XXX --X XX density

Large industrial -XXXXXX XX sites

Harbors XXXX---X X

Railways XXXXXXXXXX

Tunnels X XX

Major motorways XXXXXXXX

Earthquakes -XXX X

Storms / bad X -XXX X weather conditions

Water catchment -X X - n/a n/a area

Flooding XX X-XX XX

Cultural heritage, X X X X n/a n/a X X X old cities

Touristic attraction XXXX-XXX

Cultural Events X XX

Calls / Missions 17135 19500 1432 3942 1467 59458 0 1700000 240000 140000

Copyright © 2015, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global is prohibited. 44 International Journal of Information Systems for Crisis Response and Management, 7(2), 36-58, April-June 2015

3.6. Operational Tasks (Question B1-B3)

All participants share the task of responding to the public, or have a direct role in responding to emergency incidents. Case 5 is an exception to this as it holds a higher level role in emergency management. With the exception of case 7, most of the participants are involved in notifying the public and providing information. Similarly, case 10 is the only organization in the study, not directly responsible for receiving emergency calls. Table 4 summarizes the results. Table 5 identifies how each participant is involved in the four stages of the emergency management cycle1.

3.7. Inbound Communication (Question B4)

The communication channel most frequently reported and used as a source of input messages from citizens is the telephone. Another frequently reported channel is SMS, which in some cases is also used by people with hearing disabilities. Social media is usually not reported as an inbound message source, with the exception of two cases. In the first case, it is reported that social media is so tightly integrated in our daily life and communication channels, that is too important to ignore nowadays and it would be unthinkable not to respond to it. In the second case – although social media is not officially acknowledged as an inbound message source – messages are received, read and certainly taken into account. In both cases and in the context of this study, social media is used as a communication channel for inbound information in practice, while officially it is not acknowledged or promoted as an information channel. Other communication channels reported: Case 1 and 3: fax (deaf and persons hard of hearing); case 2 is planning to implement SMS and email channels in the near future, as it is important to widen the ways to reach the 112 PSAP; case 8: WAP, Video Call. Table 6 summarizes the results of all cases.

Table 4. Operational Tasks (Source: own elaboration)

Case Case Case Case Case Case Case Case Case Case Case 1 2 3 4 5 6 7 8 9 10 11 Receiving urgent X X X - - X X X n/a - X calls from 112 Receiving direct XXXX-XXXX-X urgent calls Notification X X X X - X - X X X X Providing XXX--X-XXXX information

Table 5. Participant involvement in four stages of an emergency (Source: own elaboration)

Case Case Case Case Case Case Case Case Case Case Case 1 2 3 4 5 6 7 8 9 10 11 Prevention n/a - X n/a n/a X - - X X X Preparedness n/a - X n/a n/a X X - X X X Response n/a X X n/a n/a X - X X - X Recovery n/a - n/a n/a n/a - - - X - X

3.8. Outbound Communication (Question B5)

The purpose of sending messages to citizens follows the same patterns as with receiving messages and finds the telephone as the most commonly used channel. All other channels are similarly used as for inbound messages (see Table 7).

3.9. Broadcasting (Question B6)

Authorities appear less reluctant in using social media for broadcasting messages. Most participants use the radio for broadcasts, although it was commented there has been a shift from the press and the radio to the online channels. The use of their websites, other online media websites and social media provide the opportunity to inform the public very fast and allow the message to spread rapidly. In some cases, social media, such as Twitter, were previously used for communicating with the media, and have proved to present an excellent opportunity for providing public information. The results are summarized in Table 8.

Table 6. Communication Channels for inbound messages (Source: own elaboration)

Case Case Case Case Case Case Case Case Case Case Case 1 2 3 4 5 6 7 8 9 10 11 Telephone calls X X X X n/a X X X X X X SMS - - - - n/a X - X X - - Paging - - - - n/a ------Email - - - - n/a - - - - - X Webform - X - - n/a - - - - X X Radio Traffic X - - - n/a X X X - - - Social Media - - - - n/a - - - - X X

Table 7. Communication Channels for outbound messages (Source: own elaboration)

Case Case 2 Case 3 Case Case 5 Case 6 Case 7 Case 8 Case 9 Case Case 11 1 4 10

Telephone calls X X X X n/a X X X X X X

SMS - X - - n/a X - X X - -

Paging - X - - n/a X - X - - -

Email - - - - n/a X - X - - -

Webfor - - - - n/a - - X - - -

Radio Traffic - X - - n/a X X X - - -

Social Media - - - - n/a X - - - X X

Copyright © 2015, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global is prohibited. 46 International Journal of Information Systems for Crisis Response and Management, 7(2), 36-58, April-June 2015

Table 8. Communication Channels for broadcasting warning messages (Source: own elaboration)

Case Case Case Case Case Case Case Case Case Case Case 1 2 3 4 5 6 7 8 9 10 11 Radio XXXX--X-XXX Local TV - - - X - - X - X X - SMS-Broadcast ------Social Media - - - - - X - - X X X Sirens ------X--XX Special warning 1) ------Offline media & -X-X----XXX press Online media - X - X - - - - X X X 1)Systems like KatWarn or AmberAlerts

4. SOCIAL MEDIA IN EMERGENCIES 4.1. Current Use (Questions D1-D14)

Most of the analyzed case studies use social media (except case 4 and 5) for different purposes that will be described below. Twitter is often used, sometimes Facebook and YouTube. Other services are rather seldom used. The following table (see Table 9) gives an overview about what kind of social media is present at the institutions/authorities that were part of the survey. The different tools were not predefined, but collected during the interviews.

4.2. Potential Use (Question D15-D23)

The analysis of the first stage interviews highlighted that, in general, social media is seen as an important tool for the future, among others for gathering of information, interpretation of incidents and communication to/with the public. It is seen as one of the fastest ways to disseminate

Table 9. Current use of social media (Source: own elaboration)

Case Case Case Case Case Case Case Case Case Case Case 1 2 3 4 5 6 7 8 9 10 11 None - - - X X - X X - - - Twitter X X X - - X - - X X X Facebook X X - - - X - - X - - YouTube - X ------X - - Flickr ------X - - LinkedIn ------X - - Tweedeck - - X* - - - - - X - - Hootsuite - - X* ------*Individual key players

Copyright © 2015, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global is prohibited. International Journal of Information Systems for Crisis Response and Management, 7(2), 36-58, April-June 2015 47 information about prevention or to inform the public with warnings or notifications before, during and after a major incident. Further, social media supports a bi-directional communication and can support all stages of an emergency management cycle: starting with identification of potential hot spots and informing or warning the public about certain facts regarding an incident (prevention or mitigation), informing the public about certain warnings or directions regarding an incident like spread of hazardous substances (preparedness), informing the people about forthcoming evacuation out of exposed areas, and reduce or avoid the possibility of emergency escalations (response), inform the people about medical aid and further behavior (recovery). The intensity and degree of “information exchange” concerning relevant phases of the emergency management cycle making use of social media is diverse. Case 1 for example distinguishes between the foci “operation” and “general”. Concerning the operation, the focus is set up on the receipt of relevant information for operation. This information is used for e.g. necessary intervention concerning the process of communication (e.g. information provided by e.g. media) and thereby control and regulation of information (e.g. instructions concerning behavior, revision and correction of wrong information). In general it concerns public relations, information about training positions and advertisements. Even if the use of social media is seen as important, there are some special restrictions concerning the “exchange” (gathering and mining) of data, not only due to the limited resources, but also limited tools and authorization. Case 10 differentiates the responsibility according to the dimension: in special cases and big emergencies the responsibility lies with the police communications department. In other “simpler” or pure information cases everyone in the dispatching center is responsible for social media use. Due to missing technical possibilities and tools (“Internet lock out”) for example in case 1 only the staff of the press office (fire department) are allowed to gather data (here data filtering/gathering is done only – if at all – when this is connected to a clear operation). A similar situation is seen in case 2, where the communication service local (government) and communication service (fire service) is only allowed to spread data (responsibility belongs to the experts in the service) or case 6 the dispatching center (responsible there: dispatch officer, IT staff) or as in case 11 the officer in command. Nevertheless, some of the analyzed cases already use (more or less) social media tools. Case 2 uses the Twitter platform, Hootsuite, Storify, Social Mention and aims to use Coosto in future. Here, the observation of current communication processes is used. Some other examples are crowd sourced maps including OpenStreetMap, Google Crisis Map or crowd sourced information management, crowd sourced pictures, information about smoke spread is used by e.g. case 2. Case 10 was said to use several tools (Hootsuite, Tweetdesk and Coosto) when preparing for a big event but not in urgent emergencies. During an emergency one or two people are gathering the data. Case 11 did not use any specific tools or applications. In one case, the use of a specialized Twitter service, Twitter Alert was reported. Twitter Alert was launched in late 2013 and aims to enable authorities to keep people well informed by providing accurate and important information during an emergency. It is open to be used by authorities, government services and NGOs worldwide. Beside the limited human and technical resources, the financing of using new social media is quite problematic and constrains its use. In some cases (case 1, 4, 5) there is no financing for social media. Nevertheless, in other cases (2, 3, 6, 10, 11) a budget is available or could be provided (local, regional, own budget) to guarantee this task, even if it is not part of the regular budgets and not marked as a use for social media. The strategy for implementation into the organization differs through the case studies. Some case studies have no clear plans (like case 1); other organizations (case 3) are experimenting with the use of social media in case of an emergency. There have been experiments developed

Copyright © 2015, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global is prohibited. 48 International Journal of Information Systems for Crisis Response and Management, 7(2), 36-58, April-June 2015

throughout the use of Coosto but these experiments were sporadic and without reciprocal rec- onciliation. Others (case 2) have clear expectations and define the first steps how to use and implement social media. Case 2 is running experiments with the active role of citizens during emergencies: they use social media on a daily basis to monitor and to gather information about incidents. Up to now this is performed manually, but in the near future it is aimed to implement software to automate this process and to implement social media in the dispatching rooms of the fire service and police (in the police, social media monitoring through specific software is already used to improve intelligence). In case 6, fire officers and dispatchers are using social media without a clear and determined goal. This way, it is present in their daily routines. However none of the staff have received any training on how to use it so far. They plan to provide warning messages via Twitter and Facebook in incidents, where more than two fire brigade units are present at the spot or during incidents with hazardous materials. Nevertheless and as mentioned before, there might be some problems with integrating it with the existing software. As any new software has to be checked and verified by the PSAP director, this might take a while (in future, there might be a chance to integrate it into the existing software). While there are no formal or documented implementation strategy or guidelines in case 10, search for information and monitoring of social media is done continuously and in parallel with all normal activities. This is similar to the situation in case 11, where the information is used (one person in the PSAP is responsible for posting messages, all people are responsible for reading messages), but no formal workflow, method, or guidelines exist and no special tools are used. Case 2 aims at a distinction between expert social media users and basic users: Basic users are, for example, PSAP operators who just have to get information from social media when it’s relevant. For these users, case 2 is focusing on buying software that will allow it to build a dashboard to show the right information when it is available. Expert users are the ones who control the dashboard and choose the search protocol. They are also the users that provide crisis communication on social media, as there is a requirement to be very familiar with social media “language” to be able to communicate in the most appropriate way. At this moment, implementation is still a problem as a lot of the necessary knowledge is only held by experts in the organization.

4.3. Research (Questions D24-D27, F5)

The importance and position of new social media will increase in future. This is stated in the cases analyzed, although most interviewees are not active in the field of research (e.g. cases 6, 7, 8 and 11). Therefore, a further use is in principle welcome and recommended. Nevertheless due to missing technical and time resources, an implementation in case 1 is currently not possible. Case 2 plans to use social media for e.g. gathering information, communication to the public, cooperation between services (for example Google Hangouts to communicate between the gold and silver command). There are concerns to follow research and development on social media as stated by case 2 in the field of monitoring tools (e.g. implementation of Coosto), continuous evaluation of social media use (what works and what doesn’t), strengthen the cooperation with universities and the “team D5”, a newly formed team of public communication specialists who help each other during crisis (“team D5” have a specific section on social media monitoring). Case 11 states that possibilities of social media are currently being considered, even there are no formal plans for using it. Here, the verification of information is important as well as the possibility of an account being hacked is considered a serious risk, especially if at the same time there is a serious emergency.

4.4. Benefits (Question E1)

Basically all possible benefits (warning, reachability, situation awareness for authorities and citizens) have been mentioned by all interviews. The following table (see Table 10) shows an overview about the benefits of social media use not only for the public but also for authorities. In times of an incident, inhabitants want to be informed by the authorities in the same way as they daily gather news. That is dependent of the involvement and curiosity of the inhabitants. Case 11 states social media has taken away some of the distance between the public and the Police, resulting in the police coming closer to citizens. Case 10 highlighted that people know they get valid information from emergency services and that they can trust this information. Here, trust is seen as a prerequisite for an appropriate relationship between local / regional authorities and the society. Trust has a key role in dealing with given risks and communication and should be regarded as fundamental for risk interpretation and awareness of the public between “real” and “perceived” risks (interpretations of “risk” differ according to individual and social contexts). It has to be kept in mind that public decision-making, which is based only on the factual “scientific” dimension of risk leads to distrust, not taking into account the “socio-cultural” dimension, which includes how a particular risk is viewed when values and emotions are concerned (e.g. whether a risk is judged acceptable, tolerable or intolerable by society is partly influenced by the way it is perceived to intrude upon the value system of society). In addition, it contributes to the vulnerability of institutional settings as well as affected individuals. Case 3 stated that when the government takes a position in the social media networks, they will be able to deliver an interpretative perspective and handling perspective to the inhabitants in the area of the incident. This causes effects on the resilience of a community, because only those who are well informed and integrated in the process will accept the decisions made by different authorities and undertake the right choices/decisions in cases of risks. Furthermore it “proves” the image of the organization. Case 3 highlighted the fire service has a fine (positive) image towards the inhabitants compared to different other services and the government, where their image is significantly perceived worse. The government must be a reliable partner concerning information exchange in the daily life, before they will be considered reliable at times of an incident. By quickly providing the “correct” information (concerns quality and quantity), the inhabitants will be provided only this kind of information they really want and especially need. By this way, the government can prevent panic situations during a big incident or a critical event.

Table 10. Benefits of Social Media use (Source: own elaboration)

Case Case Case Case Case Case Case Case Case Case Case 1 2 3 4 5 6 7 8 9 10 11 Better public warning X X n/a X n/a X X X X X X Better reachability of X X n/a X n/a X X X X X X citizens Improvement of the X X n/a X n/a X X X n/a n/a X situation awareness for authorities Improvement of the X X n/a X n/a X X X n/a n/a X situation awareness for citizens

Copyright © 2015, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global is prohibited. 50 International Journal of Information Systems for Crisis Response and Management, 7(2), 36-58, April-June 2015

4.5. Challenges (Question E2)

In parallel to the benefits, the interview study highlights additional challenges (even if some of the case studies have no experience with it, the difficulties could be estimated, as seen in the table below (Table 11). The biggest challenge is reliability of the sources, followed by the liability, believability and objectivity. Less important challenges are timeliness, reproducibility, understandability and legal uncertainty. Additionally to this, case 3 stressed that some identified difficulties are the speed information is released with, it is never possible to quickly discover in which communities the most critical information is shared; you do not know who needs the information at most and messages cannot be quickly or sufficiently checked concerning their truth. In fighting a crisis, there is a crucial process of validation of the facts to prevent the ex-post evaluation and the wrong conclusions.

4.6. Future Plans (Questions F1-F4)

Our interview study reveals that concerning future plans for using social media in emergencies the impact is estimated increasing (cases 1, 2, 4, 5, 6, 7, 8). Furthermore, case 2 stated that in 5-10 years crowdsourcing and big data will be standard technologies in crisis management. Therefore, it is necessary to focus on such items. In a crowd sourced and big data world (cf. Ludwig et al., 2015), verification becomes the main skill of the PSAP operator and the information manager. Nevertheless case 1 voiced the concern that as long as the technical and personnel circumstances will stay as they are, there will be no significant change. In cases 4 and 5 there are no examples of using social media in emergences by public administration. Probably some will be observed as important issue only when this gap will be filled by citizens. For sure social media will be used mostly in the field of situation monitoring and public warning. Case 6 dispatching center will promote the implementation of notifications such as warnings or alerts. The attention should be moved also to gathering information from the public. Case 8 has no clear concept, but will place the social media growth and usage on the own agenda shortly. Case 11 will consider the use of Instagram and Facebook, but there are no precise plans at the moment (“we know what social media is only for a few years”).

Table 11. Difficulties of using social media (Source: own elaboration)

Case 1 Case 2 Case Case 4 Case Case 6 Case 7 Case 8 Case Case Case 11 average 3 5 9 10 Reliability of the high medium n/a high n/a high high high n/a high high High different sources (2,9) Liability of the medium medium n/a high n/a high high high n/a high high High different sources (2,8) Believability / medium medium n/a high n/a high high high n/a high high High Verifiability (2,8) Objectivity high medium n/a medium n/a medium high high n/a high high High (2,6) Timeliness low low n/a no n/a medium high high n/a high medium Medium (2,0) Reproducibility low low n/a medium n/a medium high high n/a high n/a Medium (2,1) Understand- low low n/a medium n/a low high high n/a high n/a Medium ability (2,0) Legal uncertainty low low n/a high n/a high high high n/a low n/a Medium (2,1) High=3; medium=2; low=1; no=0

Concerning new technologies, case 2 is focused on Coosto and case 6 aims at providing information to the general public using one app. This app should post a warning message to everybody on that area, regardless of the follower status. Case 6 and case 1 highlighted social media will be present in the daily routines in dispatch centers, PSAPS, etc. and it is possible that the reporting of the fire departments will focus more toward social media and the “older” solutions (such as a website) will lose importance. The units (as in case 3) are aware about the development around social media and know that they will have to work with it. Nevertheless, for government agencies this is a new phenomenon. At this moment, there are only experiments, for example Coosto, put together to look how and what you are able to do with social media during a crisis. However, all interviewed institutions / units agree in opinion that research projects like EMERGENT are helpful to gain new insights.

5. CONCLUSION

Social media is and has already been widely used during a variety of emergencies. After presenting a definition for social media, various related studies are presented. Our performed study includes a variety of respondents who helped collecting the reported results. Participants were from different European countries and their profile varied in terms of their role in emergency management, operational size, methods and tasks. The main results are given in Table 12. Social media provides a direct and fast channel for communicating with the public and spreading accurate and trusted information, especially during emergencies, when it is even more important for people to have access to trusted information. Social media provides the possibility for bi-directional communication and it is currently used in the majority of the cases. However, the current practice shows higher use of broadcasting information to citizens, rather than responding to individual cases. The use of social media during the different stages of an emergency management cycle highlighted the importance of information gathering and validation. This is considered as one of the great challenges in the use of social media and in some cases it may prohibit or restrict further use. To overcome this barrier, an alternative model of operation is explored by some authorities, where the purpose of using social media is distinguished between use for operational purpose versus informational purpose, or use during an emergency versus use during normal operation. The emerging difficulty in handling the data streams from social media, trigger the opportunity for supporting these processes with applicable tools. Authorities already experiment with the

Table 12. What is the use of social media by emergency services in Europe?

Question Answer Current Use Twitter often, sometimes Facebook and YouTube. Other services rather seldom. Potential Use Gathering of information, interpretation of incidents and communication to/with the public Research Most interviewees are not active in the field of research Benefits Basically all possible benefits (warning, reachability, situation awareness for authorities and citizens) have been mentioned by all interviews. Challenges Reliability of the sources, followed by the liability, believability and objectivity. Less important challenges are timeliness, reproducibility, understandability and legal uncertainty Future Plans Impact is estimated increasing

Copyright © 2015, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global is prohibited. 52 International Journal of Information Systems for Crisis Response and Management, 7(2), 36-58, April-June 2015

use of different tools, although the current study reports a lack of a clear understanding of what tool to use or which tool is better in different situations. While implementation strategies or models for integration with current workflows are clearly interesting topics and some authorities currently explore or discuss different options, no clear or documented plans or goals were identified. Interviewees were not active in the research field of social media and emergency management. However, all shared an interest to follow the advances of this theme and expect the importance of social media in emergencies to rise in the future. To achieve a broader view across Europe, larger studies are necessary. The project EmerGent conducted an additional European survey in 2014 and 2015 (a) among 700 people (working within emergency services) across 27 countries and (b) among 1000 citizens to investigate their experiences and visions on how to use social media nowadays and in the future. These qualitative and quantitative results will give the project directions on how to use social media more effectively in the future and how to support both emergency services and citizens in a better way during emergencies.

ACKNOWLEDGMENT

The research project EmerGent’ was funded by a grant of the European Union (FP7 No. 608352). We would like to thank all members of our project for their remarks and for supporting in the work. This work has been conducted as part of EmerGent’s work package 3: “Analysis of Social Media in Emergencies Today and Tomorrow”.

REFERENCES

American Red Cross. (2012a). More Americans Using Mobile Apps in Emergencies. Retrieved from http:// www.redcross.org/news/press-release/More-Americans-Using-Mobile-Apps-in-Emergencies American Red Cross. (2012b). Social Media in Disasters and Emergencies. Retrieved from http:// a1881.g.akamai.net/7/1881/26640/v0001/redcross.download.akamai.com/26640/PollData/Social Baxter, K., & Courage, C. (2005). Understanding Your Users: A Practical Guide to User Requirements Methods, Tools, and Techniques. Morgan Kaufmann. Bergstrand, F., Landgren, J., & Green, V. (2013). Authorities don’t tweet, employees do! Proceedings of the MobileHCI 2013. Birkbak, A. (2012). Crystallizations in the Blizzard : Contrasting Informal Emergency Collaboration In Facebook Groups. Proceedings of the Nordic Conference on Human-Computer Interaction (NordiCHI) (pp. 428–437). doi:10.1145/2399016.2399082 Canadian Red Cross. (2012). Social Media during Emergencies. Retrieved from http://www.redcross.ca/ cmslib/general/pub_social_media_in_emergencies_survey_oct2012_en.pdf Cobb, C., McCarthy, T., Perkins, A., & Bharadwaj, A. (2014). Designing for the Deluge: Understanding & Supporting the Distributed, Collaborative Work of Crisis Volunteers. Proceedings of the Conference on Computer Supported Cooperative Work (CSCW), Baltimore. doi:10.1145/2531602.2531712 Endsley, T., Wu, Y., & Eep, J. (2014). The Source of the Story: Evaluating the Credibility of Crisis Informa- tion Sources. Proceedings of the Information Systems for Crisis Response and Management, 1, 158–162. Flizikowski, A., Hołubowicz, W., Stachowicz, A., Hokkanen, L., & Delavallade, T. (2014). Social Media in Crisis Management – the iSAR + Project Survey. Proceedings of the Information Systems for Crisis Response and Management (ISCRAM) (pp. 707–711). Fuchs, G., Andrienko, N., Andrienko, G., Bothe, S., & Stange, H. (2013). Tracing the German Centen- nial Flood in the Stream of Tweets: First Lessons Learned. Proceedings of the SIGSPATIAL Interna- tional Workshop on Crowdsourced and Volunteered Geographic Information, Orlando, USA (pp. 2–10). doi:10.1145/2534732.2534741 Heverin, T., & Zach, L. (2010). Microblogging for Crisis Communication: Examination of Twitter Use in Response to a 2009 Violent Crisis in the Seattle-Tacoma, Washington Area. Proceedings of the Information Systems for Crisis Response and Management (ISCRAM), Seattle, USA. Hughes, A. L., & Palen, L. (2009). Twitter Adoption and Use in Mass Convergence and Emergency Events. In J. Landgren, & S. Jul (Eds.), Proceedings of the Information Systems for Crisis Response and Manage- ment (ISCRAM), Gothenburg, Sweden. doi:10.1504/IJEM.2009.031564 Hughes, A. L., St. Denis, L. A. A., Palen, L., & Anderson, K. M. (2014). Online Public Communications by Police & Fire Services during the 2012 Hurricane Sandy. Proceedings of the Conference on Human Factors in Computing Systems (CHI). doi:10.1145/2556288.2557227 Jennex, M. E. (2012). Social Media – Truly Viable For Crisis Response? In L. Rothkrantz, J. Ristvej, & Z. Franco (Eds.), Proceedings of the Information Systems for Crisis Response and Management (ISCRAM), Vancouver, Canada (pp. 1–5). Kaplan, A. M., & Haenlein, M. (2010). Users of the world, unite! The challenges and opportunities of Social Media. Business Horizons, 53(1), 59–68. doi:10.1016/j.bushor.2009.09.003 Latonero, M., & Shklovski, I. (2011). Emergency Management, Twitter, and Social Media Evangelism. International Journal of Information Systems for Crisis Response and Management, 3(4), 1–16. doi:10.4018/ jiscrm.2011100101

Copyright © 2015, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global is prohibited. 54 International Journal of Information Systems for Crisis Response and Management, 7(2), 36-58, April-June 2015

Liu, S., Palen, L., & Sutton, J. (2008). In search of the bigger picture: The emergent role of on-line photo sharing in times of disaster. In Proceedings of the Information Systems for Crisis Response and Manage- ment (ISCRAM), Washington D.C., USA. Ludwig, T., Reuter, C., Siebigteroth, T., & Pipek, V. (2015). CrowdMonitor: Mobile Crowd Sensing for Assessing Physical and Digital Activities of Citizens during Emergencies. Proceedings of the Conference on Human Factors in Computing Systems (CHI), Seoul, Korea. ACM Press. doi:10.1145/2702123.2702265 Machado, G. (2014). PSAPs Organisation in Europe. Retrieved from http://www.eena.org/ressource/static/ files/2013_04_16_psaporgeurope.pdf Murphy, T., & Jennex, M. E. (2006). Knowledge Management, Emergency Response, and Hurricane Ka- trina. International Journal of Intelligent Control and Systems, 11(4), 199–208. Nagy, A., Valley, C., & Stamberger, J. (2012). Crowd Sentiment Detection during Disasters and Crises. In Proceedings of the Information Systems for Crisis Response and Management (ISCRAM), Vancouver, Canada (pp. 1–9). Perng, S.-Y., Büscher, M., Wood, L., Halvorsrud, R., Stiso, M., Ramirez, L., & Al-Akkad, A. (2012). Pe- ripheral response: Microblogging during the 22/7/2011 Norway attacks. Proceedings of the Information Systems for Crisis Response and Management (ISCRAM) (pp. 1–11). Plotnick, L., Hiltz, S. R., & Kushma, J. a, & Tapia, A. (2015). Red Tape : Attitudes and Issues Related to Use of Social Media by U. S. County- Level Emergency Managers. Proceedings of the Information Systems for Crisis Response and Management (ISCRAM), Kristiansand, Norway. Qu, Y., Wu, P. F., & Wang, X. (2009). Online Community Response to Major Disaster : A Study of Tianya Forum in the 2008 Sichuan Earthquake. Proceedings of the Hawaii International Conference on System Sciences (HICSS), Waikoloa, USA. Reuter, C., Heger, O., & Pipek, V. (2013). Combining Real and Virtual Volunteers through Social Media. Proceedings of the Information Systems for Crisis Response and Management (ISCRAM) (pp. 1–10). Reuter, C., Ludwig, T., Kaufhold, M.-A., & Pipek, V. (2015). XHELP: Design of a Cross-Platform Social- Media Application to Support Volunteer Moderators in Disasters. Proceedings of the Conference on Human Factors in Computing Systems (CHI), Seoul, Korea. ACM Press. doi:10.1145/2702123.2702171 Reuter, C., Marx, A., & Pipek, V. (2012). Crisis Management 2.0: Towards a Systematization of Social Software Use in Crisis Situations. International Journal of Information Systems for Crisis Response and Management, 4(1), 1–16. doi:10.4018/jiscrm.2012010101 San, Y. S., Wardell III, C., & Thorkildsen, Z. (2013). Social Media in the Emergency Management Field: 2012 Survey Results, June. Shklovski, I., Palen, L., & Sutton, J. (2008). Finding Community Through Information and Communication Technology During Disaster Events. Proceedings of the Conference on Computer Supported Cooperative Work (CSCW), San Diego, USA, ACM-Press. St. Denis, L. A., Hughes, A. L., & Palen, L. (2012). Trial by Fire: The Deployment of Trusted Digital Volunteers in the 2011 Shadow Lake Fire. Proceedings of the Information Systems for Crisis Response and Management (ISCRAM) (pp. 1–10). Starbird, K. (2013). Delivering Patients to Sacré Coeur : Collective Intelligence in Digital Volunteer Com- munities. Proceedings of the Conference on Human Factors in Computing Systems (CHI) (pp. 801–810). doi:10.1145/2470654.2470769 Starbird, K., & Palen, L. (2010). Pass It On?: Retweeting in Mass Emergency. Proceedings of the Informa- tion Systems for Crisis Response and Management (ISCRAM), Seattle, USA (pp. 1–10). Starbird, K., & Palen, L. (2011). Voluntweeters: Self-Organizing by Digital Volunteers in Times of Crisis. Proceedings of the Conference on Human Factors in Computing Systems (CHI), Vancouver (pp. 1071–1080). ACM-Press. doi:10.1145/1978942.1979102

Starbird, K., & Palen, L. (2012). (How) Will the Revolution be Retweeted? Information Diffusion and the 2011 Egyptian Uprising. Proceedings of the Conference on Computer Supported Cooperative Work (CSCW) (pp. 7–16). Sutton, J. (2010). Twittering Tennessee: Distributed networks and Collaboration Following a Technological Disaster. In S. French, B. Tomaszewski, & C. Zobel (Eds.), Proceedings of the Information Systems for Crisis Response and Management (ISCRAM), Seattle, USA. Van Gorp, A. F. (2014). Integration of Volunteer and Technical Communities into the Humanitarian Aid Sector: Barriers to Collaboration. Proceedings of the Information Systems for Crisis Response and Man- agement (ISCRAM) (pp. 620–629). Vieweg, S., Hughes, A. L., Starbird, K., & Palen, L. (2010). Microblogging During Two Natural Hazards Events : What Twitter May Contribute to Situational Awareness. Proceedings of the Conference on Human Factors in Computing Systems (CHI) (pp. 1079–1088). doi:10.1145/1753326.1753486 Wilensky, H. (2014). Twitter as a Navigator for Stranded Commuters during the Great East Japan Earthquake. Proceedings of the Information Systems for Crisis Response and Management (ISCRAM) (pp. 695–704). Wulf, V., Misaki, K., Atam, M., Randall, D., & Rohde, M. (2013). “On the Ground” in Sidi Bouzid: Inves- tigating Social Media Use during the Tunisian Revolution. Proceedings of the Conference on Computer Supported Cooperative Work (CSCW) (pp. 1409–1418). New York: ACM. Yang, S., Chung, H., Lin, X., Lee, S., & Chen, L. (2013). PhaseVis: What, When, Where, and Who in Visualizing the Four Phases of Emergency Management Through the Lens of Social Media. In Proceed- ings of the Information Systems for Crisis Response and Management (ISCRAM), Baden-Baden, Germany (pp. 912–917).

ENDNOTES

1 See for example: https://www.providenceri.com/PEMA/about/emergency-management; often is “mitigation” used instead of prevention (we define mitigation as part of the prevention stage).

Copyright © 2015, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global is prohibited. 56 International Journal of Information Systems for Crisis Response and Management, 7(2), 36-58, April-June 2015

Christian Reuter studied Information Systems at the University of Siegen, Germany and the École Supéri- eure de Commerce de Dijon, France (Dipl.-Wirt.Inf.; M.Sc) and received a PhD for his work on (inter-) organizational collaboration technology design for crisis maagement (Dr. rer. pol.) with summa cum laude. He has worked as a web developer, consultant and researcher and has published more than 60 scientific articles. He is voluntary founding spokesman of the section “human computer interaction in security relevant systems” of the German Informatics Society. Thomas Ludwig studied Information Systems at the University of Siegen (Germany) and University of Newcastle (Australia). Currently, he is a research associate and PhD student at the Institute for Informa- tion Systems at the University of Siegen. He finished his study on information systems in 2012. During his study, he worked as a student assistant in the research project ‘INFOSTROM’ dealing with IT-support for inter-organizational crisis management. Now he works in the research project ‘EMERGENT’, which deals with the impact of citizen-generated content of social media in crisis management and he leads the research project ‘KOKOS’ that deals with the integration of citizens into the official crisis response work. He further examines HCI issues of complex as well as critical software and hardware systems and researches how the users’ appropriation of those systems can be supported by technology. Therese Friberg, Dipl.-Inf., research assistant at the University of Paderborn, has been engaged in the field of information quality since 2008. Her diploma thesis was distinguished by the German Association of Information Quality (DGIQ) wherein she is part of an expert group focusing on the standardization of definitions and measurements of information quality. She has published several national and international publications in the field of information quality and human influences; her paper was awarded as best paper at the International Conference of Information Quality (ICIQ’ 09). Ms. Friberg was involved in six successfully finished national research projects and the EU-Projects PRONTO and Secur-ED, all in the field of public safety and security. Currently she is the project manager of EmerGent, a FP7 security project focusing on social media in emergencies. Furthermore she has built up a broad network in the domain of Public Safety and Security (f.i. GESA, EENA, ISCRAM) and also organises workshops on an international level (ISCRAM 2011-2015). Sylvia Pratzler-Wanczura is the principal engineer of the Institute of Fire Service and Rescue Technol- ogy of the City of Dortmund - Fire Department. After her academic studies of spatial planning at the TU Dortmund University (Germany) she worked several years as research assistant and received her Dr.-Ing. In her PhD-Thesis she dealt with methods for measurement of an effective and efficient coordination in the risk governance process (communication, management, assessment and analysis of risk) for flood risk. She has over 10 years of experience in the organisation and management of several European as well as national safety & security related research projects. Alexis Gizikis holds a Master of Engineering (MEng) from the University of Aberdeen in Scotland and has worked for companies and research institutions in the UK and Greece. Alexis has been involved in the ICT sector since 1999 and has acquired experience in business analysis, requirements engineering, HCI and software engineering. He currently works for the European Emergency Number Association (EENA) and he is responsible for planning and monitoring EENA’s participation in research projects.

APPENDIX A: INTERVIEW QUESTIONS

Section A: Introduction to the Organization

• A1. Specify the type of your organization. • A2. Specify your PSAP model. • A3. General Information about you • A4. Geographical and regional information • A5. What are the main risks in your region? • A6. Specify the vulnerability of your region: • A7. Estimate the average number of calls / emergency calls per anum for the following categories. • A8. Estimate the average number of emergency operations per anum for the following categories. • A9. Specify the number of your workforce (ft/pt) • A10. On which administrative division is your organization operating in your state?

Section B: Information about the PSAP and/or Dispatching Center You Are Part Of, or Responsible For

• B1. Select the main tasks within your PSAP (Multiple choices are welcome) • B2. Are there any further tasks your PSAP is responsible for? • B3. Specify in which phases of the emergency management cycle your organization operates and describe the main tasks. • B4. Select the communication channels that your PSAP supports to handle input messages (i.e. messages from citizens). • B5. Select the communication channels that your PSAP supports to send out messages (i.e. messages to inform or support citizens) • B6. Select which communication channels your PSAP uses to broadcast warning messages. A warning message should inhabitants make aware of a danger. • B7. Please describe the special warning systems that your PSAP uses to send out warning messages. • B8. Select the responsible roles for output messages.

Section C: Technical Information about the PSAP and/or Dispatching Center You Are Part Of, or Responsible For

• Not analyzed for this article.

Section D: The Use of Social Media in Emergencies

• D1. Does your organization already use Social Media, e.g. to gather or spread information? • D2. Select the Social Networks that are used by your organization. • D3. Who is allowed to post/send out data? • D4. Is information from Social Media provided by citizens used in authorities? • D5. Which tools do you use to analyze/collect data from Social Media (e.g. Twitcident)? • D6. Who is allowed to use gathered data from Social Media in your organization? • D7. Does your organization use other web 2.0 technologies?

Copyright © 2015, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global is prohibited. 58 International Journal of Information Systems for Crisis Response and Management, 7(2), 36-58, April-June 2015

• D8. In which phases of an emergency do you use Social Media? Please select and describe the usage. • D9. Please try to categorize the main purposes in using Social Media. • D10. What is the main purpose in using Social Media? Please describe in detail. • D11. How does your organization finance Social Media usage? • D12. Please try to categorize the implementation strategy that would fit at most to your current implementation. • D13. Describe your organizations’ implementation strategy on Social Media in detail. • D14. Please describe how you or your organization has implemented Social Media into workflows or processes within your organization. Try to describe the main workflows and how the implementation is done. • D15. Is Social Media an interesting topic for your organization? • D16. Select the Social Networks which are most interesting for your organization in order to inform, encourage or engage citizens. • D17. Please try to categorize the main purposes that you would like to focus in Social Media. • D18. Describe the main purposes in the use of Social Media. • D19. How would you finance the usage of Social Media? • D20. Please try to categorize the implementation strategy that would fit at most. • D21. Please try to describe a suitable implementation strategy on Social Media for your organization. • D22. Please describe how you or your organization would implement Social Media into workflows or processes within your organization. Try to identify the main workflows and describe how the implementation is done. • D23. Please describe why Social Media isn’t an interesting topic for you. • D24. Does your organization make research and development on Social Media? • D25. Please describe the research and/or development that your organization does on Social Media. • D26. Is your organization following or would like to follow research and development on Social Media? • D27. Please describe the research topics you’re interested in.

Section E: Challenges and Benefits on Social Media in Emergencies

• E1. What are the main benefits of Social Media use? • E2. What are the main difficulties or challenges experienced with working with Social Media? • E3. If you have some interesting references to cases studies, SWOT- or risk-analysis regarding Social Media, especially Social Media in emergencies, please paste them below.

Section F: Future Plans

• F1. How do you think the impact of Social Media in emergencies will change in the next years? • F2. How will your organization deal with Social Media in the next years? • F3. How will workflows and tools evolve? • F4. What are the new technologies you are looking at? • F5. Do you think research projects like EMERGENT are helpful to gain new insights?

Contents lists available at ScienceDirect

Int. J. Human-Computer Studies

journal homepage: www.elsevier.com/locate/ijhcs

Emergency services' attitudes towards social media: A quantitative and qualitative survey across Europe

Christian Reuter a,n, Thomas Ludwig a, Marc-André Kaufhold a, Thomas Spielhofer b a University of Siegen, Institute for Information Systems, Germany b The Tavistock Institute of Human Relations, United Kingdom article info abstract

Article history: Various studies show that social media is used in emergencies – and that in spite of possible challenges Received 1 July 2015 for emergency services, beneficial use cases can be identified. However, relatively little empirical data is Received in revised form available regarding the attitudes of emergency services towards social media, and almost none of a 13 January 2016 comparative nature. This article summarizes the findings of a survey conducted of the EU project Accepted 8 March 2016 ‘EmerGent’ with 761 emergency service staff across 32 European countries from September to December Communicated by E. Motta Available online 15 March 2016 2014. The main aims of the survey were to explore the attitudes expressed by emergency service staff towards social media for private and organizational use as well as the levels and main factors influencing Keywords: the current and likely future use of social media in their organizations. Based on our results, we discuss Attitudes possible enhancements of the emergency management cycle using social media. Emergency services & 2016 Elsevier Ltd. All rights reserved. Europe Social media Survey Quantitative Qualitative

1. Introduction monitoring of, and intervention in, rapidly changing events takes place are not always fully recognized. The terms “disaster, crisis, Social media can be defined as a “group of Internet-based catastrophe, and emergency management are sometimes used applications that build on the ideological and technological foun- synonymously and sometimes with slight differences, by scholars dations of Web 2.0, and that allow the creation and exchange of and practitioners”,asHiltz et al. (2011) suggest. In recent years the user-generated content” (Kaplan and Haenlein, 2010). User- use of social media has increased and at the same time the nature “ generated content refers to the sum of all ways in which people of that use has shifted towards a more collaborative model. Recent ” “ make use of social media and describes the various forms of examples of such collaborative coping can be found, among others, media content that are publicly available and created by end- in the 2013 European floods (Kaufhold and Reuter, 2014), the 2013 users” (Kaplan and Haenlein, 2010). The most common players are Colorado floods (St. Denis et al., 2014) or the 2012 hurricane Sandy Facebook with about 1.36 billion active users monthly or the (Hughes et al., 2014). microblogging platform, Twitter, counting approximately 284 Across various studies of emergencies, or disaster events, million active users monthly. YouTube (1 billion),1 WhatsApp (600 numerous positive and negative aspects of social media have been million), Googleþ (343 million), Instagram (300 million) and identified. However, existing studies either focus only on citizens, Tumblr (230 million) (Statista, 2015) also all have more than 100 only the emergency services in the US, or only on one particular million active users monthly and are widely distributed. emergency service. An open question still remains concerning the For almost 15 years social media has been used in crisis management: After the terrorist attacks of 9/11 2001, for instance, perception of different emergency services in Europe towards the – wikis, created by ordinary citizens, were already being used to use of social media. We assumed that different cultural even – collect information about missing people (Palen and Liu, 2007). At within Europe as well as legal backgrounds will have a sig- fi fl ’ the same time, the various different conditions under which the ni cant in uence on the emergency services opinions and attitudes, and the survey we describe below tests that hypothesis. Additionally, different countries in Europe have been affected by n Corresponding author. different types of crisis ranging from natural disasters, e.g. the E-mail address: [email protected] (C. Reuter). fl 1 http://www.businessinsider.com/the-worlds-largest-social-networks-2013- 2013 European oods in the central and eastern countries to ter- 12#ixzz30MjZz5TK. rorist attacks, e.g. the 2005 suicide bomb attacks in England or the http://dx.doi.org/10.1016/j.ijhcs.2016.03.005 1071-5819/& 2016 Elsevier Ltd. All rights reserved. C. Reuter et al. / Int. J. Human-Computer Studies 95 (2016) 96–111 97

2004 train bombings in Spain, which may lead to different types of organizational utilization of social media (Latonero and Shklovski, experiences and attitudes. 2011). Authorities already experiment with the use of different Within this article we sought to explore the attitudes of Eur- tools, although the current study reports a lack of a clear under- opean emergency service staff towards social media for private standing of what tool to use or which tool is better in different and organizational use as well as the levels and main factors situations, as detected in an interview study in 7 European coun- influencing their current and likely future use in their organiza- tries (Reuter et al., 2015a). Based on interviews with representations. With regard to the examination of related work (Section 2), tives from large international disaster response organizations, we describe the methodology of our survey (Section 3) and pre- Tapia and Moore (2014) point out that “emergency responders sent its quantitative as well as qualitative results (Section 4). already operate with less than reliable, or ‘good enough’, infor- Afterwards the findings are discussed with regard to an extension mation in offline practice, and that social media data are useful to of the emergency management cycle, leading to a final conclusion responders, but only in specific crisis situations”. A study of the (Section 5). response after the 2012 hurricane Sandy shows that communication differed between different types of organizations and across various types of media (Hughes et al., 2014). Accordingly, Twitter 2. Related work tends to be used for real-time notification, and Facebook for community engagement (St. Denis et al., 2014). However, some This section summarizes literature findings on the use of social scenarios require cross-platform collaboration of relief activities media in emergencies. We will first present some fundamentals (Reuter et al., 2015b). A comparative study of the police in the 2011 (Section 2.1). We will then highlight emergency services’ work London riots observed communication approaches through Twit- practices (Section 2.2) and later their perception of social media ter (Denef et al., 2013) and classified them into an ‘instrumental’, (Section 2.3). Finally, we present the results of surveys which have formal, depersonalized style of communication that emphasized already been conducted (Section 2.4) to clarify the research gap the gap between the police and the public as well as an ‘expres- this study aimed to fill (Section 2.5). sive’, highly personalized, informal style that allowed direct interaction with individual followers, required high maintenance 2.1. Social media in emergencies and had issues on easily overstepping boundaries like auditing the legality of information publication and learning when to engage Almost a decade ago, Palen and Liu (2007) described the and how to resource it. Reuter and Schröter (2015) revealed emergence of social media in emergencies in these terms: “the pragmatic (linguistic) barriers resulting from irony, wordplay, and role held by members of the public in disasters […] is becoming ambiguity; Ludwig et al. (2015a) reveal challenges regarding more visible, active, and in possession of greater reach than ever information quality of citizen generated content. seen before”. In the same year they asserted that “the availability In summary it can be stated that the “landscape of the use of of mobile, networked information communication technology social media data in crisis response is varied, with pockets of use (ICT) in the hands of ordinary people makes information exchange and acceptance among organizations” (Tapia and Moore, 2014). increasingly potent” (Palen et al., 2007). Nowadays, the broad acceptance of social media has widened the possibility for crisis 2.3. Emergency services’ perception of social media communication in a variety of different scenarios (Reuter et al., 2012), such as from authorities to citizens (A2C), from citizen to Some studies claim that “additional information provided by authorities (C2A) (Kaufhold and Reuter, 2016)aswellasthe volunteers can improve the work of emergency services” (Reuter coordination of voluntary activities among citizens (C2C) and by et al., 2013) and that “volunteered individual reports, especially emergency services (Ludwig et al., 2015b). Reuter et al. (2013) pictures, are of particular value” (Ludwig et al., 2015b). These suggest distinguishing between the real activities of “emergent claims are based on qualitative data concerning the perception of groups” (Stallings and Quarantelli, 1985), which usually act in the volunteer activities by emergency services. A recent contribution form of neighborly help and on-site work, and virtual activities of on the 2013 European flood moreover confirms the potential of “digital volunteers” (Starbird and Palen, 2011), which originate on Twitter as a distributed ‘social sensor’, but at the same time the Internet and work online. These groups naturally partially highlights some caveats in interpreting immediate results (Fuchs overlap, and both intersect with emergency services at various et al., 2013). points. The term “Crisis informatics” covers all of this. It “views Volunteer activism – also in social media – may result in emergency response as an expanded social system where infor- negative impacts; for instance, by increasing the emergency ser- mation is disseminated within and between official and public vices’ pressure to act (Perng et al., 2012). Therefore, in several past channels and entities. Crisis informatics wrestles with methodo- cases, such as the 2011 Shadow Lake Fire, volunteers are actively logical concerns as it strives to develop new theory and support deployed as “trusted volunteers”–a virtual team designed to sociologically informed development of both ICT and policy” manage and monitor social media communications in support of (Palen et al., 2009). emergency incident response. A study of the 2011 Thailand flooding disaster highlights the authorities’ actions taken to cor- 2.2. Emergency services' use of social media rect the mistakes caused by the “emerging risks of the chaotic use of social media” (Kaewikitipong et al., 2012). Alongside the use of social media for communication among Group interviews with 24 county officials about government citizens (C2C), the role of authorities as producer (A2C) or con- officials' social media use in Virginia in 2010, led to the identification sumer (C2A) of information has also been studied. Some years ago, of challenges, such as the overwhelming amount of data and the many emergency managers and agencies adopted social media in recognition of relevant and timely information (Kavanaugh et al., some way, and people increasingly expect agencies to use social 2011). Hughes and Palen (2012) described the burden on emergency media to meet their informational needs (Lindsay, 2011). The 2009 responders to receive and filter a substantial amount of incoming study of public information officers of the Los Angeles fire information. Limited resources that hamper the collaboration of department highlights the importance of the information evan- humanitarian aid organizations and Volunteer and Technical Com- gelist: a person who promotes the use of new forms of media and munities (V&TCs) – technically trained volunteers – using social technology within authorities to achieve an effective media (Gorp, 2014) have also been commented on. 98 C. Reuter et al. / Int. J. Human-Computer Studies 95 (2016) 96–111

According to a study comprising eleven semi-structured although the “largely untapped resource” of digital volunteers interviews with US public sector emergency managers, the could “help to alleviate some personnel issues”. The study also major barriers are mainly organizational rather than technical revealed that 85% of US authorities already use social media. (Hiltz et al., 2014). Most reported statements concern the lack of A further survey of 241 US emergency managers at the county personnel or time to work with social media, the lack of policies level in 2014 shows that only about half of these agencies use and guidelines as well as official prohibitions for its use. They social media (Plotnick et al., 2015). Most of them also do not have therefore reference a structured approach for designing social any formal policies to guide their use. Of those that do have media policies (Hrdinová et al., 2010). Further responses identify formal policies, about one quarter actually forbid the use of social a lack of appropriate technology and training and issues of media. A lack of staff, guidance and skills have been identified as trustworthiness. Hiltz et al. (2014) suggest that interviewees the main barriers for A2C; the main barriers for C2A are lack of were “enthusiastic” about the potential usefulness of affordances staff, trustworthiness and information overload. The authors from current research systems, which have already been sum- conclude that “the agencies and their representatives are not yet marized in a small number of research and commercial approa- ready to embrace SM and use it to its fullest potential. For the ches under different scopes (Pohl, 2013; Reuter et al., 2015b). most part, current SM use is for dissemination of information, not In summary, the overall picture suggests a degree of ambivalence the collection of it”. Furthermore “in addition to technological concerning social media use, with some potential identified (addi- advances, policy and management changes are needed as well, to tional information, pictures, social sensor), but also some risks and remove the “red tape” (lack of guidelines or even prohibitions barriers (pressure to act, chaotic use, overwhelming amount, limited against use) that impedes the effective use” of social media resources, appropriate technology, trustworthiness). (Plotnick et al., 2015). The International Association of Chiefs of Police (IACP) con- 2.4. Previous research on social media use among citizens and ducts an annual quantitative survey about law enforcement's use emergency services of social media to report on “the current state of practice and the issues agencies are facing in regard to social media” (International Several surveys on the possible use of social media already Association of Chiefs of Police, 2015). With over 500 participating exist: With over 1000 participants, a comparative study of the law enforcement agencies across the United States each year, the Canadian Red Cross (2012) aimed to explore the extent Canadian survey provides comparable results on how attitudes and adoption citizens use social media and mobile devices in crisis commu- rates gradually shifted in the recent six years. Comparing the first nication and what they expect from the emergency services both (International Association of Chiefs of Police, 2010) and last sur- currently and in the future. This study emphasizes the require- veys, the agencies’ use of social media increased from 81% (77% ment of trained social media personnel and the credibility issues Facebook, 37% Twitter, 16% YouTube) to 96% (94% Facebook, 71% of citizen-generated content but also shows the benefits of reas- Twitter, 40% YouTube) and the number of social media policies surance for citizens, providing situational information and mon- increased from 35% to 78%. The 2015 survey also highlights that itoring. Social media were seen as a support for existing channels, 74% of responding agencies that are not currently using social but not as a replacement for them. It is noteworthy that the media are considering its adoption, 86% report that social media Canadian Red Cross employs “trusted volunteers” to support offi- helped to solve crimes and 84% state that social media has cial response via social media. improved police-community relations in their jurisdiction. Fur- The American Red Cross (2012) also studied citizens’ use of thermore, some aspects agencies are “very concerned” regarding social media during emergencies, with 1.017 online and 1.018 social media use are: Online radicalization and violent extremism telephone survey respondents. According to the study, 12% of the (26%), criminal use of social media (25%), fake accounts targeting general public, and respectively 22% of high school graduates, have law enforcement (25%), privacy (22%), employee safety (21%), and used social media to share or obtain information during emer- keeping informed of changes in technology (20,9%). gencies and disasters or in severe weather conditions. Users were Finally, Flizikowski et al. (2014) present the only recent survey most likely to seek information about weather, traffic, damage within Europe, conducted among citizens (317 respondents) and caused and information on how other people were coping. Beyond emergency services (130 respondents plus 33 interviews from Fin- that, users shared not only weather information, safety reassur- land, France, Portugal, Norway, Ireland, Great Britain and Poland). ances and their feelings about the emergency but also their loca- The study focuses on the identification of user needs concerning tion, and eyewitness information. In terms of trustworthiness, crisis management with the support of social media and mobile friends, family, news media (or reporters) and local emergency devices. The main goal of the study was to identify possibilities and officials were the most trusted sources, while unknown people in challenges of social media integration into crisis response manage- the general vicinity of the emergency were the least trusted. ment. Generally, the participants had a positive attitude towards Another comparative study published by the American social media. During the study, both citizens and emergency services National Emergency Management Association (NEMA) contains identified the same challenges, such as a lack of knowledge, trained the results of a survey conducted in 2012 among members of personnel, uniform terms of use, credibility of citizen-generated emergency services from all 50 Federal States of the US (San et al., content, and accessibility for older generations. 2013). The survey, which involved 505 respondents, focused on the current degree of use of social media in crisis situations by 2.5. Research gap emergency services and the future development of the organizations in respect of possible use. Additional questions were also It has been shown that social media is used “with pockets of asked regarding general opinions of social media and the trust- use and acceptance among organizations” (Tapia and Moore, worthiness of citizen-generated information. Although the 2014). Benefits - such as additional information, or pictures that respondents indicated a positive attitude towards social media in are of particular value; and challenges, such as the pressure to act, general and valued its suitability for information dissemination, chaotic use, overwhelming amount, limited resources, appropriate 75% of them mentioned the requirement of verifying citizen- technology and trustworthiness – have been summarized. Find- generated content, and otherwise questioned its credibility. ings regarding the perception of emergency services are often However, the main barrier identified was the lack of personnel, based on small numbers of qualitative interviews (Hiltz et al., experience and knowledge to take on additional responsibilities, 2014; Kavanaugh et al., 2011; Tapia and Moore, 2014). C. Reuter et al. / Int. J. Human-Computer Studies 95 (2016) 96–111 99

A few quantitative studies concerning social media and emer- 3. Part III: Use of social media by one's own organization – three gency services already exist. Surveys with several hundred sets of closed questions to gauge current usage, what informa- responses often focus on citizens’ perception (American Red Cross, tion is seen as useful and the main factors to ensure the use of 2012; Canadian Red Cross, 2012). Just four of the studies focus on social media by the organization. This was supplemented by the attitudes of the emergency services: The first one is an annual two open-ended questions to provide further details. survey conducted in the US since 2010 (International Association 4. Part IV: A series of closed questions and one open-ended of Chiefs of Police, 2015); the second one builds on a survey con- question to explore expected changes in the future use of ducted in the US in 2012 (San et al., 2013); the third was con- social media. ducted in the US as well in 2014 (Plotnick et al., 2015), and only the fourth was conducted in Europe (Flizikowski et al., 2014). All We designed the survey based on a strategy aimed at triangula- studies attest a positive attitude towards social media and all tion of micro- (referring to individual perceptions) and macro- level identify challenges in terms of credibility, knowledge and per- (referring to organizational responses) attitudes. This methodological sonnel. San et al. (2013) furthermore reference the knowledge triangulation involved a combination of questions that focus on more required to take on additional responsibilities, and Flizikowski qualitative aspects of the emergency services’ intentions towards et al. (2014) acknowledge a lack of uniform terms of use. However, social media and their usage before, during and after an emergency there is clearly a lack of recent strong evidence of attitudes at a micro-level as well as more quantitative aspects to obtain a towards social media usage in Europe, with most of the evidence comprehensive picture of emergency services’ attitudes towards coming from the US. social media within emergencies at a macro-level. While the study by Flizikowski et al. (2014) impartsinsightsabout The survey was created using the open-source survey applica- social media use and challenges across multiple European countries tion LimeSurvey (〈http://www.limesurvey.org〉). In early Septem- utilizing mainly qualitative survey questions (open questions, the ber 2014, we sent out the link to the online survey to different main intention of which is to identify respondents’ ideas and opi- networks of emergency services as well as to different national/ nions on how social media can be used in crisis response efforts), our international mailing lists, like the Federation of the European study seeks additionally to build on this by providing a combined Union Fire Officer Associations, various Fire and Rescue Units (e.g. analysis of qualitative and quantitative survey questions as well as Fire Brigade Ljubljana and also Dortmund), Firefighters 112 Social shifting the focus to private and organizational attitudes because we Network, EENA Emergency Services Staff Network (ESSN) and assume that differences exist between acting as a private person and Norwegian regional authorities, the civil defense department at acting as an emergency service unit. We also use methods of utili- the County Government and others (see Acknowledgments). zation with regard to both private and organizational social media usage in the present and in the future. With this study we therefore 3.2. Characteristics of survey participants provide a recent insight into the attitudes towards the current and future social media usage during emergencies from the perspective We received 761 survey responses from emergency service staff of European emergency services. across 32 countries. It is important to emphasize that the sample of emergency service staff responding to this survey represents an opportunity sample and, as such, provides a heuristic device for 3. Methodology exploring some questions which are relevant to this study. The largest number of respondents came from Germany (269) fol- Based on our goal to produce a comparative analysis of emer- lowed by Slovenia (134), Poland (117), Denmark (65), Finland (28), gency services’ attitudes towards social media across several Eur- Norway (28), Belgium (23), Italy (17), the Netherlands (11) and opean countries, we decided to conduct an online survey with other countries (70) (Fig. 1). 310 participants (40%) also answered closed (quantitative) as well as open-ended (qualitative) ques- at least one of the qualitative free-text fields. The large majority of tions. The survey have been conducted as part of the EU funded respondents (92%) were male, although the survey did include 54 project “EmerGent”. This section presents the methodology of our female emergency service staff (8%). The largest proportion of study, whereby we will first present the survey design (Section respondents was aged 30–39 years old (29%) and the smallest 3.1), including questions, technical realization and channels of aged less than 20 years old (6%), although, overall responses were distribution. Then we will present a characterization of our par- fairly well distributed across age groups, with similar proportions ticipants (Section 3.2) followed by a description of our quantitative of responses (around 20%) received from those aged 20–29, 40–49 (Section 3.3) and qualitative (Section 3.4) analysis design. and 50 years or older (Fig. 2). The majority of survey participants were full-time employees 3.1. Survey design in Fire Departments (39%), Volunteer Fire Brigades (23%) or the German Federal Agency for Technical Relief (‘Technisches Hilfs- Our survey aimed to identify the attitudes of emergency ser- werk’) (23%). The remaining 16% of participants included a rela- vices, both as a whole as well as individual staff, towards their own tively small number of staff working at Public Service Answering and their organization's current and future use of social media. Points (PSAPs), for Emergency Medical Services, the Police and The survey was designed with the aim of collecting a mixture of other relevant organizations (Fig. 3). The largest proportion of quantitative and qualitative evidence. It consisted of four parts (see Appendix I for details), as follows: 50% 40% 35% 1. Part I: Demographic details of survey participants (age, gender, 30% 18% country of origin, role, type of organization) to explore any 20% 15% 16% differences in responses depending on the characteristics of 9% 10% 4% 4% participants. 0% 2. Part II: Attitudes towards social media – a combination of closed questions (eight-point Likert scale (Likert, 1932) asking participants to rate on a scale of 1–5 how much they agree with a series of statements) and open-ended questions. Fig. 1. Country (Q8). 100 C. Reuter et al. / Int. J. Human-Computer Studies 95 (2016) 96–111 respondents described themselves as ‘members of the crew’ (31%), 2. Factor-analysis of the eight Likert scale questions (see above) on while 23% were Heads/Supervisors of their particular emergency participants’ attitudes towards social media. To measure service. This, as well as the average experience (Fig. 4), suggests respondents’ attitudes towards the use of social media for both that the survey achieved a good cross-section of senior as well as private and organizational purposes, we used the statistical more junior and supervisory-level staff (Fig. 5). technique of factor analysis. Factor analysis is a technique used in research to identify groups or clusters of variables, which, taken together, represent an underlying construct or variable of 3.3. Quantitative analysis interest in the study (Field, 2009). The analysis showed that the factor, consisting of the eight questionnaire items, had high α¼ For the quantitative analysis, the survey data was extracted and reliability with a Cronbach alpha score 0.774 (Cronbach, analyzed using Excel, a free software environment for statistical 1951), which is used to indicate reliability of the scale used. 3. One-way Analysis of Variance - ANOVA (Field, 2009) was then computing and graphics called “R” (http://www.r-project.org)as used to measure any significant differences between the types well as Statistical Package for the Social Sciences (SPSS), a software of respondent in relation to this factor. package for analyzing quantitative data (IBM, 2014). The analysis consisted of three key steps: 3.4. Qualitative analysis

1. Exploring basic frequencies for each question and using cross- The analysis of our free-text survey questions was based on the fi tabulations to explore any signi cant differences across different inductive approach of grounded theory (Strauss, 1987). We used types of respondents. open coding associated with grounded theory to derive categories from the more qualitative free-text answers by careful reading and 50% the aggregation of categories. 40% The firststepwastoextracttheentiredatasetfromthesurvey 29% fi 30% 23% 22% platform into an Excel (*.xls) output le. Accordingly, a second sheet was added which contained only the qualitative results including the 20% 16% response identifier and original language identifier. As the survey had 6% 4% 10% been distributed all across Europe, responses made by the emer- 0% gency service staff were in different languages wherefore there was a under 20-29 30-39 40-49 50-59 60 + need for translation. For each question, two columns for translation 20 and categorization were added. Thereafter each response was read Fig. 2. Age (Q6). manually and translated into English, if required. The translation was performed by native speakers of the respective languages. Easy translations were performed by translation services, such as Google 50% 39% Translate and supplemented with dictionaries, if single words could 40% not be translated automatically or the translations needed manual 30% 23% 23% adjustment for better intelligibility. These translations have later on 20% 16% been checked by a native speaker. 10% To use the grounded theory-oriented method, the open-ended 0% questions were coded openly and participants’ statements were Fire Volunteer Fire Federal Agency Others divided into categories. Each response was then assigned one or Department Brigade for Technical Relief multiple categories to achieve a quick overview of the interesting and relevant topics. The previously acquired knowledge from the Fig. 3. Organizations (Q1). literature review and quantitative analysis was used to increase theoretical sensitivity. Within the next section we only present 50% 45% those responses that show both positive and negative perceptions of social media and its use by emergency services from an orga- 40% nizational as well as individual perspective. Each quotation is 30% referenced with the participants’ response identifier (e.g. R391). 19% 17% 18% Several similar answers are indicated with a number (e.g. 12). 20% 10% 0% 4. Empirical results under 5 5-9 10-14 15 + In the following sections we present the results of our survey. Fig. 4. Experience (Q5). First we present results regarding personal attitudes towards the

35% 31% 30% 23% 25% 20% 16% 16% 15% 11% 10% 4% 5% 0% Head/supervisor Incident Section leader Member of the Communication Other of organisation commander crew Officer (incl. press)

Fig. 5. Role (Q2). C. Reuter et al. / Int. J. Human-Computer Studies 95 (2016) 96–111 101 use of social media (Section 4.1), then on the current organiza- 45% 39% tional use (Section 4.2), types of information (Section 4.3), 40% important factors (Section 4.4), and insights about the future use 35% of social media (Section 4.5). 30% 27% 25% 4.1. Attitudes towards the use of social media (Q7) 20% 14% 13% 15% 10% 7% Following our questions focusing on the characteristics of the 5% participants themselves (Section 3.2), the second step was to 0% obtain insights into their attitudes towards the use of social media Strongly Agree Neutral Disagree Strongly for private purposes. We asked them to rate their agreement or Agree Disgree disagreement on a scale from 1 to 5 to a series of statements. Fig. 6 Fig. 6. In my private life, I use social media very often (Q7). shows, for example, that almost two-thirds (66%; 27% strongly agreed and 39% agreed) of respondents stated that they used social media very often in their private life. As described in Section 3.3 above, eight of these statements were combined using factor analysis to provide an overall factor score for each respondent on their attitudes towards the use of social media for both private and organizational purposes. We identified significant differences between different types of emergency service staff in their attitudes towards the use of social media for private purposes (as measured by this ‘factor’). We found out that female emergency service staff are much more open-minded and have a more positive attitude towards social media than their male counterparts (po0.05). The significance of the statement, however, is mitigated by the low number of female Fig. 7. Attitudes of emergency service staff (Q7) depending on their age (Q4). participants (8%). In addition to gender, it is unsurprising that the age of the participants has an impact on the attitudes concerning a closer relationship with emergency services (R127), and poten- social media usage. As Fig. 7 shows younger emergency service tially develop a more positive attitude towards institutional staff are more positive towards using social media than older staff emergency service activities: members (po0.01) – this difference was particularly significant when comparing those aged less than 20 years with those aged 50 “A safe community must communicate! It's our job to stay in or over. Emergency service staff in countries with high levels of touch with those we serve each and every day, proving our- social media use2 were, on average, more positive than those liv- selves to be a reliable service for prevention and also to over- ing in countries with lower levels of use (po0.001, Kruskal–Wallis come challenges and crises. We have the knowhow, it's worth test; see Field (2009) as well as Kruskal and Wallis (1952)). sharing. Save lives and spare suffering, at almost no cost. To Almost 60% of all emergency service staff think that social make sure the public trusts us and to maintain their confidence media is important for their organization (Fig. 8). They thought we have to communicate with them wherever they are 24/7!” that potential use cases could be sharing information with citizens (R1901). (83%), keeping in touch with citizens (67%) and improving the “ overview of a situation and therefore raising situational awareness While one participant pointed out that these days, radio and fi ” “ (66%). However, 44% state information is not reliable and 27% television are no longer suf cient as more and more people use ” thought that emergency services are in general too busy to use social media (R391), another argued that social media need to be social media data (Fig. 8). used to deliver information and save lives: When we asked for further comments to explain the partici- “Social media is a valuable part of an emergency organization pants’ attitude (either positive or negative) towards the use of […] because it is quite useful to deliver information to the social media for private purposes, most of them answered this population as well as providing them with a wider view of the question from the point of view of their professional role of their situation. We need to integrate this channel because it helps to particular emergency service unit. There were no significant dif- save lives and it is a very valuable tool for civil protection, given ferences in attitude depending on the staff role of the respondents, the citizens’ tools to protect themselves before, after, during although, unsurprisingly, communications officers appeared to and in the aftermath of a disaster” (R632). have a more positive attitude than other staff members – although fi the sample only included 28 staff working as press or commu- In Germany in particular, the re services mainly rely on fi nication officers. volunteers. On the one hand, it is usually dif cult for them to assign additional personnel to deal with social media (R2101). 4.1.1. Additional comments on the participants' attitude towards Often, this is the result of a lack of time and required expertize social media (Q8) among staff (R3164). On the other hand, social media offers an “ ” The open-ended responses asking participants to provide important medium to recruit voluntary helpers (R2708). How- additional comments about their attitude showed that, for some ever, the survey participants mentioned some caveats regarding emergency service staff, social media provides the opportunity for data quality as well as privacy and issues of social media adoption. organizational self-presentation (n¼10), such that citizens can In some instances, it cannot be guaranteed that the data is correct ’ better appreciate the work of the emergency services (R130), have as the individuals perception may complicate the situation: “How useful data in social networks is depends on who gives 2 http://www.slideshare.net/fullscreen/wearesocialsg/social-digital-mobile-in- this information to the public. It also depends on the very europe/. technical possibilities of information transfer and public access 102 C. Reuter et al. / Int. J. Human-Computer Studies 95 (2016) 96–111

to this data on social networks. So, information […] is always reserved for state authorities” and “it is unlikely they will give it provided by very different people […]. It can happen that the up and we are not certain of being competent” (R654). Further information is inaccurate, accidents exaggerated or data is participants appraise social media as a useful technology that they incorrect […]” (R327). “inevitably […] have to deal with” (R885), without mentioning a precise idea of its use (n¼18): The information provided by unauthorized people or unofﬁcial “ sources might be unreliable (R848, R1733) and could mislead It is really hard to ignore the impact social media has on the organizations (n¼33). Therefore, “emergency services must be way we communicate today; it can be a powerful tool in very careful about using information received on social media shaping the way we want to be perceived and the relationships ” during emergencies” and “information may be inaccurate and may we have with others (R884). not provide a sufﬁcient overview of what is actually happening” (R562). Another challenge is the issue of privacy (n¼6), “because 4.2. Current organizational use (Q9) information is disseminated within seconds and also victims have a right to privacy” (R1028) and organizations possibly “have no In contrast to the general positive attitudes towards social media, the actual organizational use paints a different picture and experience in media ethics” (R1246). For instance, “the right of it has been shown that only a relatively small proportion of control over a picture of your own body could be compromised” respondents utilized such data frequently, particularly during an (R1041). To overcome some issues, participants argued for the emergency. As Fig. 9 shows, almost half share (A2C) information standardization of information processing (n¼6): with the public ‘sometimes’ or ‘often’ before an emergency occurs “There is a need for standards. The use could be very reliable (44%); most organizations have never actually shared any infor- and important but it must be very well organized” (R635). mation with the public during emergencies (34%) and 83% (Q7) of emergency services staff nevertheless think sharing information Moreover, other participants argued that the wider adoption of with citizens is an important use case. social media by emergency services is likely to take time (R861) The survey also showed that 19% of the respondents said that and secondly, that “communication concerning major events is social media was used ‘sometimes’ or ‘often’ for receiving

Strongly Agree Agree Neutral Disagree Strongly Disgree 100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% useful to useful to useful for important emergency use in used by information share keep in situational for own services are private live friends not reliable information touch awareness organization too busy Strongly Disgree 7% 1% 2% 2% 3% 4% 1% 7% Disagree 13% 6% 5% 8% 12% 11% 16% 30% Neutral 14% 16% 11% 22% 18% 26% 39% 36% Agree 39% 52% 50% 44% 45% 41% 33% 19% Strongly Agree 27% 25% 33% 23% 21% 18% 11% 8%

Fig. 8. Attitudes towards social media (Q7).

Often Somtimes Occasionally Never Don't know 100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% share information share information search social media with the public two-way with the public sites to gain receive messages about how to avoid communication with about how to situational from the public accidents or the public behave awareness emergencies Don't know 6% 7% 10% 13% 11% Never 27% 34% 33% 42% 48% Occasionally 23% 27% 26% 23% 22% Somtimes 29% 23% 22% 16% 15% Often 15% 9% 9% 7% 4%

Fig. 9. Current use of social media by emergency services (Q9). C. Reuter et al. / Int. J. Human-Computer Studies 95 (2016) 96–111 103 messages from the public during emergencies (C2A) and only 4% upcoming hazards” and using social media to refer “spontaneous said this happened often in their organizations. This is also the volunteers to existing coordination initiatives” (R688), thereby case for situational awareness. Although 66% (Q7) of emergency providing an example of how volunteers could be integrated into services think social media can be used to obtain an overview of relief efforts. Further important tasks were seen to be countering the situation and to raise situational awareness, actually only 23% criticism and providing trusted information to citizens: have often or sometimes used social media sites for this purpose. “The most important task is countering criticism during a dis- Such results clearly show that there is a huge gap between rhetoric aster [for instance during a forest fire], so as to provide trusted and reality in the use of social media by emergency services within information and avoid the collapse of motorways or hospitals, emergency management. which should be available for the emergency services” (R632). However, at the same time, analysis of the data suggested that overall those emergency service staff already using social media For the purpose of information, documentation or posting of frequently in their organizations were significantly more likely to equipment, participants’ organizations shared photos (n¼12): indicate that they expected their organizations to increase their “Updated information on the transportability of the main roads use of social media than other respondents. For example, of 112 during the flood (approx. 15 minute interval). Documented with emergency service staff who said they were already using social photographs. The response was excellent” (R1522). media to share information with the public about how to avoid A widespread use of social media was the publication of a accidents or emergencies, 109 (97%) said they thought that their report (n¼45) – sometimes supplemented with pictures (n¼11) – organizations were likely to increase their use of social media for after the emergency; for example, press releases that recap the this purpose in the future. In contrast, among the 206 who said emergency to “avoid countless questions about what happened” their organization currently never did this, only 95 (46%) thought (R146). Moreover, clean-up activities were part of the effort, for their organization would do so in future. This contrast was even instance “after storms when power lines are down and roads are more striking concerning messages received from the public dur- closed” to locate “fallen trees in a large area” (R1072) or to provide ing emergencies: 94% of the 34 who said they received messages guidance in terms of reconstruction: from the public often thought their organization would increas- “ ingly do so in future, compared with only 37% of the 364 who said After an emergency, we share information. If recovery and this was not currently their practice. rebuilding is necessary, we provide information on damage The qualitative analysis revealed that 18% (26 of 138) of those assessment, how citizens can indicate their losses to the fi who answered the question about personal attitudes towards the authorities, what the of cial procedures are, where and how use of social media in emergencies have concrete concerns donations can be made and what kind of help is needed, and so ” regarding the credibility of citizen-generated content. Simulta- on (R1733). neously, 21% (29 of 138) argued in favor of one-way communica- Furthermore, from a more general perspective, some partici- tion towards citizens. There was little overlap between the two pants reported to represent the organizations’ work practices groups (only 5 individuals belonged to both groups), implying that (R173) or to get information by reading feeds from other autho- overall levels of concern were appreciably higher than might be rities (R912). Also, the use of social media was said to serve as an fi apparent at rst glance. additional channel supporting internal organizational communication (R2911); for instance, to exchange experiences of former 4.2.1. Examples of the use of social media (Q10) similar cases prior to an emergency (R2573). As part of the survey, participants were asked to provide con- To prepare for a possible electricity shortage in Belgium, an crete examples of some of the ways in which their own organi- organization processed social media data: zations had recently used social media during the emergency management cycle. Overall, their comments suggested that such “Extracted the useful data gained from monitoring social use most commonly included: (a) providing warnings, advice and media, and converted it into information to communicate guidance to citizens on how to cope with or prevent emergencies advice at the federal crisis center” (R547). or disasters; (b) disseminating hints and advice on how to behave Although not specified in detail, one participant mentioned the during an emergency as well as coordinating the help of volun- way social media could be used to control the spread of rumors teers, and (c) sharing summary information or reports with citi- (R891) to diminish the dissemination of misinformation. Other zens after the emergency, and coordinating clean-up activities. participants simply gave examples of recent incidents in which While most participants named several activities their organi- they had used social media, including wildfires, floods, pier or city zation had undertaken generally or explicitly before (n¼37), dur- fires, freezing rain, traffic management during incidents and res- ing (n¼64) or after (n¼62) an emergency, others either reported cue efforts. However, some participants reported the use of spe- that their organization did not use social media (n¼25), did not cific social media (Facebook (n¼23), Twitter (n¼11), YouTube specify the use (n¼11) or answered with “do not know” (n¼5). (n¼2), and WhatsApp (n¼2)). Although the sets are rather small, Although partially covered by the quantitative questions, many Facebook was used to provide information and updates (n¼5), participants mentioned the dissemination of preventive measures disseminate articles or reports (n¼3), and to seek or monitor (R560) and, respectively, general behavioral advice (n¼12), infor- information (n¼4): “When it rains heavily, it is soon reported on mation (R688) or warnings (n¼13), and media, like videos (R146) Facebook which city has been worst affected and you can see the first before an emergency. One participant said that their organization images of the actual situation of the people affected” (R3176). In one used Virtual Operations Support Teams (VOST) which “provide case the police asked for information about a car accident via recommendations about how to act before a disaster strikes (for Facebook (R3435). Twitter was mostly used to provide information instance, what to do in case of floods or heavy rain)” (R632). and updates (n¼6) and to disseminate alerts (n¼2). During an emergency, information about the forwarding of From a more critical point of view, a participant identified units (R280), status updates on the current state of emergency issues concerning the reachability, information reaction and (R2647) on the one hand (n¼16), or general tips (R560) relating to overload: different kinds of safety advice (R2102) and recommendations regarding citizens’ behavior on the other hand (n¼10) were dis- “The problem I see with my experience working with poor or seminated. One participant also mentioned “early warning of lower social classes is that many times they do not have the 104 C. Reuter et al. / Int. J. Human-Computer Studies 95 (2016) 96–111

chance to access any type of information or technology. At organizational culture (78%) as well as the skills of staff in using other times they can ignore or over-react to information. I social media (77%). The conditions which were deemed to be less would say that the emergency organizations are overcrowded important were the provisions of funding for staff time (56%), and with useless information and are too close to the sources of the availability of equipment (68%) or software (68%) to access or information. So we cannot build a system that includes a useful analyze the data. Further analysis also suggested that older staff social media tool” (R720). (aged 30 to 49) were significantly more likely to regard the skills of personnel as being important than younger staff (aged 29 or Among those answering that their organization did not use below) in enabling their organizations to use social media – this social media, few mentioned a concrete reason. Either (political) could reflect the fact that older staff have lower confidence in the authorities prevent or prohibit the use (R1417) or the perceived use of social media. unreliability influences them: “Social media is too unreliable as a source of information on the latest threats. In addition, you cannot 4.4.1. Other important conditions enabling social media use (Q13) reach everyone this way” (R3142). 20% of those answering (26 In summary, the open-ended responses largely reflected those from 130) mentioned barriers in terms of trust in citizen- of the quantitative survey in that participants emphasized the generated content. importance of staff skills and knowledge to ensure that informa- These key issues or attitudes mentioned above might hamper tion from social media is accessed and used in their organizations. the successful integration of social media into the relief efforts of In addition, qualitative responses highlighted the need for good emergency services; a point to be discussed within the following practice examples and awareness of recent trends and legal fra- survey question. meworks to ensure social media is used effectively by emergency services. 4.3. Types of information (Q11) To ensure the wider use of information from social media, participants pointed out the importance of experience, which Our survey researched what types of information shared on requires training (n¼9). They designated personnel (n¼8) to be social media by the public would be useful during emergencies responsible for the access of such information. For example, as one (Fig. 10). It reveals, unsurprisingly, that general situational updates participant observed, this could include “a person with knowledge on a current emergency are considered to be more useful or very in this field or a responsible member of society (a volunteer in our useful (73%) than specific information such as details about inju- case) who has the appropriate knowledge and equipment to ries or damage to property. Almost two-thirds of the respondents communicate with the public” (R391), or staff “training and gui- think that both photos (67%) and videos (59%) are the type of dance for simple use” (R877) and “online identity, credibility and information publicly shared on social media which it is very excellent communication skills” (R635). Concerning technical worthwhile to receive during an emergency. However, once again requirements (10), a reliable Internet infrastructure (R2708) and further analysis suggested that all types of information were most accessibility to the scene of the incident must be ensured (R2573). likely to be seen as useful by those who already used social media In addition, there is the need for software to enable the easy dis- to receive or share information with the public often (or at least semination of messages into multiple social media networks: sometimes). “Software should be designed to access all selected social 4.4. Important enabling conditions (Q12) media directly whenever it is used, and to enable the one-click dissemination of text into all (selected) networks” (R1414). Since we assumed a gap between the potential use and the actual current use of social media, we asked which conditions Besides those action- and technology-oriented factors, several could ensure that social media is widely used by the emergency participants emphasized the importance of the organizational services within one's own organization (Fig. 11). The analysis culture (n¼13); for instance, personnel having a positive attitude suggested that the most important enabling conditions were (R861) or even enthusiasm (R2102) towards using social media,

100%Very useful Useful Neutral Not useful Not at all useful 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% General situational Information about Photos Videos Specific information updates the public mood Not at all useful 4% 3% 5% 4% 5% Not useful 8% 6% 10% 9% 13% Neutral 21% 18% 26% 25% 26% Useful 41% 51% 37% 44% 39% Very useful 26% 22% 22% 18% 17%

Fig. 10. Usefulness of different types of information shared on social media by the public during emergencies (Q11). C. Reuter et al. / Int. J. Human-Computer Studies 95 (2016) 96–111 105

Very important Important Neutral Not important Not at all important 100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% Staff skills to Software to Funding for Organisational use social Equipment access social staff time to use culture media media social media Not at all important 2% 1% 1% 3% 5% Not important 4% 3% 6% 7% 9% Neutral 17% 17% 24% 22% 30% Important 44% 50% 45% 45% 38% Very important 33% 28% 23% 23% 18%

Fig. 11. Main perceived enabling conditions for social media use (Q12). supplemented by knowledge of “examples of good practice, to “When we talk about social media and players, we're not encourage the development of using such social networks” (R487). talking about people, information, explanation, mediation, Moreover, a participant quoted the requirement of rethinking reality. The timing of the information may not match reality emergency management scenarios: […]. Also we may believe that we are delivering great information and often we forget the other side (recipient) of the “Understanding that the rules have changed and we live in a information […].” (R1075). completely new scenario, where citizens expect to receive advice and help through social media and where first respon- In any case, a “backup plan” was always said to be required in ders and authorities listen to what the citizens’ share […], case the internet infrastructure or social media communication because those affected by a disaster are an important source of failed (R2215). information. This is a two way channel and it should not be used only to extract information as it has been traditionally 4.5. Future use of social media (Q14) done with press, radio and other communications technologies” (R632). The majority of emergency service staff said they expected their organizations to increase their use of social media in future, On the other hand, survey participants identified factors that particularly in relation to sharing information with the public could limit organizational use, including the possibility that inter- before and during emergencies (Fig. 12). Thus, around three- acting with the public through social media might be restricted for quarters of respondents thought that their organization would legal reasons: increase their use of social media to share information with the public about how to avoid accidents (74%) and how to behave “ A clear allocation and restriction of access to the accounts of during an emergency (73%). A lower proportion, but still more fi the emergency services (is required). It must be clearly de ned than half (54%), also thought that their organization would be which employees can and may perform the information dis- more likely to utilize social media to receive messages from the closure and publication. Free communication between all public during emergencies. fi employees on behalf of the ES is not allowed! Speci c infor- However, at the same time, analysis of the data suggested that mation is sometimes very important and sometimes also overall it was the emergency service staff already using social ” punishable by law! (R1034). media often in their organizations who were significantly more likely to indicate that they expected their organizations to increase In our context, participants also name issues regarding data their use of social media than other respondents. This suggests ¼ “ credibility and protection (n 5). Requiring that the privacy and that although there is a willingness even among those who cur- ” “ data protection must be ensured , especially photos of the danger rently only seldom use social media in the emergency services to ” area (no matter from which side) may constitute a problem expand their use of social media, the growth of this practice and “ (R2973). Additionally, there are risks that information is the uptake of more sophisticated systems is likely to be greatest ” manipulated in social media, or a one-sided depiction takes place among so called ‘early adopters’ or ‘converts’ (Rogers, 2003) – who “ (R2865) and that few posts in social media are objective and already use such technology and have the most positive attitudes effective” (R3562), which is why “you cannot rely unrestrictedly towards social media. on them” and “you must also provide your own impression” (R2865). Also, the compliance with the command structure is 4.5.1. Comments on the role of social media for organizations in 5– important; for instance, there must be clear rules regarding 10 Years’ time (Q15) whether a WhatsApp message constitutes an official order or not Reflecting the findings of the quantitative survey, most (R2215). A participant furthermore emphasized the complexity of responses to a qualitative open-ended question asking participants the information space with its dimensions and possible risks, to explore the role social media might play for their organizations whose questions must be answered with maximum importance in a horizon of 5–10 years time, highlighted the way social media (R1075): was likely to play an increasingly important role. Concrete 106 C. Reuter et al. / Int. J. Human-Computer Studies 95 (2016) 96–111

Very likely Likely Neutral Not likely Not at all likely 100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% share information share information with the public search social with the public two-way receive during media sites to about how to communication messages from emergencies gain situational avoid accidents with the public the public about how to awareness or emergencies behave Not at all likely 3% 3% 4% 6% 5% Not likely 8% 7% 14% 15% 14% Neutral 15% 17% 29% 26% 26% Likely 39% 42% 36% 36% 37% Very likely 35% 31% 18% 17% 17%

Fig. 12. Expected increase of future use of social media in own organization (Q14).

scenarios of the expanded use of social media in future included participant, the process of adoption itself and responsible sub- using it to recruit employees (R2880) and as the “main commu- sequent use must be planned precisely to ensure social media nication channel to promote volunteers” (R3021) or to coordinate complements existing practices effectively: their involvement during emergencies: “Firstly we must create a map of risks on social media. Then we “The ﬂoods in 2013 showed that Facebook and others can reach have to draw up a strategy for 15–20 years. We also need to broad bands of the population, who, however, tried to help create tactics that correspond to each one of the dangerous spontaneously in quite an uncoordinated way. At this point, situations, disasters or incidents which can arise. Create all the control centers and headquarters could intervene to lead such mechanisms for integration into contingency plans, etc.” volunteers to some extent or to keep them away from incident (R1075). scenes in especially dangerous situations […]” (R1414).

Participants also anticipated that “the advent of younger gen- 5. Discussion and conclusion erations” (R280) working for emergency services would inevitably lead to an increasing role for social media. Others emphasized that Recently, by using social media, citizens have acquired new emergency service staff of all ages needed to keep up with the means of communication both within their daily lives and as a latest social trends, including the use of social media, to ensure means of mobilization during emergencies. Emergency services effective emergency management: are confronted with the problem of how to integrate such new “The public are increasingly using social media and therefore methods of communication into their work practices. As other emergency services have to work with this- In the UK, younger studies have already revealed, it is quite obvious from the citizens’ people use mobile phones less often to make a person-to- point of view that the professional emergency services are person telephone call, but they make use of a wider number of expected to recognize citizen-generated content within social apps to connect with a large audience. Emergency services media (American Red Cross, 2012; Canadian Red Cross, 2012). As need to keep up with developing social trends, to find out how San et al. (2013) have shown, most emergency services within the our communities communicate with each other” (R562). US (85%) already use social media, compared to approximately 19– 44% of emergency service staff (Q9) found in this survey focussing In contrast, the participant also commented on the fact that on Europe: 19% often or sometimes receive messages (C2A); 31% access to technology is limited in some areas and that social media do two way communication and 32% share information with the is not used throughout society, which means that emergency public about how to behave or 44% how to avoid accidents or services should not rely solely on information derived from social emergencies. media when making decisions: Thus with regard to attitudes towards social media, the emer- “However we must remember that not everyone uses social gency services in the US cannot necessarily be compared to those media in the same way and some of the older generation use of Europe. How European emergency services are disposed social media less frequently, if at all. Emergency services need towards the use of social media for private and organizational use fl to recognize the diverse communities they serve as well as the as well as the levels and main factors in uencing their current and technological difficulties in some areas, where mobile phone likely future use of social media in their organizations is therefore signals and 3G or 4G is poor” (R562). still an open question, which we tried to address within this article. Other survey participants indicated that the adoption of social media depends on “…the future decision-makers. Today's gen- 5.1. Main results eration is denied the current possibilities of active communication with citizens; a short introduction to these information and Emergency service staff reported that their organizations were communication channels is unlikely” (R1034). According to one currently most likely to use social media to share information with C. Reuter et al. / Int. J. Human-Computer Studies 95 (2016) 96–111 107 the public (A2C) about how to avoid accidents or emergencies knowledge and excellent communication skills are required. On (Q7). However, only 15% (Q9) of respondents said they did this the technical side, it demands an available and reliable Internet often. Less than half the respondents said social media was used to infrastructure, including software artefacts that support a user in receive messages from the public (C2A) at least occasionally and dealing with multiple social networks. A positive attitude and only 5% said this happened often in their organizations (Q9). examples of good practice could influence the use of social media Nevertheless, the survey - and especially the qualitative answers positively. Moreover, emergency services must keep up with (Q10)-revealed several organizations already use social media in changing scenarios, trends, and habits of communication in social several phases of the emergency management cycle (Fig. 13): media. Barriers of usage may result from legal concerns such as European emergency service staff uses social media for differ- data protection, internal organizational compliance issues or ent purposes within different phases of an emergency: (a) Before, mistrust, as well as from the perceived complexity of social media they tend to use it to release preventive information and recom- information spaces. mendations; (b) During an emergency, organizations disseminate This study also has limitations: We do not differentiate tips, safety advice, status updates, and warnings, or they monitor between different social media services. Furthermore, we do not social media activity. Moreover, social media is sometimes used for analyze differences between answers from the 32 countries, since internal communication and sharing experiences; (c) After emer- the number of participants of each country was not high enough to gencies, it is used to share reports enriched with multimedia draw clear conclusions from it. Because participants from certain content or to coordinate clean-up activities through social media. countries and emergency service roles as well as females were Problems discussed here are the unequal reachability of citizens, rather underrepresented in this sample, and since we do not dis- as lower social classes or older generations potentially have no or tinguish between different social media services, upcoming stu- only limited access to social media. dies should take these issues into account to be able to compare The majority of emergency service staff expected their organiza- e.g. culture-, role- and social-media-specific differences. Also, tions to increase their use of social media in future (Q14), particularly further in-depth research is required to analyze how to overcome to share information with the public about how to avoid accidents negative attitudes towards social media use. and how to behave during emergencies. However, the emergency ‘ ’ service staff already using social media often in their organizations 5.2. Relationship with related work were significantly more likely to indicate that they expected their organizations to increase their use of social media than other We partly confirmed the results previously obtained by Flizi- respondents. This means that while some emergency services are kowski et al. (2014) and extended it in relation to the way in which likely to increase their usage over the coming years, others may not current usage tended towards one-way communication, i.e. the do so at all or only in small incremental ways. provision of information (A2C) and the collection of information as The main factors seen as enabling the use (Q12) of social media part of the monitoring process (C2A). Also, both studies come to by emergency services were seen to be staff skills and an organi- the conclusion that the use of social media is generally regarded zational culture open to the use of such information (Fig. 14). The positively. We also obtained results with regard to the concerns open-ended question (Q13) exposed additional enabling condi- and challenges regarding future use. Challenges for the future use tions: To ensure wide use, trained personnel, appropriate became apparent from the statements expressed by the respondents: In both cases, lack of expertize and human resources as well as uniform usage conditions were expressed. Similarly, a lack of trust in relation to citizen-generated information was mentioned. A comparison with the study of the Canadian Red Cross (2012) is more difficult, as the geographical and cultural conditions differ from those in Europe. This can lead to different results when evaluating the surveys, particularly as not only natural disasters but also cultural and political events such as riots or attacks are very different from those in Europe. Consequently, the local emergency services and citizens experience such exceptional situations in a very different manner too (Flizikowski et al., 2014, Fig. 13. Emergency Management Cycle enhanced with Social Media (based on Q9 p. 709). Furthermore, both studies differ in that in Canada only and Q14) (own illustration). citizens were interviewed about their expectations regarding the

Guidelines Examples of good practice Enablers Positive Attitude Technological access

Strategic alignment Social trends

Future Organisational Internal factors External environment Use of Social Media

Human resource allocation Data protection

Organizational scepticism Data credibility & quality Barriers (mistrust, complexity) Law compliance

Fig. 14. Internal and external enablers and barriers (own illustration). 108 C. Reuter et al. / Int. J. Human-Computer Studies 95 (2016) 96–111

Fig. 15. Infographic of the main results (own illustration). use of social media in relation to the emergency services. However, responding (San et al., 2013, p. 50). In our survey, one of the most the case studies contained in the paper are particularly interest- frequently mentioned concerns was the trustworthiness of social ing: They show not only an example of how social media in media data. Canada is already being used, but they draw parallels between the One major difference was found: While the details of the atti- Canadian and European authorities. The police in Toronto trained tudes of European and US authorities towards social media in 300 employees to deal with social media to improve the response disaster situations are mostly similar, the main difference is the to disaster situations in terms of providing situational information extent of their practical implementation: As already stated, within (A2C) as well as for monitoring social media (C2A) to respond to the American study all national and 85% of local authorities emergency calls if necessary or to correct misinformation (two- reported using social media already (San et al., 2013, p. 2). The way-communication). This shows that at least in some Canadian regular use rate detected in our study however lies between just emergency services there is a demand for skills in dealing with 19% and 44% of emergency service staff, depending on the appli- social media. Meeting these needs is, according to our study, one cation. The type of use, the authorities are agreed, still resides of the main prerequisites to enable its future use. mainly in the dissemination of information to the population The study published by the American National Emergency (A2C) (Fig. 15). Management Association (San et al., 2013) as well as the study on US emergency managers (Plotnick et al., 2015) are geographically and thus also culturally and politically based in the US. However, Acknowledgments they are similar in so far as both surveys are addressed towards emergency services and include both quantitative and qualitative The research project ‘EmerGent’ was funded by a grant of the questions which have many substantive parallels. When compar- European Union (FP7 No. 608352). We would like to thank all ing the results of these studies, it becomes apparent that there are members of our project for their remarks and for distributing our many similarities in the use of social media by the emergency survey within the following networks: Federation of the European services in America and in Europe, despite the varying environ- Union Fire Officer Associations (FEU), EENA Emergency Services mental conditions. Most obvious is the pace - but not the direction Staff Network (ESSN), German Federal Agency of Technical Relief - of change and that it varies a great deal as a result of cultural and (THW), Fire Brigade Ljubljana, Slovenia Firefighters 112 Social organizational factors. Network, District Headquarters of the SFS (Poland), Fire Depart- Thus, all studies conclude that the setting for the use of social ment Dortmund (FDDO), German Fire Service Association (DFV), media is generally positive, even if the widespread implementation of their use is often hampered by a lack of resources, German Fire Protection Association (vfdb), Association of Fire experience or lack of knowledge, or is at least limited by these Departments in North-Rhine Westphalia (VdF), Association of fi shortcomings. The most important factor is staff time as the heads of German re services (AGBF), Norwegian regional monitoring and the active use of social media is time consuming. authority, and the Global Fire Service Leadership Alliance. Organizations, especially in larger or more protracted civil cases, cannot use social media without additional, competent, staff. Appendix I. Survey description and questions Similarities can also be found regarding concerns about trust: 75% of respondents wanted to check the messages first before See Fig. A1 here. C. Reuter et al. / Int. J. Human-Computer Studies 95 (2016) 96–111 109

Fig. A1. Online survey.

The survey aims to find out the attitudes of emergency service Officer (incl. press); PSAP Supervisor (public-safety answering staff towards their own and their organization's current and future point); PSAP Operator (public-safety answering point); Member use of social media. Social media, such as Facebook, Twitter or of the crew; Other) YouTube, allow people to create, share or exchange information in 3. Q3: *How many years have you been working for Emergency virtual communities and networks. Such information could be Services? (under 5, 5–9, 10–14, 15 þ) useful to Emergency Services during emergencies. 4. Q4: *What is your age? (under 20; 20–29; 30–39; 40–49; 50– I agree to complete this questionnaire for the EmerGent pro- 59; 60þ) ject, asking me about my opinions on social media and that my 5. Q5: *What is your gender? (Female; Male) participation is voluntary. The results of this survey will be used 6. Q6: *What country do you live in? for the purpose of research and not for profit; all information gathered in this survey will be retained, accessed and analyzed by Part II: Your own attitudes towards social media the project researchers for the purpose of this project only. My anonymity is guaranteed and I will not be identified in publica- 1. Q7: *Please indicate how strongly you agree or disagree with tions or otherwise without my express written consent. the following statements. (Strongly Agree; Agree; Neutral; Dis- Welcome, the following questions will take only about 5–10 agree; Strongly Disagree) minutes to answer. We thank you in advance for your participa- - In my private life, I use social media very often.-Most of my tion. For the completion of the questionnaire it is mandatory to friends use social media to keep in touch.-Information provided fi answer all questions, unless the question speci es otherwise. on social media during an emergency is often not reliable.-It is important for Emergency Services to use social media to keep in Part I: Personal details touch with the public during emergencies.-Social media are an important tool for Emergency Services like the one I work for.- 1. Q1: *What type of organization do you work for? Fire Depart- Social media could be useful for gaining situational awareness ment; Volunteer Fire Brigade; Emergency Medical Service; Police information during emergencies.-Social media could be a useful PSAP (Public Safety Answering Point); Other type of organization tool for Emergency Services to share information with citizens.- 2. Q2: *What is your main role in this organization – the role you Emergency Services are too busy to use social media. spend most time on each day? (Head/supervisor of organiza- 2. Q8: Add any other comments about your attitude towards tion; Incident commander; Section leader; Communication social media 110 C. Reuter et al. / Int. J. Human-Computer Studies 95 (2016) 96–111

Part III: Use of social media by your own organization Fuchs, G., Andrienko, N., Andrienko, G., Bothe, S., Stange, H., 2013. Tracing the german centennial flood in the stream of tweets: first lessons learned. In: SIGSPATIAL International Workshop on Crowdsourced and Volunteered Geo- 1. Q9: *Does your organization use social media? (Often; Some- graphic Information. Orlando, USA. pp. 2–10. times; Occasionally; Never; Don't know) Van Gorp, A.F., 2014. Integration of volunteer and technical communities into the humanitarian aid sector: barriers to collaboration. In: Proceedings of the Information Systems for Crisis Response and Management (ISCRAM). pp. 620– - To share information with the public about how to avoid acci- 629. dents or emergencies?-To share information with the public during Hiltz, S., Kushma, J., Plotnick, L., 2014. Use of social media by US public sector emergencies about how to behave?-For two-way communication with emergency managers: barriers and wish lists. In: Proceedings of the Informa- tion Systems for Crisis Response and Management (ISCRAM). pp. 600–609. the public?-To receive messages from the public during emergencies?- Hiltz, S.R., van de Walle, B., Turoff, M., 2011. The domain of emergency management To search social media sites to gain situational awareness during information. In: Van De Walle, B., Turoff, M., Hiltz, S.R. (Eds.), Information emergencies? Systems for Emergency Management. M.E. Sharpe, New York, USA; London, United Kingdom, pp. 3–20. 1. Q10: Please give an example of the way(s) in which your Hrdinová, J., Helbig, N., Peters, C.S., 2010. Designing social media Policy for Gov- organization used social media recently before, during or after ernment: Eight Essential Elements. Center for Technology in Government, an emergency or incident? Albany, NY, Retrieved from:〈http://www.ctg.albany.edu/publications/guides/ social_media_policy/social_media_policy.pdf〉. 2. Q11: *Which of the following types of information shared on Hughes, A.L., Denis, L.A.S., Palen, L., Anderson, K.M., 2014. Online public commu- social media by the public would you find useful to receive nications by police & fire services during the 2012 Hurricane Sandy. In: Pro- during an emergency? (Very useful; Useful; Neutral; Not useful; ceedings of the Conference on Human Factors in Computing Systems (CHI). fi Not at all useful) Hughes, A.L., Palen, L., 2012. The evolving role of the public information of cer: an examination of social media in emergency management. J. Homeland Sec. - General situational updates (relating to the emergency)-Spe- Emergency Manag. 9 (1). http://dx.doi.org/10.1515/1547-7355.1976. cific information (injuries, damage to property etc.) - Informa- IBM, 2014. Statistical Package for the Social Sciences. Retrieved from: 〈https:// 〉 tion about the public mood (panic or calm)-Photos of the www14.software.ibm.com/ . International Association of Chiefs of Police, 2010. Social Media Survey Results. emergency situation - Videos of the emergency situation Retrieved from: 〈http://www.iacpsocialmedia.org/Portals/1/documents/Survey. 3. Q12: *How important do you think are the following factors to Results.Document.pdf〉. ensure that social media is widely used by Emergency Services International Association of Chiefs of Police, 2015. Social Media Survey Results. Retrieved from: 〈http://www.iacpsocialmedia.org/Portals/1/documents/FULL like yours? (Very important; Important; Neutral; Not important; 2015.Social.Media.Survey Results.pdf〉. Not at all important) Kaewikitipong, L., Chen, C., Ractham, P., 2012. Lessons learned from the use of social - Staff skills to use social media - Organizational culture - mediaincombatingacrisis:acasestudyof2011Thaillandflooding disaster. In: Proceedings of the International Conference on Information Systems (ICIS). pp. 1–17. Funding for staff time to use social media - Equipment - Soft- Kaplan, A.M., Haenlein, M., 2010. Users of the world, unite! The challenges and ware to access social media opportunities of Social Media. Business Horizons 53 (1), 59–68. http://dx.doi. 4. Q13: Are there any other factors that are important? org/10.1016/j.bushor.2009.09.003. Kaufhold, M.-A., Reuter, C., 2014. Vernetzte Selbsthilfe in Sozialen Medien am Beispiel des Hochwassers 2013/Linked Self-Help in Social Media using the Part IV: Future use of social media by your own organization example of the Floods 2013 in Germany. (V. Pipek, C. Reuter, Eds.) i-com - Zeitschrift für interaktive und kooperative Medien, 13(1), 20–28. Kavanaugh, A., Fox, E.A., Sheetz, S., Yang, S., Li, L.T., Whalen, T.F., Fox, E.A., 2011. 1. Q14: *Please indicate the extent to which you expect your Social Media Use by Government: From the Routine to the Critical. In: Pro- organization to increase its use of social media in future (Very ceedings of the International Digital Government Research Conference, Mary- land. pp. 121–130. likely; Likely; Neutral; Not likely; Not at all likely) Marc-André Kaufhold, Christian Reuter, 2016. The Self-Organization of Digital - To share information with the public about how to avoid Volunteers across Social Media: The Case of the 2013 European Floods in accidents or emergencies?-To share information with the public Germany. In: Journal of Homeland Security and Emergency Management (JHSEM) (akzeptiert). during emergencies about how to behave?-For two-way com- Kruskal, W., Wallis, W.A., 1952. Use of ranks in one-criterion variance analysis. J. munication with the public?-To receive messages from the Am. Stat. Assoc. 47 (260), 583–621. public during emergencies?-To search social media sites to Latonero, M., Shklovski, I., 2011. Emergency management, twitter, and social media evangelism. Int. J. Inf. Syst. Crisis Respons. Manag. (IJISCRAM) 3 (4), 1–16. gain situational awareness during emergencies? Likert,R.,1932.Atechniqueforthemeasurementofattitudes.Arch.Psychol.140,1–55. 2. Q15: What role do you think social media could play for your Lindsay, B.R., 2011. Social Media and Disasters: Current Uses, Future Options, and organization in 5–10 years’ time? Policy Considerations. CRS Report for Congress. Retrieved from: 〈https://www. 〉 3. Q16: *Are you willing to receive the results of the survey and to fas.org/sgp/crs/homesec/R41987.pdf . Ludwig, T., Reuter, C., Pipek, V., 2015a. Social haystack: dynamic quality assessment take part in a follow-up survey later on in the project? (if yes: e- of citizen-generated content in social media during emergencies. Trans. Hum. mail address) Comput. Interaction (ToCHI) 21 (4), Retrieved from: 〈http://dl.acm.org/citation. cfm?id¼2798442.2749461〉. Ludwig, T., Reuter, C., Siebigteroth, T., Pipek, V., 2015b. CrowdMonitor: Mobile Crowd Sensing for Assessing Physical and Digital Activities of Citizens during Emergencies. ACM Press, Seoul, Korea. References Palen, L., Hiltz, S.R., Liu, S.B., 2007. Online forums supporting grassroots participation in emergency preparedness and response. Commun. ACM 50 (3), 54. http: American Red Cross, (2012). Social Media in Disasters and Emergencies. Retrieved //dx.doi.org/10.1145/1226736.1226766. from 〈http://a1881.g.akamai.net/7/1881/26640/v0001/redcross.download.aka Palen, L., Liu, S.B., 2007. Citizen communications in crisis: anticipating a future of mai.com/26640/PollData/Social〉. ICT-supported public participation. ACM Press, San Jose, USA. Canadian Red Cross, (2012). Social Media during Emergencies. Retrieved from Palen, L., Vieweg, S., Liu, S.B., Hughes, A.L., 2009. Crisis in a networked world: 〈http://www.redcross.ca/cmslib/general/pub_social_media_in_emergencies_ features of computer-mediated communication in the April 16, 2007, Virginia – survey_oct2012_en.pdf〉. Tech Event. Soc. Sci. Comput. Rev. 27 (4), 467 480. http://dx.doi.org/10.1177/ Cronbach, L.J., 1951. Coefficient alpha and the internal structure of tests. Psycho- 0894439309332302. metrika 16 (3), 297–334. Perng, S.-Y., Büscher, M., Wood, L., Halvorsrud, R., Stiso, M., Ramirez, L., Al-Akkad, A., Denef, S., Bayerl, P.S., Kaptein, N., (2013). Social media and the police — tweeting 2012. Peripheral response: microblogging during the 22/7/2011 Norway practices of british police forces during the August 2011 Riots. In: Proceedings attacks. In: Proceedings of the Information Systems for Crisis Response and – of the Conference on Human Factors in Computing Systems (CHI), pp. 3471– Management (ISCRAM); 2012. pp. 1 11. 3480. Plotnick, L., Hiltz, S.R., Kushma, J.A., Tapia, A., 2015. Red Tape : Attitudes and Issues Field, A., 2009. Discovering Statistics Using SPSS, Third Edition Sage, London, United Related to Use of Social Media by U. S. County- Level Emergency Managers. In: Kingdom. Proceedings of the Information Systems for Crisis Response and Management Flizikowski, A., Hołubowicz, W., Stachowicz, A., Hokkanen, L., Delavallade, T., 2014. (ISCRAM). Kristiansand, Norway. Social media in crisis management – the iSARþproject survey. In: Proceedings Pohl, D., 2013. Social media analysis for crisis management: a brief survey. Special 〈 of the Information Systems for Crisis Response and Management (ISCRAM). Technical Community on Social Networking, Retrieved from: http://stcsn.ieee. pp. 707–711. C. Reuter et al. / Int. J. Human-Computer Studies 95 (2016) 96–111 111

net/e-letter/vol-2-no-1/social-media-analysis-for-crisis-management-a-brief- San, Y.S., Wardell III, C., Thorkildsen, Z., 2013. Social Media in the Emergency survey〉. Management Field: 2012 Survey Results, (June). Reuter, C., Heger V. Pipek, 2013. Combining Real and Virtual Volunteers through St. Denis, L.A., Anderson, K.M., Palen, L., 2014. Mastering social media : an analysis Social Media. In: Proceedings of the Information Systems for Crisis Response of jefferson county's communications during the 2013 Colorado Floods. In: and Management (ISCRAM). pp. 1–10. Proceedings of the 11th International ISCRAM Conference. pp. 737–746. Reuter, C., Ludwig, T., Friberg, T., Pratzler-Wanczura, S., Gizikis, A., 2015a. Social Stallings, R.A., Quarantelli, E.L., 1985. Emergent citizen groups and emergency media and emergency services? Interview study on current and potential use in management. Public Administration Rev. 45 (Special Issue), 93–100. 7 European countries. Int. J. Inf. Syst. Crisis Respons. Manag. (IJISCRAM) 7, 2. Starbird, K., Palen, L., 2011. Voluntweeters: Self-Organizing by Digital Volunteers in Reuter, C., Ludwig, T., Kaufhold, M.-A., Pipek, V., 2015b. XHELP: design of a cross- Times of Crisis. ACM-Press, Vancouver, Canada. platform social-media application to support volunteer moderators in disasters. Statista, (2015). Leading social networks worldwide as of January 2015, ranked by ACM Press, Seoul, Korea. number of active users (in millions). Retrieved from 〈http://www.statista.com/ Reuter, C., Marx, A., Pipek, V., 2012. Crisis management 2.0: towards a system- statistics/272014/global-social-networks-ranked-by-number-of-users/〉〉. atization of social software use in crisis situations. Int. J. Inf. Syst. Crisis Respons. Strauss, A.L., 1987. Qualitative Analysis for Social Scientists. Cambridge Press. Manag. (IJISCRAM) 4 (1), 1–16. http://dx.doi.org/10.4018/jiscrm.2012010101. Tapia, A.H., Moore, K., 2014. Good enough is good enough: overcoming disaster Reuter, C., Schröter, J., 2015. Microblogging during the European ﬂoods 2013: what response organizations slow social media data adoption. Comput. Supported twitter may contribute in German emergencies. Int. J. Inf. Syst. Crisis Respons. Cooperative Work: J. Collaborative Comput. (JCSCW) 23 (4–6), 483–512. Manag. (IJISCRAM) 7 (1). Rogers, E.M., 2003. Diffusion of innovation, 5th ed. Free Press, New York. Heinrich C. Mayr, Martin Pinzger (Hrsg.): INFORMATIK 2016, Lecture Notes in Informatics (LNI), Gesellschaft für Informatik, Bonn 2016 1805 Semi-Automatic Alerts and Notifications for Emergency Services based on Cross-Platform Social Media Data – Evaluation of a Prototype

Christian Reuter, Christoph Amelunxen, Matthias Moi1

Abstract: The convergence of social networking and mobile media technology is shifting the way how people communicate and gain or share information. People are using social media to agreater extent, also in emergency situations. During disasters throughout the world, such as the 2010 Haiti earthquake, the 2013 European floods, or the terror attacks 2015inParis and2016 in Brussels, this has been illustrated again. Often information about disasters even finds its wayfaster to social media than it reaches regular news companies and emergency services. However, approaches for processing and analysing the vast quantities of data produced have even more potential.Yet many emergency services still havenot found a way to put this potential to an effective use. Within our project EmerGent we are developing a system to process and analyse information from social media particularly tailoredfor the specific needs of emergency services. Theaim is to transform the high volume of noisydata into alow volume of rich content that is useful to emergency personnel. In the first part of this paperwepresent our approach from a user interface perspective. The second part deals with the evaluation of the approach and the derivation of future potentials of the approach. Keywords: social media, emergency management, evaluation

1Introduction

Social media data might be of value for emergency services. However, many studies have identifiedbarriers regarding the high amount of dataemergency services need to deal with before, during and after emergencies. Astudy with 24 district officers in Virginia has demonstrated that the amount of dataand theseverity to detect or recognise relevantinformation is amain barrier [Ka11].Asurvey with 241 emergency services in the US has shown that information overloadisoften the case [Pl15].Another studyin Thailand emphasized ensuing risks of wrong decisions andcorrection of mistakes [Ka12]. As aconclusion from this,there is a request for proper tools to deal with the high amount of social mediadata [Re15a],asidentified in an interview study on current and potentialuse of social media in 7European countries. Basedonasurvey with 761 emergency service staff across 32 European countries the “majority of emergency

1 EU-Project EmerGent (www.fp7-emergent.eu) Matthias Moi, M.Sc. (University of Paderborn; [email protected]) Dipl.-Wirt.Ing. Christoph Amelunxen (University of Paderborn; [email protected]) Dr. Christian Reuter (University of Siegen; [email protected]) 1806 Christian Reuter et al. services has positive attitudes towards social media”, however it is “more used to share information (44%) than to receive messages (19%)”, while a future increase is expected [Re16, S.3]. In our project the strategy is to process and analyse social media datastreams before, during and after emergencies including the steps of (A) information gathering (incl.data preparation), (B) data enrichment, (C) semantic information modelling with ontologies and information mining, (E) informationquality and(F) alert detection [Mo15]. The alert detectionaimstoobserve on-going conversations and trying to identify arising alerts fromthe overall noise. It operates on mined and quality-rated informationtodetect emergency-related eventssuch as “two injured persons detected” [Mo15]. Emergency services might want to know more upon request. E.g. if they receive an alert, they need the option to investigate the originorthe reason forit. Because of that it has to be possible to visualise the dependencies between alerts and single posts, whichinitiated an event (temporal representations, graph-based visualisations). In this paper we present theevaluation of ourconcept,which has been implemented as a prototypefor emergency services, including identified potentials.

2Related Work

For therealisation of the need for information for decision-makers, theuse of social media analytics has to meet specific requirements. As abasis for thecomprehensive analysis of social media asufficient number of platforms has to be connected. Especially in acrisis context, it is notsufficient onlytolook at microblogging services such as Twitter, emergency services should base their decisions on information from various media [Re15a, RV14]. Social mediaisalso highlyheterogeneous, so that the ways of distribution of different types of information varies greatly. An additional factor, which terminated the size of the available data base, is the type of connection. The different platforms offer only limited possibilities to detect an amount of data in agiven period [RS14]. The number of different methods of analysis as wellasthe functional areas, which are covered by this, constitute aseparate criterion in selectingappropriate practices for social media analysis. Thus, applications differ formarket analysis fromthose of crisis managementonthings likefocusing on reputation management, marketing and the customer relationship management [St14]. Other methods of analysis point to ageneral validity for theanalysis of social media, such as sentiment analysis [Sh10]. Additionally, there are methods which are especially suitable for the context of crisis management so that an individual assessment of the provided functionality is required. In addition to automated informationreview and selection,the possibility of individual weighting and validation by experts is an important point in the use of social media data [Ke10, Re15b]. Basically such systems should be able to adapt to changes in thesetting Semi-Automatic Alerts and Notifications for Emergency Services 1807 and to supportemergent collaboration, such as ad hoc needed work in unforeseenevents [Re14]. Especially in relation to the processing of unforeseen patterns it is necessary to activate this expertise anduse. Some approaches try to combine virtual and real activities of volunteers [Re13].Based on this approacheslike XHelp focus on the citizens themselves andthe support of their volunteer activities [Re15c], other approaches suggest to allow monitoring their activities [Lu15]. Market surveysreveal that the market for social media intelligence software is very heterogeneous andfast moving, which is also supported by related research [Po13, Tr15]. The reason is that alarge number of designed systems often provide identical functionality in analysing social media. Existingapproaches focus on the analysis of social media (e.g. Brand Watch, Public Sonar akaTwitcident, Signal Socialmention), provide functionality for organisational interactionand communicationinsocial media (e.g. Coosto, Hootsuite, SproutSocial, TweetDeck) or provide specific functionalities or other approaches likecrowdsourcing (e.g. AIDR: Artificial Intelligence for Disaster Response, SensePlace2, Tweedr, TwitInfo, Ushahidi). The transitions between the groups seem fluidand no clear-cut delineation is possible. Systems trying to generate alerts for emergency services have not been detected.

3Alerts and Notifications: The EmerGent Semi-Automatic Concept for Emergency Services

Following thestrategy [Mo15] to process social media data before, during and after emergencies,the visualisation of relevant information was defined as an importantpart for emergency services in order to handle the vast amount of social media data. In Reuter et al.[Re12]the datagenerating way of communication is defined as C2A(citizen to authorities), whereas citizens report informationrelated to an incident or emergency. Considering social media this can be done directly or indirectly. Fordirect communication social apps or direct messaging canbeused. For indirect C2A communication citizens do notaddress messages to concrete recipients (for instance the public timeline on Twitter). To analyse this information, API of different social media have to be used [RS14]. We also plan for A2Ccommunication using a feedbackchannel allowing answering on posts from social media, e.g. with a warning. In the requirementanalysis, basedonseveralempirical studies with emergency services [Re15a, Re16], we identified that it is not feasible for emergency services to deal with every single message that might be of interested or emergency related [Ak16]. During an emergency thousands of potentially relevantmessages may comeupand this would lead to an information overload [HP13]again. Information has to be transformed into meaningful and manageable parts. To target this issue, we introduced theconcept of alerts and notifications [Mo15] with the aimof“transferring high volume, but unclear information contentinto low volume and rich content suitable for emergency services” (Moi et al 2015., p. 47).Analert is a 1808 Christian Reuter et al. set of classified messages sharing a similar context, which is of particular interest for emergency services. The context is defined by but notlimited to attributes like date, time, location, full text, identified eventtypes or language. The main aimofthe interface is nottopresent hundreds of messages to the user,but to select relevant messages and to cluster them into alerts. Each notification consists of severalmessages fromTwitter, Facebook, Google+, InstagramorYouTube, as long as they belong to the samesetting. To do so, within the information mining process relevantdata is filtered,classified and automatically categorised into these alerts andnotifications. Furthermore, messages are evaluatedfor the informationquality process to estimate differentcriteria’s like timeliness, understandability, believability or completeness in order to supportESin the evaluation of information [Mo15]. Off course the user also has the possibility to adapt the process with specific keywords. Therefore, we call it a semi-automatic progress.

The EmerGent ES-Interface is part of an IT-System (see Figure 1).

Figure 1 Overview of the EmerGent IT-System Semi-Automatic Alerts and Notifications for Emergency Services 1809

The basic concept of the user interface (UI) is shown in Figure 2 where processed messages are aggregated into alerts and notifications.

Figure 2 UI concept for visualising alerts & notifications (incl. zoom)

Followed by this concept a first prototypewas developed to letemergency services interact with social media in acomfortable way. The relevantparts aresummarised in Figure 3. The first functionality is information gathering, e.g. to edit your own search criteria based on location and keywords(Examples: you are acommander and need generalinformation of allevents(e.g. flood); or:you are a special unit (dike expert) and need special information (e.g. #broken, destroyed, damaged, #dikes, dam, hill, #sandbag)). Based on this, different alerts are displayed in a list (Figure 3, top) as well as on amap (Figure 3, bottom left). Using filter criteria, it is possible to filter the notifications/alerts to your specific requirements. The information quality is represented by differentcriteria like timeliness, understandability, believability or completeness (Figure 3, bottom right).Apost popup includes detailed informationand functionality such as sharing, replying and reporting. 1810 Christian Reuter et al.

Figure 3: The EmerGent Emergency Service Interface

4Qualitative Evaluation with Emergency Services

In order to evaluatethe concept,weconducted an enduserorientedevaluation with the aimtoreceive feedbackfrom practitioners/emergency services. It followed the structure of asituated evaluation [Tw94] andwas carried out in different countries. It is important that we do not focus on feedback such as the change of colour or the size of the system. We mainly wanted to receive feedbackonthe functionality of the intendedsystem and how it would be of value fortheir work. The questions were split in two parts: personal details and 10 questions about the system. Semi-Automatic Alerts and Notifications for Emergency Services 1811

4.1Methodology

Scenarios are recognised as an importantstrategy towards understanding the interface between the environment and the systemaswell as ameans to elicit and specify software behaviour [Le97]. Broadly, ascenario is an evolving description of situations in the environment. This scenario approachisapart of theevaluation strategy of EmerGent in twodifferent ways. Thefirst onehappens in theform of astory about a floodin Germany. The story shows underwhichreal life circumstances social media might provide added value to emergency services and to citizens in distress. The story connects a series of events resultingfromthe flood regarding the possible uses of social media [Ak16]. The story is a tool to support the interviewer. It helps to introduce the user to the overall situationwhere the interviewee could use this kind of system. Depending on the individual interviewee, the interviewer decides whethertouse the storyornot. The second one happens in form of precise examples for everyES-Interface function. It demonstrates aconcrete situationinwhich waythisfunction couldbeuseful. This method builds the system context,which is defined by the contextand system boundary [Po09]. The stakeholders areavery importantknowledge source to evaluate the system. If importantstakeholders are unconsidered, it could have negativeeffects on the results [Po09]. The above mentioned evaluation was end user oriented, whichprovides strong basic conditionsfor theidentification of relevant stakeholders. The following stakeholder identificationapproachisbased on an emergency services process analysis. To coverthe different ES processes of the countries it is necessary to abstract them. In this case they were related to time- and local-aspects and the rank of the end user. Local- related aspects regard the place and environment where stakeholders work, existing infrastructure and working conditions. This lead to the identification of three working areas: emergency services in the field (e.g. incident commander), an incident support back office (e.g. public-safety answering point (PSAP), incident commander staff) and a general back office or other authorities (e.g. press and public relations). Regarding alerts and notificationstwo timestamps are relevant: before andafter an official emergency call. Before an emergency call one cannot search for acertain event or place. After an emergency call a concrete event in a determined place or area is existent. The rank reflects, in combination with the work years, theexperience, education and the perspective to situations. Regardingall these conditions, five relevant roles are identified (Table 1). These five roles serve as targetroles for the selection of theinterviewees. In the first evaluation twelve interviewshave been conducted within aperiod of four weeks.All target roles were represented, what was the highest priority in order to get a brought feedback. As theavailability of gold level officers is much more limited than silver and bronze, we decided to have themrepresented in thefinal evaluation. An overview about the distribution of participants is given in Table2. In total, twelve emergency service staffs (E1-E12) from three countries (Germany,Poland andSlovenia) participated in the 1812 Christian Reuter et al. evaluation.

Table 1: identified relevant end-user roles

Table 1: Participants ID Main role Command level Work Age experience in years E1 Incident Commander Silver Under 530-39 E2 Section Leader, Incident Commander Silver 15+ 50-59 E3 Incident Commander, Communication Silver 15+ 40-49 Officer E4 Member of the crew Silver, Bronze 5-9 20-29 E5 Member of the crew Silver, Bronze 10-14 40-49 E6 Head/supervisor of organisation Bronze 15+ 50-59 E7 Incident commander Bronze 5-9 20-29 E8 Head/supervisor of organisation Bronze 15+ 40-49 E9 Member of the crew Bronze 5-9 30-39 E10 Sectionleader Bronze 15+ 40-49 E11 Section leader Bronze 15+ 40-49 E12 Member of the crew, PSAP Operator Silver, Bronze 5-9 30-39

4.2Results I: Potentials of Alerts and Notifications

First Impression (Q1): Nine of the 12 interviewees liked the presented interface or thoughtitwould be helpful,however three could notsee apossible use for it or were not interested. E1 andE3saw its potential especially for thecontrol centre or the high command, E3 also mentionedthe possibility to forward information as very useful. E1 Semi-Automatic Alerts and Notifications for Emergency Services 1813 and E8 said that additional informationwould always be positive. However, the interface still seems to need some improvements as mentionedbyE4and confirmed by E5 with his addition “when it’s finished”. This is based on the limitations of the current development. E12 was alreadysatisfied with the current state and thought it was “simple to understand”. E1 mentioned “each additional source of information is helpful, but it has to be consideredas a tool for a larger scale operation”. Potentials (Q2): Almost all participants were able to identify usefulpotential uses of the ES-interface. Thisincluded forwarding information to the head of operations or the officer-in-charge as wellasinforming the citizens which was mentioned multiple times by the interviewees. Most of them saw its potentialmostly forbig scale incidents and in getting information directly from the scene. E1,E10 andE12 also saw potential for identifying risksoremergencies beforehand. “It could be used to define the development of some event via timeline and geo-location. Therefore, it could be useful to predict the future development of some event – so we can plan ahead with the RSue units’ deployment, logistics” (E12). Thefilteringfunction and location map were also perceived as helpfulbyE1and E12. Just one participantcould not see any potential – this was thesame user (E9) who thought that the interface would not be useful for them overall. Most Useful Functionality (Q5): Forwarding information was once again mentioned as one of the favourite functionalities by theinterviewees (E1, E2, E3,E7). E8 andE10 found the connectionoflocation and incidents also very useful. E6 andE12 mentioned the option to filter the informationthrough owncriteria andtherefore extracting useful “information out of thousands of posts” (E12).

4.3Results II: Challenges of Social Media in Emergencies

Weaknesses and Risks (Q3): The main concern expressed in nearlyevery answer was false information and its impact on operations or the people causing panic. E1 and E12 said that the systemcannot be accurate a100%, ether missing importantinformation due to language barriers or overreaction due to the used wording. E3 thought the main barrier of using the application would include the needed skills in social media and the pure manpower neededtohandle the amount of information. Also E1,E2and E10 mentioned concerns in terms of privacy and dataprotection: “Authorities keep citizens and their social media platforms under surveillance. Citizens could feel monitored” (E1). Additional Functionality Needed (Q7): Only fourparticipantsprovided an answer to this question. E2 said that a better connectionisneeded to other involved actors suchas the police or other emergencyservices, while E3 suggested that the interface needs to distinguish between privateand official accounts. Official ones should be treated as more trustworthy and the interface should also allow the usertoidentify particular accounts as more trustworthy than others. 1814 Christian Reuter et al.

Challenges of Social Media in Emergencies (Q8): The interviewees saw false information and possible negative consequences on acting on such false informationas the main challenges. This is followed by aneed to be able to filter out unimportantand unhelpful information, as well as lacking skills in the handling of social media.

4.4Results III: Impact on the Work

Work Improvement (Q4): Themajority of the participants thought that the application could helpthem, especially in the use case of information and communication. As the additional textanswers also showed, it was particularly seen as potentially helpful for bigand mass events. Generally, the opinion seems to be positive towards the interface andits capability of supporting the interviewees’ jobs. However, the answers differ in which use case it would helpbut, with the exception of only one, they all got at least six approvals. (Q9) Themajority of the participants woulduse the application for the reasons already mentioned in the previous questions; the main reason for not using it was lack of time or not using social mediaatall. Asuggestion mentions: “keep it simple on the UI and complex in the backend” (Q12).

5Discussion and Conclusion

Social media plays a role in emergency management, howeverthe amount of information sometimes exceeds the ability to work with them.Sometools exist to deal with this amount, although fewer systems were tailored foremergency services whodo not have time to work with this system in full time (section2). Therefore – based on related work –our approach comprises alerts, notifications andgrouping of messages from social media with apossible relevance for emergency services. These are generated in asemi-automatic way, which means that thegeneration progress is adjustable to the individual needs of the officer in charge (section 3). As acontribution of this article, a system, including this functionality, has been evaluated (section 4), detecting the following findings (table 1):

Table 2: Summary of the findings Aspect Summary Potentials forwarding informationtothe head of operations, identifyingrisks, filteringusingown criteria –precise information(notifications) are needed(Q2, Q5), especially in mass events (Q4), semi-automatic approaches can help Weaknessesfalse information (Q3) and negative consequences (Q8) Final remarks “keep it simple on theUIandcomplex in thebackend” (Q10) Semi-Automatic Alerts and Notifications for Emergency Services 1815

The majority of the participants expressed a positive attitude towards the ES-interface and they see potential for supportingthem in their work. Only a few did notthink it would be helpful, mostly because they do not have the time to use it or are notfamiliar with social media at all. Interviewees thought that being abletoforward social media informationtoothers is oneofthe most useful features and they wanted the interface to allow them to communicate with other services and organisations involved in crisis situations. Another positively perceived feature is filtering: it helps ES to deal with the massive amount of posts andtherefore allows focusing on the helpful informationonly. On theother hand,participantsthoughtthatthis might also be problematic, as an automatic system cannot replace the human expertise and might therefore lead to missing important information. In this context,E1andE2recommend to checkthe EmerGent information in theback office (e.g. public-safety answering point) by using humans and to share relevant informationwith theincident commander.However, EmerGent notifications are meant to do this in a semi-automatic way. Unfortunately, most participants were not able to answer the question about the EmerGent notification as it was impossible to demonstrate this in the current version of the interface. Themain challenge andproblem of social media use in general, in emergency services andincrisis situations at all, is the possibility of false information and how this could lead to wrong decisions during acrisis. Interviewees were worried that false information or rumours shared viasocial media could cause panic – this was one of the main reasons why many were reluctant to rely on social media during emergencies. In future work we aimtoimprove our system in order to detect alerts even more appropriate and to address the weaknesses.

Acknowledgements

The research project EmerGent was funded by a grant of theEuropeanUnion (FP7 No. 608352). Thisarticle is based on Deliverable 3.7 “Potentials of Social Media Usage by EMS and citizens’ involvement in the EMC enabled by EmerGent”.

References

[Ak16]Akerkar, Rajendra; Friberg, Therese; Amelunxen,Christoph: EmerGent Deliverable 3.5: User Requirements, Version 2. Paderborn, 2016 [HP13]Hiltz, SR; Plotnick, Linda: Dealing with Information Overload When Using SocialMedia for Emergency Management: Emerging Solutions. In: Proceedings of the Information Systems for Crisis Response and Management (ISCRAM). Baden-Baden, Germany, 2013, S. 823–827 [Ka12] Kaewkitipong, Laddawan; Chen,Charlie; Ractham, Peter: Lessons Learned from the Use of Social Media in Combating a Crisis: A Case Studyof2011 Thailand Flooding Disaster. In: Proceedings of the International ConferenceonInformation Systems (ICIS). 1816 Christian Reuter et al.

Orlando, USA, 2012, S. 1–17 [Ka11] Kavanaugh, Andrea; Fox, Edward A; Sheetz, Steven; Yang, Seungwon; Li, Lin Tzy; Whalen, Travis F.; Shoemaker,Donald; Natsev, Paul; u. a.:Social Media Use by Government: From the Routine to the Critical. In: Proceedings of the International Digital Government Research Conference. Maryland, USA, 2011. [Ke10] Keim, Daniel; Mansmann, Florian; Thomas, Jim; Keim,Daniel: Visual Analytics : How Much Visualization and How Much Analytics ?In: ACM SIGKDD Explorations Newsletter Bd. 11 (2010), Nr. 2, S. 5–8 —ISBN 1931-0145 [Le97] Leite, J C S P; Rossi,G;Balaguer, F; Maiorana,V;Kaplan, G; Hadad, G; Oliveros, A: Enhancing a requirements baseline with scenarios. In: Requirements Engineering,1997., Proceedings of the Third IEEE International Symposium on, 1997, S. 44–53 [Lu15] Ludwig, Thomas; Reuter, Christian; Siebigteroth,Tim; Pipek, Volkmar: CrowdMonitor: Mobile Crowd Sensing for Assessing Physical and Digital Activities of Citizens during Emergencies. In:CHI ’15: Proceedings of the 33rd Annual ACM Conference on Human Factors in Computing Systems. Seoul, Korea : ACM Press, 2015, S. 4083–4092 [Mo15] Moi, Matthias; Friberg, Therese; Marterer, Robin; Reuter, Christian;Ludwig, Thomas; Markham, Deborah; Hewlett, Mike; Muddiman,Andrew: Strategyfor Processing and Analyzing Social Media Data StreamsinEmergencies. In: Proceedings of the International Conference on Information and Communication Technologies for Disaster Management (ICT-DM). Brest, France, 2015 [Pl15] Plotnick, Linda; Hiltz, Starr Roxanne; Kushma, Janea;Tapia, Andrea: RedTape : Attitudes and Issues Related to Use of Social Media by U.S.County-Level Emergency Managers. In: Proceedings of the Information Systemsfor Crisis Response and Management (ISCRAM). Kristiansand, Norway, 2015 [Po09] Po, Klaus: Basiswissen requirements engineering.Aus- und Weiterbildung zum Certified Professional for Requirements Engineering : dpunkt, 2009 [Po13] Po, Daniela: Social Media Analysis for Crisis Management: A Brief Survey. URL http://stcsn.ieee.net/e-letter/vol-2-no-1/social-media-analysis-for-crisis-management-a- brief-survey. [Re12] Reuter, Christian; Marx, Alexandra; Pipek, Volkmar: Crisis Management 2.0: Towards a Systematization of Social Software Use in CrisisSituations. In: International Journal of Information Systems for Crisis Response and Management (IJISCRAM) Bd. 4(2012), Nr. 1, S. 1–16 [RS14] Reuter, Christian; Scholl, Simon: Technical Limitations for Designing Applications for Social Media. In: Koch, M.; Butz, A.; Schlichter, J. (Hrsg.): Mensch & Computer 2014: Workshopband. München, Germany : Oldenbourg-Verlag, 2014, S. 131–140 [Re13] Reuter, Christian; Heger, Oliver; Pipek,Volkmar: Combining Real and Virtual Volunteers through Social Media. In: Comes, T.; Fiedrich,F.; Fortier,S.; Geldermann, J.; Müller, T. (Hrsg.): Proceedings of the Information Systems for Crisis Response and Management (ISCRAM). Baden-Baden, Germany, 2013, S. 1–10 [Re14] Reuter, Christian: Emergent Collaboration Infrastructures: TechnologyDesign for Inter- Organizational Crisis Management (Ph.D. Thesis). Siegen, Germany : SpringerGabler, Semi-Automatic Alerts and Notifications for Emergency Services 1817

2014 — ISBN 978-3-658-08585-8 [Re15a]Reuter,Christian; Ludwig, Thomas; Friberg, Therese; Pratzler-Wanczura, Sylvia; Gizikis, Alexis: Social Media andEmergencyServices?Interview Study on Current and Potential Use in 7European Countries. In: International Journal of Information Systems for Crisis Response and Management (IJISCRAM) Bd. 7 (2015), Nr. 2 [Re15b] Reuter, Christian; Ludwig, Thomas; Ritzkatis, Michael; Pipek,Volkmar: Social-QAS: Tailorable Quality Assessment Service for Social Media Content.In: Proceedings of the International Symposium on End-User Development (IS-EUD). Lecture Notes in Computer Science, 2015 [Re15c] Reuter,Christian; Ludwig, Thomas; Kaufhold, Marc-André; Pipek, Volkmar: XHELP: Design of aCross-Platform Social-Media Application to Support Volunteer Moderators in Disasters. In: Proceedings of the Conference on Human Factors in Computing Systems (CHI). Seoul, Korea : ACM Press, 2015 [Re16]Reuter, Christian; Ludwig, Thomas; Kaufhold, Marc-André; Spielhofer, Thomas: Emergency Services Attitudes towards SocialMedia: A Quantitative andQualitative Survey across Europe. In: International Journal on Human-Computer Studies (IJHCS) (2016) [RV14]Ruggiero, Aino;Vos, Marita: Social Media Monitoring for Crisis Communication: Process, Methods and Trends in the Scientific Literature. In: Online Journal of Communication and Media Technologies Bd. 4 (2014), Nr. 1, S. 105–130 [St14] Stieglitz, Stefan;Dang-Xuan, Linh; Bruns, Axel; Neuberger, Christoph: Socialmedia analytics.In: Businessand Information Systems Engineering Bd. 6 (2014), Nr. 2, S. 89– 96 [Sh10] Sheth, Amit; Purohit, H; Jadhav, A; Kapanipathi,P;Chen, L: Understandingevents through analysis of social media. In: The Ohio Center of Excellence in Knowledge- Enabled Computing (Kno.e.sis) (2010), S. 1–4 [Tr15] Trateral Research & Consulting: Comparative Review of Social Media Analysis Tools for Preparedness. URL http://Trateralresearch.com/wp-content/uploads/2015/08/ GDPC_SMAT_Short-Report-for-GDPC_Final.pdf [Tw94] Twidale, Michael; Randall,David; Bentley, Richard: Situated evaluation for cooperative systems Situated evaluation for cooperative systems. Lancester, UK, 1994

Appendix: Evaluation Questions

Part I: Personal Details

1. What type of organisation do youwork for? 2. What is your main role (the role youspend most time on) in this organisation? 3. What is your command level? 4. How many years have you been working for Emergency Services? 5. What is your age? 1818 Christian Reuter et al.

6. What is yourgender? 7. What country do you live in?

Part II: 10 Questions on the Application

1. What is your first impression? 2. What potentials does the applicationinvolve? 3. What weaknesses andrisks does the application have? Howcould they be improved? 4. Would a final version of the application helpyou doing your job better? In which use cases (not)? 5. What functionality of the applicationdoyou find most useful? 6. What is your evaluation of the “EmerGentNotifications” (summary of social media posts) and of the “InformationQuality”? 7. Is there any additional functionality youwould like the applicationtohave? 8. What are the greatest challenges of using social media in emergencies? 9. If the application was operationally available, would you use it?Why (not)? 10.Doyou have final remarks? Homeland Security & Emergency Management 2016; aop

Marc-André Kaufhold and Christian Reuter* The Self-Organization of Digital Volunteers across Social Media: The Case of the 2013 European Floods in Germany

DOI 10.1515/jhsem-2015-0063

Abstract: For almost 15 years, social media have been regularly used during emergencies. One of the most recent, and instructive, examples of its widespread use during a large scale scenario in Europe were the 2013 European floods. Public reporting during the event indicated, and our analysis confirms, that Twitter, Facebook (FB), Google Maps and other services were frequently used by affected citizen and volunteers to coordinate help activities among themselves. We conducted a qualitative analysis of selected emergent volunteer communities in Germany on FB and Twitter among others, and subsequently conducted interviews with FB group founders and activists. Our aim was to analyze the use of social media during this particular event, especially by digital volunteers. Our study illustrates the relevance of social media for German citizens in cases of disaster, focusing especially on the role of the moderator. Our specific emphasis was the embedding of social media in the organizing work done by said volunteers, emphasizing both the patterns of social media use and the challenges that result. We show that different social media were used in different ways: Twitter was used in the main for status updates while FB-pages were mostly intended to provide an overview. FB-groups also coordinated a multitude of activities.

Keywords: emergencies; ethnography; Europe; Germany; qualitative; social media; volunteers.

1 Introduction

Social media have been used in a wide variety of disaster scenarios occurring in the recent past, one prominent example being the 2013 European floods. During the flood, a large number of professional and volunteer emergency

*Corresponding author: Christian Reuter, Institute for Information Systems, University of Siegen, Kohlbettstr. 15, 57072 Siegen, Germany, e-mail: [email protected] Marc-André Kaufhold: Institute for Information Systems, University of Siegen, Kohlbettstr. 15, 57072 Siegen, Germany

Angemeldet | [email protected] Autorenexemplar Heruntergeladen am | 05.03.16 09:51 2 Marc-André Kaufhold and Christian Reuter personnel as well as citizen volunteers were involved in relief activities. The latter are relevant because official authorities only have limited resources and hence volunteers may have an advantage in relation to the immediacy of information. Based on a shorter, preliminary analysis that has been published in German (Kaufhold and Reuter 2014), this article aims to uncover real and virtual self-help activities that are initiated and coordinated in the context of social media. Widespread social media such as Facebook (FB) and Twitter in particular are now commonly used during emergencies. Founded in 2004, the social network FB is one of the most visited sites worldwide, counting some 1.36 billion active users per month. Users generally maintain a profile, a friendship network, provide text, videos or pictures inputs and can set up or join groups. Twitter, founded in 2006, is a micro-blogging service with 284 million active users per month. Public short messages (Tweets) with a maximum length of 140 characters are at the core of Twitter. Crisis mapping efforts can also be observed in tools like Google Maps (Meier 2012), a web mapping service that offers, satellite imagery, route planning and collaborative mapping functionality with annotation support, among others. These services provide appropriate structures and functions for supporting communication and coordination between stakeholders and volunteers before, during or after emergencies. In our study, we focus on Germany, where 56% of all internet users are active on FB. Active Twitter usage is however considerably lower, counting only 6% of internet users (BITKOM 2013). Given the disparity in overall usage, the analysis of FB is of particular interest to us. Despite evidence that FB is used more extensively, a lot of research on social media in emergencies tends to concentrate on the role of Twitter, as indicated by Reuter et al. (2012). We therefore suggest that paying more attention to other media such as FB or Google Maps may add value to the research discourse. We contribute to this issue with a case study that examines the use of FB, Google Maps, Twitter and additional media during the 2013 European floods. Our research questions are as follows: How were different types of social media used during the European floods? How was social media used to organize volunteers? And: What are the challenges for collaboration support?

2 Related Work

The use of social media in emergencies has been extensively researched. For nearly 15 years, the public has used social media in crisis situations (Reuter et al.

Angemeldet | [email protected] Autorenexemplar Heruntergeladen am | 05.03.16 09:51 The Self-Organization of Digital Volunteers across Social Media 3

2012): after the terrorist attacks of September 11th 2001 for example, wikis created by citizens were used to collect information on missing people (Palen and Liu 2007), while photo repository sites were used by citizens to exchange information following the 2004 Indian Ocean tsunami (Liu et al. 2008).

2.1 Cases of Social Media Use

Many published research papers, as we have suggested, have mainly focused on the use of Twitter during various disasters in the USA since 2008. This in turn leads to the observation of differences between the use of Twitter in crisis situations as compared to general use (Hughes and Palen 2009). Observing the Tennes- see River technological failure in 2008 outlined the phenomenon of broadcasting emergency information (Sutton 2010) while the Red River Floods in 2009 highlighted the broadcasting by people on the ground as well as activities of directing, relaying, synthesizing, and redistributing (Vieweg et al. 2010). Studies on the Oklahoma Fires of 2009 highlight the role of retweeting (Starbird and Palen 2010), and the attack on four police officers in Lakewood, Washington in 2009 also demonstrates Twitter’s ability to organize and disseminate crisis-related information (Heverin and Zach 2010). Other studies have provided a more international backdrop. The analysis of the 2008 Sichuan earthquake demonstrated that people gather and synthesize information (Qu et al. 2009). For instance, the case of the 2010 Yushu earthquake in China shows that microblogging is used to seek information about other people’s statuses (Qu et al. 2011). Moreover, a study of the 2011 Egyptian uprising shows how the crowd expresses solidarity and does the work of information processing through recommendation and filtering (Starbird and Palen 2012). The mass panic at the Love Parade music festival in Germany in 2010 as well as the 2010 outbreak of the volcano Eyjafjallajökull in Iceland outlined the need for two-way communication (Reuter et al. 2012). This also came apparent in a study of the attacks, which took place in Norway in 2010, where the notion of peripheral response has been developed in relation to emergent forms of agile and dialogic emergency response (Perng et al. 2012). Reuter et al. (2016) summarize the state of the art on emergency services’ perception of social media. Their survey with 761 emergency service staff across Europe points out that the majority of emergency services have positive attitudes towards social media and that it is more used to share information (44%) than to receive messages (19%) (Reuter et al. 2016). A survey with 241 US emergency managers at the county level shows that only about half of these agencies use social media. Although authorities already experiment with the

Angemeldet | [email protected] Autorenexemplar Heruntergeladen am | 05.03.16 09:51 4 Marc-André Kaufhold and Christian Reuter use of different tools, there is a lack of a clear understanding of what tool to use, as detected in interviews in different European countries (Reuter et al. 2015a). Further research investigates the interrelation between the activities of the authorities, those of professional response teams and of volunteers. A study on the 2011 Vancouver Riots revealed “unintended do-it-yourself-justice” whereby citizens overruled authorities and enforced justice on their own terms and by their own means through social media (Rizza et al. 2014). Another study examined the work of Public Information Officers and argues that members of the public “have a changed relationship to the institution of emergency response” through their use of social media (Hughes and Palen 2012). In relation to use by formal institutions, a study on 2012 hurricane Sandy reveals that very few departments used online channels in their response efforts and that communication differed between fire and police departments and across media type (Hughes et al. 2014). Moreover, it has been shown that citizens can handle those activities that are unlikely to be done by official emergency services, e.g. recov- ering lost pets (White et al. 2014). Another study suggests that volunteer groups in emergencies will in the future be challenged to mature and improve according to these enhanced possibilities, so that “professional responders will begin to rely on data and products produced by digital volunteers” (Hughes and Tapia 2015). Research regarding types of users active on social media began by identifying individual roles and proceeded with the development of role typologies. Hughes and Palen (2009) initially identified information brokers who collect information from different sources to help affected citizens. For Starbird and Palen (2011), the second step was to recognize the actions of remote operators as digital volunteers who progress from simple internet-based activities like retweeting or translating tweets to more complex ones, e.g. verifying or routing information. To further differentiate potential user roles, Reuter et al. (2013) distinguish between activities in the “real” world as opposed to the “virtual” world: real emergent groups (Stallings and Quarantelli 1985), whose involvement usually takes the form of neighborly help and work on-site, and virtual digital volunteers (Starbird and Palen 2011), who originate from the internet and work mainly online. Ludwig et al. (2015b) build on it and address these groups by enabling the detection of physical and digital activities and the assignment of specific tasks to citizens. Based on a timeline and qualitative analysis of information and help activities during the 2011 Super Outbreak, Reuter et al. (2013) suggest a more specific classification of Twitter users in different roles: helper, reporter, retweeter, repeater and reader. Kaufhold and Reuter (2014) additionally suggested the role of the moderator.

Angemeldet | [email protected] Autorenexemplar Heruntergeladen am | 05.03.16 09:51 The Self-Organization of Digital Volunteers across Social Media 5

2.2 Studies of the 2013 European Floods

There is a body of work available which uses data from the 2013 European floods. Mildner (2013), for instance, describes his experience of setting up the flood map of the German city of Dresden. Conducting semi-structured qualitative interviews with members of the administrative staff in the German city of Passau, Wittmann et al. (2015) show that a hybrid governance structure was employed throughout the flooding. Fuchs et al. (2013) confirm the potential of Twitter as a distributed “social sensor” but at the same time highlight some caveats in interpreting immediate results. Kaufhold and Reuter (2014) highlight the excellence of Twitter as a platform for status updates whilst at the same time showing that a variety of virtual and real relief activities was coordinated, mainly in FB-groups. Analyzing Twitter during the European floods, Reuter and Schröter (2015) revealed pragmatic (linguistic) barriers resulting from irony, wordplay, and ambiguity, as well as in retweet-behavior. They argue that existing software has to be adapted and improved for German language characteristics. All these papers identified several challenges and areas for improvement. Some papers have tried to address these challenges: Reuter et al. (2015b) designed, implemented and evaluated “XHELP,” a cross-platform social media application. They conclude that “a cross-platform search, especially with time and location filtering options, contributes to the channeling of the information flow.” Backfried et al. (2015), “aim to take the cross-media approach one step further” by not only including data from multiple social media platforms, but also by combining it with information from traditional media. They “describe some of the links and linking behavior encountered during the floods […] from the perspective of Twitter and FB.” Besides these approaches, which specifically arose from the European floods, Sackmann et al. (2014) present “Hands2Help,” an app-based coordination system intended to support incident commanders and control centers in volunteer coordination.

2.3 Research Gap, or: State of Play

To summarize the state of the art, most studies have focused on the use of Twitter, perhaps due to its frequency of use in the US (Reuter et al. 2012). However, in many European countries such as Germany – where our study pertaining to the Euro- pean floods was conducted – 56% of all internet users actively use FB, whereas the active usage of Twitter is significantly lower at just 6% (BITKOM 2013). One recent case of social media use in emergencies in Europe are the Euro- pean floods. Some studies already engaged with this event. However, either they

Angemeldet | [email protected] Autorenexemplar Heruntergeladen am | 05.03.16 09:51 6 Marc-André Kaufhold and Christian Reuter are design-oriented (Sackmann et al. 2014; Backfried et al. 2015; Reuter et al. 2015b); do not cover the overall event by examining only one particular city (Mildner 2013; Wittmann et al. 2015); or considered “only” Twitter (Fuchs et al. 2013; Reuter and Schröter 2015). Our study aims, then, to cover various media “beyond Twitter” in Europe and study the European floods in three locations across Germany using both social media and interview data.

3 Case Study: 2013 European Floods

The European floods (Figure 1) in June 2013 were an event in which volunteer activity played a major role. Thirty-five federal states in seven European countries had to declare a state of disaster in multiple districts, including 55 districts in Germany including the Federal States of Bavaria, Lower Saxony and Saxony-Anhalt. This disaster caused the deaths of eight people in Germany and cost the states an estimated 6.7 billion Euros. The German armed forces (19,000 soldiers) and fire services (75,000 firefighters) participated in the relief efforts, along with the German Red Cross and other relief organizations who provided medical and other care. Besides the large number of professional forces and voluntary organizations who were involved, many volunteers and affected people participated by building up flood barriers, filling and piling up sandbags, donating work material and goods for victims, distributing food and providing emergency shelters. The use of social media in this case has been acknowledged and even utilized by the traditional media. Volunteer activities were planned and coordinated, emotional support was given and eyewitness reports were shared using these media. Aside from Twitter, FB was the most-used application. One hundred and fifty seven pages and groups relating to the keyword “Hochwasser” (English: flood) were created and more than 600,000 people were members of such groups or liked the pages. For 47% of these communities the focus was on local geographical locations such as counties or towns or districts. Thirty-four percentage dealt with forwarding and exchanging information and 27% considered them to be relief platforms (Karsten 2013). In the following sections we describe three cases and six related FB-communities that we examined within our study. We chose three different areas to establish a modest degree of generalizability across federal borders. We selected FB-pages that indicated above-average activity and reach (in terms of FB-likes) to act as a broadcasting medium (Sutton 2010), and FB-groups representing local (self-help) communities that also showed a brisk degree of activity and

Angemeldet | [email protected] Autorenexemplar Heruntergeladen am | 05.03.16 09:51 The Self-Organization of Digital Volunteers across Social Media 7

UK Lower Saxony (case B)

Saxony-Anhalt (casse A)

Germany

Bavaria (case C) France

Figure 1: Amount of Rainfall and Cases during the Floods. more specifically dealt with the mediation of help supply and demand (Reuter et al. 2013).

3.1 Case A: Saxony-Anhalt

In Saxony-Anhalt, the floods had a significant impact on the town of Magdeburg where about 23,000 people had to be evacuated east of the Elbe River. Also, a transformer station was threatened, meaning that a potential ingress of water would have endangered the electricity supply of 30,000 households for several months. The team of the FB-page, “Flood News Magdeburg,” for instance, aimed to browse through, evaluate, select, and share relevant information about the developing situation. Additionally, the FB-group “Flood Magdeburg – Requests for help” tried to provide a holistic situational overview with links to further groups or pages and information on hotlines and “donation camps.”

3.2 Case B: Lower Saxony

The case of Lower Saxony shows communities seeking to support help activities distributed throughout the whole federal state. The description on the FB-page “Flood Lower Saxony” defines its goal as collecting and exchanging information about the flood, e.g. water levels, photos documenting the damage, evacua- tions and situational trends regarding Lower Saxony. The FB-group “Flood Lower Saxony – Offer/Search” coordinated the communication between people affected and those offering assistance, and presented conventions of use in the group description and as a fixed post.

Angemeldet | [email protected] Autorenexemplar Heruntergeladen am | 05.03.16 09:51 8 Marc-André Kaufhold and Christian Reuter

3.3 Case C: Bavaria

In Passau (Bavaria), the water supply was temporarily contaminated by some of the heaviest floods ever to affect the Danube River. Teaching at schools in the vicinity and the local were disrupted, while in the nearby town of Deggendorf more than 6000 citizens had to be evacuated when the dam failure burst, resulting in the flooding of low-lying areas. The team of the FB-page “Passau cleans up” established a local and structured organization consisting of headquarters, coordinator and scout units to support the overall coordination of relief activities in Passau. The FB-group “Moms help” originating with a group of (pregnant) women, supported relief activities and took an active role in the response and advanced recovery phase.

3.4 Methodology

Our empirical study, which has partly been reported in German (Kaufhold and Reuter 2014), aims to examine the use of social media for the coordination of volunteer relief activities. For this, we initially analyzed six FB-communities, evaluated tweets and conducted seven semi-structured interviews (Table 1). We also make occasional use of further sources. Our work is, then, based on the case

Table 1: The Three Cases of our Empirical Study.

Case A B C

State Saxony-Anhalt Lower Saxony Bavaria FB-Pages Hochwassernews Hochwasser Niedersachsen (Flood Passau räumt auf Magdeburg (Flood News Lower Saxony) (Passau cleans up) Magdeburg) FB-Groups Hochwasser Magdeburg Hochwasser Niedersachsen – Mamas helfen – HilfsGESUCHE (Flood BIETE/SUCHEa (Flood Lower Saxony (Moms help) Magdeburg – Requests – Offer/Search) for help) Twitter Search term: flood (“hochwasser”) Interviews A (moderator) B (moderator) C (moderator) D (helper; employed at E (helper; employed at water F (moderator) police) management) G (helper; employed at fire service) aThe group was renamed to “Hochwasser – die Helfer-Gruppe (Flood – the Helper-Group)” on July 20, 2015.

Angemeldet | [email protected] Autorenexemplar Heruntergeladen am | 05.03.16 09:51 The Self-Organization of Digital Volunteers across Social Media 9 study approach described by e.g. Stake (1995) and Yin (2014) and which allows for the detailed exploration of a phenomenon within some bounded context (see also Creswell 2013). Yin, in particular, allows for comparative work across multiple cases and distinguishes exploratory and explanatory case studies. Our work on the 2013 floods clearly explores data across what we can reasonably describe as a bounded context (the floods), but which nevertheless compares across three individual cases. The FB analysis explored messages in three groups and from three pages from three federal states in order to analyze processes regarding interaction, structuring, and task sharing. Starting on June 6th, FB-group and page activities were downloaded and updated regularly as complete HTML documents to extend and refine our analysis. The last update was performed on October 31st. Additionally, news articles in the context of the floods were archived as PDFs and more than 50 screen- shots were saved in order to document the use of Google Maps in the affected areas. In addition, from June 6th to June 27th 2013 nearly 80,000 tweets containing the term or hashtag “hochwasser” (English: flood) were archived using the tool “Tweet Archivist Desktop.” We used the term “hochwasser” because it was listed among the top trends in Twitter and was mentioned in several media. Our Twitter analysis consisted of two steps: First, we explored the whole dataset to extract quantitative features (e.g. overall number and temporal distribution of tweets, retweets, media, embedded URLs; tweets and retweets per user; other hashtags used along with “hochwasser”) and conceive the structure and topics of the information space. Subsequently we focused on a reduced dataset of 1602 tweets (containing relevant keywords like “urgent,” “need,” “offer” or “help”) to examine how and in what ways relief activities were shaped in and by Twitter. In the next step of our empirical study, we recruited moderators and helpers of the FB-communities in question via e-mail and private FB-messages, conducting both local and phone interviews to gain insights into their work practices, ways of mobilizing resources (Perng et al. 2012) and self-organization processes (Starbird and Palen 2011) as well as identifying the potential ways to support them. The interviews were designed to specifically address our interests. Initially, the interviewees were asked whether they belonged to an organization of civil protection or had competencies regarding professional crisis response. Follow- ing this, we focused on the tasks and activities of the respondents, whereby we were particularly interested in the relationship between online or virtual activity and mobilization “on the ground.” The third part concentrated on task-sharing processes, during which potential cooperation with emergency services was discussed. Finally, the interviews posed questions at an ICT level by evaluating FB functions and discussing technical support potential. Interviews were audio- recorded and transcribed for further analysis.

Angemeldet | [email protected] Autorenexemplar Heruntergeladen am | 05.03.16 09:51 10 Marc-André Kaufhold and Christian Reuter

In our subsequent analysis we employed “open” coding (Strauss and Corbin 1998) i.e. gathering data into approximate categories to reflect the issues raised by respondents based on repeated readings of the data and its organization into “similar” statements. These categories reflect the structure of the following results section. While the analysis was conducted in German, the quotes selected for this article were translated to English by the authors.

4 Results I: The Use of different Types of Social Media and ICT during the Floods

Focusing mainly on the use of FB and Twitter during the floods, the following sections present the analysis of the use and reception of specific functions of FB- groups and pages, as well as the integration of social media with other ICT.

4.1 Use Differences Among Different Services

Although Twitter was used extensively in flood situations, especially for distributing information and situational updates, our results show that it generally tended to take the form of a broadcast medium for shaping cross-platform structures and extending coverage (Reuter et al. 2015b): The coordination of relief activities was rarely observed in the Twitter sphere alone, but took place within a complex ecology of both online and offline facilities. This could, to some extent, be a corollary of low Twitter penetration in Germany, but it is consistent with Hughes et al. (2014) who showed differences in use across media platforms. Accordingly Twitter seems to be used for real-time notification, and FB for community engagement (St. Denis et al. 2014). In our case, local and subject-specific groups emerged on FB, who cooperated extensively with other groups/pages. In addition, the interviewees mentioned their awareness of certain functionalities in social networking which induced them to stick to platforms they were familiar with (in this case FB) during the floods (A, 17:18; B, 50:36; C, 40:26).

4.2 Facebook Group and Page Functionalities

4.2.1 Group Discussion and Group Description

On entering a group or page in FB, the group discussion or the page chronicle (see Section 4.2.3) are the most prominently visible features. While the chronicle lists

Angemeldet | [email protected] Autorenexemplar Heruntergeladen am | 05.03.16 09:51 The Self-Organization of Digital Volunteers across Social Media 11 items sorted exclusively by the date of their creation, in a group discussion, the date of the item comments also influence the position of the item. Exceptions are “fixed postings” which are always at the top of the list and which were used in all cases either to (a) introduce the purpose of the group, (b) present the group’s conventions and use of its functions, (c) give a general overview of the emergency and existing infrastructure, or (d) prioritize important or recent information. In the cases A and B, the fixed posting more or less consisted of the group description, including policy statements, probably because the fixed posting is more accessible and central on the user’s screen. To reflect the group’s purpose, statements like these were disseminated: –– “Help offers are collected here, so that voluntarily helpers have a contact point” (Flood Magdeburg – Requests for help, case A). –– “Although we are no longer under water, there are still plenty of help requests, so we needed to find something new […]. We need YOU as a family mentor” (Moms help, Case C). –– “If ever there is another disaster in future, we would like to offer this platform to mediate help” (Flood Lower Saxony – Offer/Search, case B).

While in case A the statement reflects the general purpose of the group, case C indicates that the focus of activity was shifted during the course of emergency and case B even strives to be a platform for future disasters. In terms of conventions, case A attempts to lay out a protocol, suggesting that a posting such as a help request should contain at least an address and the goods or services (lifts, sleeping-places, etc.) needed, and that users should only publish help requests in a clearly stated fashion. In case B, interested people are told that help requests are organized on the discussion wall and files have been created with information on provisions, shelters, furniture or materials to enable subject-specific communication via comments (see Section 4.2.2). Contrary to case A, they explicitly ask posters not to “post addresses, phone numbers or contact points without consultation.” Two distinct approaches to issues of privacy could clearly be distinguished and are examined below. Both groups mention that they will delete inappropriate postings (case A) or when they have been successfully processed (case B). Furthermore, the mediation process of family mentoring is described in detail in case C. “Flood Magdeburg – Requests for help” in particular tried to give an overview of the emergency by providing (a) a list of other FB-groups and pages that have a more global (e.g. Saxony-Anhalt), regional (e.g. Magdeburg, Elbe-Havel) or topical (e.g. help offers, emergency shelters) focus and (b) a list of official donation hotlines and donation camps.

Angemeldet | [email protected] Autorenexemplar Heruntergeladen am | 05.03.16 09:51 12 Marc-André Kaufhold and Christian Reuter

4.2.2 Group Files and Documents

Group files (resp. documents) were used as descriptive sources and to enable topic- related communication (Figure 2). Interviewee B quotes her previous knowledge and the availability of group functions (files, documents and photos) as crucial factors in the use of FB during the floods, to structure efforts:

“If you create a group you can get people, who you promote to admin, so that you can divide labor. Then I can create folders in a group to sort certain things like help offers and requests in this case, which we managed with files, or that you can use photo albums to classify certain categories. You could structure this, which was crucial for me” (B, 50:26).

While in most cases the administrators managed the document’s title and content, other members used the comment section of the document for communication. For instance, “Moms help” started to mediate “family mentors” in mid-October with the help of FB-documents, whereby one document enabled potential family mentors to show their willingness and the second provided a sorted list of those already registered, ordered by postcode. Other files consisted of lists of one’s own or foreign group administrators, blocked users, group supporters or general information. In another document, members were instructed to publish their help offers, and the document itself gave an example of how to structure these. Also, the files were tagged by the abbreviations “done for now” and “current” or by an exclamation mark to express the status or the urgency (Figure 2, right). Using the file function to organize help offers was described as a critical task:

“We originally tried to do this by using files but within three days we had reached our limitations. Answers were inserted into files as comment, but once there are 200 comments; nobody reads 250 comments to find out: ‘Ah, the gumboots were up there’” (C, 50:12).

Figure 2: “Moms Help” Group Discussion and List of Files.

Angemeldet | [email protected] Autorenexemplar Heruntergeladen am | 05.03.16 09:51 The Self-Organization of Digital Volunteers across Social Media 13

4.2.3 Page Chronicle and Page Info Section

In contrast, the page chronicle is generally managed by the page team members, so that regular users can only comment on page postings and a specific comment section, kept apart from the chronicle itself. Page teams often used the page info section to clarify their objectives. The team of “Flood News Magdeburg” regularly published status updates regarding the severity of the situation, volunteer and material requirements and general information about recommended equipment as well as posts with empathetic intent, such as thanking volunteers for their efforts. They describe their task to browse through existing information, to evaluate it, to select information considered to be relevant, and to share it on its page. It can thus be thought of as generating collective intelligence (Hollan et al. 2000). In addition, the activities of structuring and synthesizing information can be observed on the wall summarizing information on several areas affected by the floods:

Update: 01:50 +++ approx. 30–40 fresh and well rested volunteers are required at Zollstraße to relieve tired helpers +++ +++ The Wissenschaftshafen is debatable as contact point according to our information; please inform us about the current situation via PM. +++ +++ Fresh helpers are also required in Pechau +++ If you have been up and about for a long time, please go to bed and rally your strength for the upcoming days.

Discussions often emerged in the comment section of these messages, where potential helpers initially expressed their need for information and elements of a verification process became visible: There was a higher probability that information would be assumed to be correct if it was confirmed by the page moderator (Table 2, left). In contrast, the statements made by other users were sometimes questioned with greater intensity, indicating a certain degree of self-regulation through the crowd’s mistrust of unconfirmed information (Table 2, right). The photo function was also utilized by several groups and pages (Figure 3). For instance, in “Flood Lower Saxony – Offer/Search” various photos are to be seen which document damage and other effects of the crisis, and the (process of) help activities such as “piling sand bags” or relief and evacuation efforts. Comments on flood-related photos on the one hand usually contained emotional messages expressing solidarity, sympathy and identification with the helpers. Interviewee F states that they published photos intentionally:

“We always uploaded pictures of people […] and I think this was very good for the emotional processing (of the floods) that this mutual photo exchange took place here. And in a way it was good for motivation too, of course.” (F, 24:00).

Angemeldet | [email protected] Autorenexemplar Heruntergeladen am | 05.03.16 09:51 14 Marc-André Kaufhold and Christian Reuter

Table 2: Two Examples of Information Verification.

[02:00] Eric: Where exactly in Pechau and [02:36] Janine: In Westerhuesen I think Zollhaus?? they are also looking (for helpers)!!!! [02:00] Basti: I am driving to Zollstrasse. [02:40] Christian: #Janine Krause sure? Where do I have to register??? Where exactly? [02:02] Thomas: It will be explained at the [03:18] Markus: New contact point. sports field in Pechau. Leave the car there Virchowstraße [02:02] Flood News Magdeburg: In Pechau, [03:20] Jana: How certain? register at the sports field! At Zollstraße, you [03:20] Simon: Is that certain, Markus???? will see it directly on arrival [03:40] Markus: Certain, but they are still [02:03] Mirko: In Penchau you will be met at the constructing on-site. It makes little sense sports field and at Zollstraße everything is lit up to come here yet

Figure 3: Photos with Emotional and Cohesive Appeal (Sources: “Flood Lower Saxony – Offer/ Search”; “Flood Lower Saxony”).

On the other hand, coordination efforts occasionally took place and new help offers were disseminated in the comment section, indicating that the emotional appeal of photos can also serve the purpose of mobilizing additional helpers and resources (White et al. 2014).

4.2.4 Problems with Private Messages: “Oh I didn’t see it”

Besides page and group functions, private messages were also used in certain situations. To filter and verify information as well as support a clear overview, interviewee B adjusted group policy in such a way that comments were not published on the wall until an administrator had confirmed them. Interviewees used private messages to contact helpers and affected people:

“At busy times, especially in the initial phase […], suddenly there were 17 private messages from people who wanted something from the group. Then I could simply open a new tab for each person and work through it and also re-sort the information in files” (B, 50:26).

Angemeldet | [email protected] Autorenexemplar Heruntergeladen am | 05.03.16 09:51 The Self-Organization of Digital Volunteers across Social Media 15

However, this procedure has the disadvantage that messages to contacts, who are not part of one’s friend list, are put into the folder “other” and the user is not notified:

“Many people aren’t aware of the folder ‘other’ and FB gives no notification of other messages. And 2–3 weeks later the answer came back: ‘Oh I didn’t see it.’ For emergency situations, it would be great if you could answer more directly” (A, 33:10).

4.3 Twitter – Retweets and Situational Updates

Representing 17.93% of the entire Twitter dataset, the account @_BB_RADIO_H2O has regularly and automatically published information on water levels based on data provided by the Federal Waterways Administration. As these tweets contain situational updates and geo-referenced information, they contribute to overall situational awareness. Tweets generated by other accounts comprise links to webcams, which give the users an idea of how the situation on the ground looks. With 13.86%, the account @Hochwasser2013 mainly published retweets. Under the accounts @FeuerwehrRT and @THW the fire service and the THW (federal agency for technical relief) retweeted tweets automatically, including the hashtags #hochwasser and #feuerwehr (fire department) resp. #thw. A great number of captured accounts were limited to or mainly disseminated retweets, which constitute 46.45% of the dataset. The news channel “MDR Sachsen-Anhalt” showed a high level of activity with 1071 tweets (1.34%) on the account @MDR_SAN. On June 3rd, 4:03 AM, MDR offered to retweet important messages: “@MDRaktuell: #Flood: Where are #Volunteers needed? Please tweet, we retweet.” Alongside MDR, many other accounts based on traditional media, especially TV and radio stations, print media and their associated websites also produced some of the highest numbers of tweets. To further analyze tweets related to civic relief activities, we reduced the data set by excluding official accounts dealing with politics, media and companies. When reflecting the remaining data, it became apparent that help activities in particular can be identified by recurring expressions such as “urgent,” “needed,” “offer,” or “help.” Filtering the data set according to these expressions alone resulted in a set of 1.602 tweets. Overall, only a small amount actually dealt with relief activities. The majority belonged to situational updates (e.g. water levels). Among the communities examined, only “Flood Lower Saxony” used the account @hochwasserniede to distribute information on Twitter, but each tweet contained a URL referencing their activities on FB. Furthermore, none of the interviewees had a personal Twitter account or considered creating one during the floods.

Angemeldet | [email protected] Autorenexemplar Heruntergeladen am | 05.03.16 09:51 16 Marc-André Kaufhold and Christian Reuter

4.4 ICT used alongside Facebook and Twitter

4.4.1 Geo Mapping with Google Maps

Social media relief activities were supplemented by a vast range of external tools. The team of the FB-page, “Flood News Magdeburg,” for instance, created a Google map (similar work was done in other cities). Flooded areas, shelters, sand bag filling stations, threatened dikes, a need for relief forces or transport facilities were illustrated with the help of various markers, polygons and attached information windows (Figure 4). In addition, the description of the map was used to introduce its purpose or additional resources. In the comment section, numerous comments were made by helpers who appreciated such maps and wanted to support the up-to-datedness of the map with situational updates (Table 3, left).

Figure 4: Google Maps in the City of Halle.

Table 3: Positive Reception and Obstacles of Google Maps.

Marko: The map is really brilliant. Keep Jschlademann: GREAT MAP!!! THANKS! Please it up-to-date, we will gladly continue to update: The physiotherapy practice on Ernst- work with it König-Strasse does not require help anymore. Eriu: Really helpful application for the The annotation can be deleted. MANY THANKS!! coordination of flood relief activities Philipp: Why is not the street at the surface Anon: Many thanks for this great map. mine listed as blocked at the map? It has been Using it yesterday, I found a suitable completely flooded for several days now place of action and after work was Joshwa: Please update, some entries are older done there, I found the next place than 8 h and make the map unclear and useless for potential helpers:/

Angemeldet | [email protected] Autorenexemplar Heruntergeladen am | 05.03.16 09:51 The Self-Organization of Digital Volunteers across Social Media 17

These comments reveal the potential of crisis maps to visualize geo-referenced information, but clearly show the challenges that lie in ensuring information is both clear and up-to-date (Table 3, right). Furthermore, interviewee B indicated problems adapting to Google Maps as she tried to “create a map where volunteers sign in […], but did not understand how to set a marker” (B, 56:59).

4.4.2 Organized Support: Live Blog and Google Docs

“MDR,” the public broadcasting company in some federal states, organized a moderated live-blog: besides selected tweets and FB-posts, manual posts by users (and their own recommendations e.g. accounts for donations or seek/offer-platforms) were made available (Figure 5, left). Furthermore, MDR used Google docs to share a public spreadsheet regarding flood-related information and guidance. In order to obtain and list volunteers’ addresses and phone numbers, the team around interviewee F created a Google form: “And then we created a Google form and distributed it via campus blog and student council Facebook pages. We got more than 1000 entries overnight […]” (F, 01:53).

4.4.3 Custom Web Tools

Custom web tools were developed in addition to the above to structure the process of capturing and matching help requests and offers. In the area around Dresden, tweets were automatically generated by the account @FluD-

Figure 5: MDR Live Blog and FluDDHilfe Website.

Angemeldet | [email protected] Autorenexemplar Heruntergeladen am | 05.03.16 09:51 18 Marc-André Kaufhold and Christian Reuter

DHilfe whenever a user had created a help request or offer on the fluddhilfe. de website (Figure 5, right). By cross-referencing in this way, the developer established a bidirectional connection between both platforms. Interviewee F also mentions the use of the flut.stomt.de website, a similar tool to collect help requests and offers in different regions, highlighting coordinative features: “And it was really nice that you could enter how many helpers were required and click “I will go there” and then the tool’s counter (of required helpers) went down” (F, 35:43). In addition to these publicly accessible artifacts, interviewee C used a local Excel spreadsheet to organize demands from people and collection points, while interviewee B, along with other administrators, used Dropbox to make certain information and documents available to a limited audience of users.

5 Results II: (Self-)Organization with Social Media

Social media was used to advance communication, to supplement existing channels and to support overall organizational efforts to overcome the floods. Data concerning information processing activities and the qualitative characteristics of social media information in this context will be presented in the following sections.

5.1 Organization and Task Sharing in Volunteer Communities

In all the cases we examined, there were elements of task sharing within the internal team of moderators. The team of the Lower Saxony case split tasks as follows:

“She went to Schönhausen and Breitenhagen and gained her own impressions. At the time I was still reacting at a distance. She then called me […]: ‘This and that is needed. Could you please check if anybody has that?” (B, 15:52).

Interviewee B followed up by checking existing FB-documents or asking for offers. In terms of other virtual communities, a private coordination process was established in a way that addresses and persons were treated confidentially for the purpose of data privacy: “That means there is a private group of moderators where there are some admins who made a reliable impression” (B, 43:50). Inter- viewee E also states that many resources were organized privately (E, 06:17) and that too few affected citizens articulated their demands over FB for efficient pro-

Angemeldet | [email protected] Autorenexemplar Heruntergeladen am | 05.03.16 09:51 The Self-Organization of Digital Volunteers across Social Media 19 vision, possibly due to lack of internet or relevant facilities and/or skills within the disaster situation (E, 27:40). According to interviewee G, trucks of resources were transferred to the disaster zone, where the respective resources were not needed at all:

“After the floods, the few deployments from the various help organizations that I saw were catastrophic. When people returned to the locality for two or three days, I cannot pull up with countless trucks containing furniture, if no houses have been cleared out yet” (G, 06:07).

Interviewee E reported a similar incident of someone “sending 120 beds to Bre- itenhagen,” although the “people of Breitenhagen did not want any old hotel beds” (E, 35:56). Besides the inefficient allocation of resources, problems with trans- portation allocation arose because the coordinators were subject to a degree of information overload (far too many messages) (E, 34:06). In Passau, interviewee C describes a cross-platform division of work where volunteers provided collection points, published them in newspapers and later on were supplied with material donations. Afterwards the moderator team arranged distribution and transport to the affected population. Within FB, and in contrast to the Lower Saxony case, a public coordination process was established in Passau: “We cooperated with Flood Help Bavaria, Passau […] and all the others who really were active there. We posted the help offers and requests everywhere” (C, 20:00). Nevertheless, the cross-platform dissemination of posts had the draw- back of potential information inconsistencies, because status updates were not always announced on every platform:

“The problem was that we had eight help requests in eight groups […]. There were conversations like ‘I already got help, everything is done’ – ‘Really? The request is still open over here’ – ‘Oh sorry, I forgot to update it’. The response between the groups or help activities did not work” (C, 55:56).

In addition to that, the relief efforts in Passau were supplemented by the student initiative, “Passau cleans up.” After using Google forms to collect volunteers’ addresses and phone numbers, a coordination network was built up:

“At the beginning we set up a mailing list. We built up a local coordination network, meaning we had […] two headquarters. Then there were (seven) coordination points with coordinators, whereby each coordinator had 3–5 scouts who walked through the streets along with fire service and THW personnel, coordinating how many volunteers and what kind of equipment was needed. This was communicated back to the coordinators, who then forwarded the information to the headquarters. From the headquarters, we sent the volunteers and their equipment to the relevant location” (F, 04:28).

Angemeldet | [email protected] Autorenexemplar Heruntergeladen am | 05.03.16 09:51 20 Marc-André Kaufhold and Christian Reuter

In order to reach a broader audience (further to their campus blog), i.e. for sharing the Google forms, clarifying conflicting or inaccurate information and reducing fearmongering, the students created a FB-page for the communication of information (F, 05:11). The page was also used to gather help offers and requests via private messages and public page posts (F, 10:11).

5.2 Roles of Digital Volunteers: The Moderator

Several social media roles (see Section 2.1) could be observed in our case studies which were connected both to the core activity of helping as well as to the gathering, amplification and moderation of information. The interviews clearly focused on the role of moderators, but also provided us with insight on their cooperation with other types of users. In the beginning, the role of the moderators, represented by interviewees A, B, C, and F, often initially encompassed simple activities but saw their efforts evolve into the more enhanced role of founding and organizing communities of affected citizens and volunteers. Interviewee C started her voluntary work by collecting information about activities in the neighborhood, but also went on to organize not only appeals for donations but also to manage collection points (C, 04:48). Likewise, interviewee B describes in detail how she mediated messages containing offers by volunteers to people affected by the floods and how she organized transports:

“Messages about hygiene articles and disinfectants were managed privately in the background. People just said. ‘I have such-and-such. Where should I send the packet?’ And then I sent them a PDF with the address” (B, 09:49).

Interviewee A started her activities on “Flood News Magdeburg” by identifying demands for help and mediating volunteers, which led to the foundation of “Flood Magdeburg – Requests for help”:

“There were plenty of requests: ‘Where can I help?’ If you registered there and visited the page 10 min later, you wouldn’t be able to find your own posting anymore. New postings were published every few seconds and you couldn’t find anything” (A, 03:45).

She thus describes her task as “providing a clear view of (relevant) information and sorting out (spam)” (A, 29:09). However, her aim was also to reduce time- consuming exposure for others by “clicking all the pages and getting the messages” (A, 28:37) and further by “updating information, answering requests and maintaining contact with on-site volunteers to get current information” (A, 05:20).

Angemeldet | [email protected] Autorenexemplar Heruntergeladen am | 05.03.16 09:51 The Self-Organization of Digital Volunteers across Social Media 21

Furthermore, according to interviewee F, the team of “Passau cleans up” exploited the style of message writing to create acceptance both for citizens and moderators, respectively:

“We paid great attention to our style of writing. We tried to formulate it in a way that made it clear help was mutual and not condescending. And of course you’ve got to try to express yourself in a way that is not too complex because there are people out there who kept on saying ‘Well it’s not just students you know but the town population, too’ because it was only students who were mentioned in the press. So you see we were a bit afraid that if we conveyed things in a complex way, we would lose volunteers; that they would pack it in because they thought we were arrogant” (F, 28:39).

Summarizing the moderators’ core activities, they entailed the mediation of volunteer activity and matched citizens’ demands with offers of help by utilizing and integrating social media functions and ICT which contributed to the process of structuring the rich information supply. A prominent role is the helper who performs real, virtual or even combined activities to support affected people. On-site helpers often commented on page postings to supplement or update published information, and used information offered by postings or Google maps to find new places requiring help or to ascer- tain which (material) resources were sought. Interviewee E, for instance, was involved in the filling of sandbags, cleanup activities (E, 03:32) and the collection of donations for affected citizens (E, 14:14). She also provided psychological assistance from home, something she was not prepared for initially (E, 2:49). Our research also identified the role of the amplifier, or retweeter in the context of Twitter. The automatic retweet services and the manual retweet offers described in Section 4.3 contribute to the amplification of specific information. Although FB does not provide direct functionality such as retweets, some FB-page teams described their task as to identify, evaluate, select and publish – and therefore amplify – relevant information. A common activity of “Flood News Magdeburg” was to disseminate structured and synthesized postings to provide a situational. Additionally, FB provides mechanisms for adjusting the relative weight of information, due to its more segmented structure and rights management: For instance, interviewee B adjusted “Flood Lower Saxony – Offer/Search” in such a way that messages were only published if a moderator approved it, thus preventing inappropriate and irrelevant information. Furthermore, successfully processed requests were deleted afterwards. Page teams typically not only performed the role of the amplifier but also that of the reporter, acting as a news channel that integrated external resources such as traditional media articles, water level services, and links to further specific or regional communities. Indeed on-site helpers, members of moderator teams (see

Angemeldet | [email protected] Autorenexemplar Heruntergeladen am | 05.03.16 09:51 22 Marc-André Kaufhold and Christian Reuter

Section 5.1) and affected citizens acted as eyewitnesses to this role, confirming the observation that citizens adopt multiple roles simultaneously (Reuter et al. 2013).

5.3 Cooperation with Official Emergency Management

In relation to official emergency management, regional differences regarding the depth of cooperation with volunteers could be observed. While interviewee A was only in touch with the fire service and THW during food distribution, B main- tained contact with local emergency service organizations via telephone, so that they could “ask incident control how many people they still needed” (B, 20:56). However, it was not usually possible to establish communication with the emergency services via FB (B, 38:45). Equally, in Passau communication was realized not through FB, but through direct contact to officials:

“Indeed, within one and a half weeks we had been invited to the ‘round table’, to the big crisis meeting which everyone important was attending. That’s where it was all coordinated, who should work together with whom” (C, 22:20).

As the administration of “Moms help” did not constitute an official emergency management organization, they joined the charitable organization “Caritas” to give their work a status (C, 22:20). The initiative “Passau cleans up” ensured contact with official emergency forces in their internal organization in a rather different way:

“Scouts walked through the streets talking to the official forces, asking how many volunteers they needed, and then passed the information ‘We need 50 people with shovels and boots’ on by word of mouth to the coordinators. They called us (the headquarters), where one member managed a spreadsheet in front of the office, equipped the volunteers and sent them to the coordinators. The scouts picked the volunteers up from there and took them along” (F, 37:20).

This respondent, though acknowledging that she did not know how to support this process using social media, strongly supported the idea that state institutions like the municipality, fire services or THW should be educated in dealing with social media. The author of “Info Page Flood Bavaria 2013” pointed out that although he gathered information from blogs and established media, he relied above all on messages from users, because such messages provided information up to 2 h in advance of other sources. According to interviewee D, some coordination difficulties resulted through reliance on official information, whereby too many volunteers sometimes went to the contact points mentioned, while there

Angemeldet | [email protected] Autorenexemplar Heruntergeladen am | 05.03.16 09:51 The Self-Organization of Digital Volunteers across Social Media 23 were too few helpers at other places (D, 02:34). Because of this suboptimal volunteer coordination, up-to-date information about required number of helpers was published on FB, which meant that indecisive helpers no longer went to these locations. In this case the incident commander could not comprehend why fewer people were arriving, because he was not aware of the current situation on FB (E, 14:14). Such circumstances indicate that the emergency services’ exploitation of social media information would improve the optimal allocation of both information and (material) resources.

6 Results III: Discussion of Challenges and Patterns

Our examination of social media use revealed some recurring patterns, extending those detected in our preliminary study (Kaufhold and Reuter 2014; Reuter et al. 2015b). Table 4 summarizes the identified challenges regarding the use of ICT

Table 4: Identified Challenges and Actions.

# Challenge identified in Actions and Implications our study

1 Clarity and – Display the community’s goals or purpose representation of – Display conventions of interaction relevant content – Apply a structure and channeling of information 2 Feedback and updates in – Ensure the timely arrival of information interaction relationships – Update outdated information 3 Organizational – Use of photos for emotional processing, solidarity and motivation identification and – “Exploit” style of message writing to maximize acceptance resilience – Design proper collaboration between involved stakeholders 4 Integration of – Model organizational embedment of ICT technologies and – Support different tasks with suitable ICT interaction types – Consider public and private ways of interaction 5 Moderation and – Use ICT to relieve moderating tasks autonomous work – Foster self-regulation of participants – Consideration of different levels of skills regarding ICT 6 Maintain quality of – Amplification of relevant information information – Provision of current information – Provision of geo-visualized information – Provision of “appealing” information

Angemeldet | [email protected] Autorenexemplar Heruntergeladen am | 05.03.16 09:51 24 Marc-André Kaufhold and Christian Reuter and, in particular, social media during the European floods of 2013, and identifies the most common actions intended to overcome these challenges that were undertaken. However, while some challenges could be addressed to a fairly satis- factory degree, most have a considerable room for improvement. These challenges will be described in the following.

6.1 Clarity and Representation of Relevant Content

The description of the FB-group, a fixed posting, and/or the page’s info section were used to display the overall goals of the community and to provide conventions on how to behave or use existing infrastructure. This seems to be a crucial aspect for optimizing the process of communication. There also seems to be a fundamental requirement to structure information flow, e.g. use of the file function to create specific documents dealing with certain topics. The annotations within the document titles furthermore indicate the need to display a status representing developments on this topic. These efforts seem to aim towards making relevant help requests more available and listenable (Hughes et al. 2014) to the users, to maintain a clear overview within the interaction platform and to support the interaction between roles, such as moderators, helpers and people in need.

6.2 Feedback and Updates in Interaction Relationships

According to our interviewees, a considerable amount of communication took place via private messages for the purposes of privacy as well. Birkbak (2012) already reported on the benefits of an architecture like FB including both public and private spaces. However, there is a high risk that timely feedback will not be received since messages received from people not in one’s friend list are filtered into the folder “other.” One major challenge is therefore to ensure the timely arrival of information. Moreover, our study showed that cross-platform collaboration requires proper information-update procedures to counter information inconsistencies and redundant resource allocation.

6.3 Organizational Identification and Resilience

To motivate affected citizens and volunteers, to generate identification with the helper’s efforts and to provide emotional support, it was common practice to upload photos showing the impact of the flood and related relief activities to

Angemeldet | [email protected] Autorenexemplar Heruntergeladen am | 05.03.16 09:51 The Self-Organization of Digital Volunteers across Social Media 25 counter it. As a consequence, comments usually contained emotional messages expressing solidarity, sympathy and identification with the helpers, showing the potential of social media to influence a community’s resilience in a positive way. Furthermore, the style of message writing seems fundamental to gain a broad acceptance among different cultural or social groups. However, the challenge on how to ensure proper collaboration among authorities, professional response and volunteer communities has to be examined in a more comprehensive and profound manner.

6.4 Integration of Technologies and Interaction Types

Regarding the organizational integration of social media, different degrees of relevance were observed. While “Passau cleans up” used social media for the most part to supplement the spread of information and gather some basic information, their central resource was the local organization with headquarters, coordinators and scouts. “Flood Lower Saxony – Offer/Search” used FB as their core platform for coordinating relief activities and information flow. In all cases, Twitter was primarily used to distribute information but not to coordinate. Also, certain degrees of privacy were implemented ranging from “Flood Magdeburg – Requests for help,” whereby people were encouraged to publicly exchange their address and contact data, to “Flood Lower Saxony – Offer/Search,” whereby sensitive data was instead exchanged by private message and managed with the support of external tools which were only visible to the creator or to a limited audience of invited people. In relation to tools, different degrees of technological integration were studied. In order to master a multitude of different tasks, “Passau cleans up” used blogs, Google docs and external web tools, “Flood News Magdeburg” integrated a Google map into their efforts and “Flood Lower Saxony” disseminated tweets alongside their core FB activities. Tools such as Google Maps and Open- StreetMap could also tighten the connection between real and virtual volunteer activities (Reuter et al. 2013) and support disaster response through voluntary community mapping (Soden and Palen 2014).

6.5 Moderation and Autonomous Work

Moderators progressed from simple activities to more complex ones (Starbird and Palen 2011). It also became apparent that the moderation of a community entails considerable overhead and is a very time-consuming process; ICT could reduce the quantity of moderation activities by fostering the self-regulation of

Angemeldet | [email protected] Autorenexemplar Heruntergeladen am | 05.03.16 09:51 26 Marc-André Kaufhold and Christian Reuter

participants. However, the interviews indicated that there were problems adapting to new technology in emergency situations, which prevented a richer integration of technology in some communities.

6.6 Maintain Quality of Information

From the perspective of information quality, several information processing activities as well as supporting functions and tools were identified. Consistent with previously conducted studies (Vieweg et al. 2010; Starbird 2013), we found that page teams and moderators sought to identify, select and amplify relevant information, and provided structured and synthesized posts of information. While geo-visualized information was a perceived aspect of improving the quality of information (Ludwig et al. 2015a), the immediacy of information was seen as a central benefit of social media in supporting the overall coordination of relief activities. In addition, the style of message writing was also seen as an aspect of quality, insofar as it influenced acceptance among of the various citizens and volunteers. For instance, emotionally supportive photos in conjunction with messages written in an appropriate style seem important to promote a positive climate among participants.

7 Summary and Conclusion

This contribution examined the actions of virtual communities in disasters by analyzing the contents of social media posts and conducting semi-structured interviews with FB-group and page founders, focusing on the use of social media and ICT in disasters with regards to organizational and information-specific characteristics. The case study of the 2013 European flood bolsters evidence for the relevance of social media during disasters for the citizens of Germany, but also shows perceived deficiencies of current emergency management practices in relation to the use of social media. In order to answer our first research question “How were different types of social media used during the European floods?” (Section 4), we found that Twitter was used as a platform for status updates, the information space of which was largely constructed through automatic retweet services. While FB-pages were mostly intended to provide an overview and to filter the vast information supply, FB-groups also coordinated a multitude of virtual and real relief activities. This confirms the findings of Hughes et al. (2014). Other tools,

Angemeldet | [email protected] Autorenexemplar Heruntergeladen am | 05.03.16 09:51 The Self-Organization of Digital Volunteers across Social Media 27 such as Google Maps, Live Blog and Google Docs and custom web tools have also been used for specific purposes often related to Facebook and Twitter. However, among the examined communities the overall importance of social media varied and different levels of privacy and technological integration were applied. To approach our second research question “How was social media used to organize volunteers?” (Section 5), we conclude that in addition to affected citizens and (digital) volunteers, some particularly dedicated volunteers adopted the role of moderator to mediate the supply and demand of help through social media and other ICT. Moderators exploited the given functionality of social media and tools they were familiar with to structure the processes of exchanging information, promoting cohesion and organizing volunteers. In contrast, considering organizational barriers of using social media (Reuter et al. 2016) the cooperation with official emergency management rather relied on local communication, meetings and telephone. To address the third research question “What are the challenges for collaboration support?” (Section 6) we outlined six challenges including (#1) clarity and representation of relevant content, (#2) feedback and updates in interaction relationships, (#3) organizational identification and resilience, (#4) integration of technologies and interaction types, (#5) moderation and autonomous work, and (#6) maintaining quality of information. Compared to related work, our approach provides several contributions: (A) While there are many studies about social media and digital volunteers in the US (as mentioned in Section 2), fewer studies investigated events in Europe. The point here is that comparative work which might draw out geographical organizational and mobilization differences ought to be valuable. (B) Looking at the 2013 European floods case, existing work focuses “only” on Twitter (Fuchs et al. 2013; Reuter and Schröter 2015), focuses on one particular city (Mildner 2013; Wittmann et al. 2015) or is rather design oriented (Sackmann et al. 2014; Backfried et al. 2015; Reuter et al. 2015b), and does not describe the overall case. In contrast, our study covered different media “beyond Twitter,” examining the event in three locations across Germany and made use of both social media and interview data. (C) We also particularly focused on the moderator as an important role in social media emergency response, something that we think has been a little underplayed.

Acknowledgments: The research project “EmerGent” was funded by a grant of the European Union (FP7 No. 608352). We would like to thank all participants of our empirical study. This article builds on a shorter, preliminary analysis (Ka ufhold and Reuter 2014).

Angemeldet | [email protected] Autorenexemplar Heruntergeladen am | 05.03.16 09:51 28 Marc-André Kaufhold and Christian Reuter

References

Backfried, G., C. Schmidt and G. Quirchmayr (2015) “Cross-Media Linking in Times of Disaster.” In: Proceedings of the Information Systems for Crisis Response and Management (ISCRAM). Kristiansand, Norway. Birkbak, A. (2012) “Crystallizations in the Blizzard: Contrasting Informal Emergency Collaboration In Facebook Groups.” In: Proceedings of the Nordic Conference on Human- Computer Interaction (NordiCHI). Copenhagen, Denmark: ACM-Press, pp. 428–437. BITKOM. (2013) Soziale Netzwerke 2013 – Eine repräsentative Untersuchung zur Nutzung sozialer Netzwerke im Internet. Berlin: BITKOM. Creswell, J. W. (2013) Research Design: Qualitative, Quantitative, and Mixed Methods Approaches. New York: Sage Publications. Fuchs, G., N. Andrienko, G. Andrienko, S. Bothe and H. Stange (2013) “Tracing the German Centennial Flood in the Stream of Tweets: First Lessons Learned.” In: SIGSPATIAL International Workshop on Crowdsourced and Volunteered Geographic Information. Orlando, USA, pp. 2–10. Heverin, T. and L. Zach (2010) “Microblogging for Crisis Communication: Examination of Twitter Use in Response to a 2009 Violent Crisis in the Seattle-Tacoma, Washington Area.” In: Proceedings of the Information Systems for Crisis Response and Management (ISCRAM). Seattle, USA. Hollan, J., E. Hutchins and D. Kirsh (2000) “Distributed cognition: Toward a New Foundation for Human-Computer Interaction Research,” ACM Transactions on Computer-Human Interaction, 7(2):174–196. Hughes, A. L. and L. Palen (2009) “Twitter Adoption and Use in Mass Convergence and Emergency Events.” In: (J. Landgren and S. Jul, eds.) Proceedings of the Information Systems for Crisis Response and Management (ISCRAM). Gothenburg, Sweden: ISCRAM. Hughes, A. L. and L. Palen (2012) “The Evolving Role of the Public Information Officer: An Examination of Social Media in Emergency Management,” Journal of Homeland Security and Emergency Management, 9(1). doi:10.1515/1547-7355.1976. Hughes, A. L. and A. H. Tapia (2015) “Social Media in Crisis: When Professional Responders Meet Digital Volunteers,” Journal of Homeland Security and Emergency Management, 12(3):679–706. Hughes, A. L., L. A. S. Denis, L. Palen and K. M. Anderson (2014) “Online Public Communi- cations by Police & Fire Services during the 2012 Hurricane Sandy.” In: Proceedings of the Conference on Human Factors in Computing Systems (CHI). Toronto, Canada: ACM-Press, pp. 1505–1514. Karsten, A. (2013) “Interoperabilität und Agilität: Sachstand und Herausforderung.” In: Interorganisationale Zusammenarbeit im Krisenmanagement. Siegen, Germany: University of Siegen. Kaufhold, M.-A. and C. Reuter (2014) “Vernetzte Selbsthilfe in Sozialen Medien am Beispiel des Hochwassers 2013/Linked Self-Help in Social Media using the example of the Floods 2013 in Germany,” i-com, 13(1):20–28. Liu, S., L. Palen and J. Sutton (2008) “In Search of the Bigger Picture: The Emergent Role of On-Line Photo Sharing in Times of Disaster.” In: Proceedings of the Information Systems for Crisis Response and Management (ISCRAM). Washington, DC, USA. Ludwig, T., C. Reuter and V. Pipek (2015a) “Social Haystack: Dynamic Quality Assessment of Citizen-Generated Content in Social Media during Emergencies,” Transactions on

Angemeldet | [email protected] Autorenexemplar Heruntergeladen am | 05.03.16 09:51 The Self-Organization of Digital Volunteers across Social Media 29

Human Computer Interaction (ToCHI), 21(4). Retrieved from http://dl.acm.org/citation. cfm?id=2798442.2749461. Ludwig, T., C. Reuter, T. Siebigteroth and V. Pipek (2015b) “CrowdMonitor: Mobile Crowd Sensing for Assessing Physical and Digital Activities of Citizens during Emergencies.” In: Proceedings of the Conference on Human Factors in Computing Systems (CHI). Seoul, Korea: ACM Press. Meier, P. (2012) “Crisis Mapping in Action: How Open Source Software and Global Volunteer Networks Are Changing the World, One Map at a Time,” Journal of Map & Geography Libraries, 8(2):89–100. Mildner, S. (2013) “Bürgerbeteiligung beim Hochwasserkampf – Chancen und Risiken einer kollaborativen Internetplattform zur Koordination der Gefahrenabwehr.” In: (T. Köhler and N. Kahnwald, eds.) Online Communities: Enterprise Networks, Open Education and Global Communication: 16. Workshop GeNeMe ’13 Gemeinschaften in Neuen Medien. Dresden, Germany: TUDpress, pp. 13–21. Palen, L. and S. B. Liu (2007) “Citizen Communications in Crisis: Anticipating a Future of ICT-Supported Public Participation.” In: Proceedings of the Conference on Human Factors in Computing Systems (CHI). San Jose, USA: ACM Press. Perng, S.-Y., M. Büscher, L. Wood, R. Halvorsrud, M. Stiso, L. Ramirez and A. Al-Akkad (2012) “Peripheral Response: Microblogging during the 22/7/2011 Norway Attacks.” In: (L. Rothkrantz, J. Ristvej and Z. Franco, eds.) Proceedings of the Information Systems for Crisis Response and Management (ISCRAM). Vancouver, Canada: ISCRAM, pp. 1–11. Qu, Y., P. F. Wu and X. Wang (2009) “Online Community Response to Major Disaster: A Study of Tianya Forum in the 2008 Sichuan Earthquake.” In: Proceedings of the Hawaii International Conference on System Sciences (HICSS). Waikoloa, USA. Qu, Y., C. Huang, P. Zhang and J. Zhang (2011) “Microblogging after a Major Disaster in China: A Case Study of the 2010 Yushu Earthquake.” In: Proceedings of the Conference on Computer Supported Cooperative Work (CSCW). Hangzhou, China: ACM, pp. 25–34. Reuter, C. and J. Schröter (2015) “Microblogging during the European Floods 2013: What Twitter May Contribute in German Emergencies,” International Journal of Information Systems for Crisis Response and Management (IJISCRAM), 7(1):22–41. Reuter, C., A. Marx and V. Pipek (2012) “Crisis Management 2.0: Towards a Systematization of Social Software Use in Crisis Situations,” International Journal of Information Systems for Crisis Response and Management (IJISCRAM), 4(1):1–16. Reuter, C., O. Heger and V. Pipek (2013) “Combining Real and Virtual Volunteers through Social Media.” In: (T. Comes, F. Fiedrich, S. Fortier, J.Geldermann and T. Müller, eds.) Proceedings of the Information Systems for Crisis Response and Management (ISCRAM). Baden-Baden, Germany: ISCRAM, pp. 1–10. Reuter, C., T. Ludwig, T. Friberg, S. Pratzler-Wanczura and A. Gizikis (2015a) “Social Media and Emergency Services? Interview Study on Current and Potential Use in 7 European Countries,” International Journal of Information Systems for Crisis Response and Management (IJISCRAM), 7(2):36–58. Reuter, C., T. Ludwig, M. -A. Kaufhold and V. Pipek (2015b) “XHELP: Design of a Cross-Platform Social-Media Application to Support Volunteer Moderators in Disasters.” In: Proceedings of the Conference on Human Factors in Computing Systems (CHI). Seoul, Korea: ACM Press. Reuter, C., T. Ludwig, M. -A. Kaufhold and T. Spielhofer (2016) “Emergency Services Attitudes towards Social Media: A Quantitative and Qualitative Survey across Europe,” International Journal on Human-Computer Studies (IJHCS), in press. Rizza, C., Â. G. Pereira and P. Curvelo (2014) ““Do-it-Yourself Justice”: Considerations of Social Media use in a Crisis Situation: The Case of the 2011 Vancouver Riots,” International

Angemeldet | [email protected] Autorenexemplar Heruntergeladen am | 05.03.16 09:51 30 Marc-André Kaufhold and Christian Reuter

Journal of Information Systems for Crisis Response and Management (IJISCRAM), 6(4):42–59. Sackmann, S., M. Hofmann and H. J. Betke (2014) “Organizing On-Site Volunteers: An App-Based Approach.” In: Proceedings of the 2014 Ninth International Conference on Availability, Reliability and Security. Washington, DC, USA: IEEE Computer Society, pp. 438–439. Soden, R. and L. Palen (2014) “From Crowdsourced Mapping to Community Mapping: The Post-Earthquake Work of OpenStreetMap Haiti.” In: COOP 2014 – Proceedings of the 11th International Conference on the Design of Cooperative Systems, 27–30 May 2014, Nice (France). pp. 311–326. doi:10.1007/978-3-319-06498-7. St. Denis, L. A., K. M. Anderson and L. Palen (2014) “Mastering Social Media: An Analysis of Jefferson County’s Communications during the 2013 Colorado Floods.” In: 11th International ISCRAM Conference. pp. 737–746. Stake, R. E. (1995) The Art of Case Study Research. New York: Sage Publications. Stallings, R. A. and E. L. Quarantelli (1985) “Emergent Citizen Groups and Emergency Management,” Public Administration Review, 45(Special Issue):93–100. Starbird, K. (2013) “Delivering Patients to Sacré Coeur: Collective Intelligence in Digital Volunteer Communities.” In: Proceedings of the Conference on Human Factors in Computing Systems (CHI). Paris, France: ACM-Press, pp. 801–810. Starbird, K. and L. Palen (2010) “Pass It On?: Retweeting in Mass Emergency.” In: Proceedings of the Information Systems for Crisis Response and Management (ISCRAM). Seattle, USA, pp. 1–10. Starbird, K. and L. Palen (2011) “Voluntweeters: Self-Organizing by Digital Volunteers in Times of Crisis.” In: Proceedings of the Conference on Human Factors in Computing Systems (CHI). Vancouver, Canada: ACM-Press. Starbird, K. and L. Palen (2012) “(How) Will the Revolution be Retweeted? Information Diffusion and the 2011 Egyptian Uprising.” In: Proceedings of the Conference on Computer Supported Cooperative Work (CSCW). Seattle, USA: ACM-Press, pp. 7–16. Strauss, A. L. and J. Corbin (1998) Basics of Qualitative Research: Techniques and Procedures for Developing Grounded Theory. New York: Sage Publications. Sutton, J. (2010) “Twittering Tennessee: Distributed networks and Collaboration Following a Technological Disaster.” In: (S. French, B. Tomaszewski and C. Zobel, eds.) Proceedings of the Information Systems for Crisis Response and Management (ISCRAM). Seattle, USA: ISCRAM. Vieweg, S., A. L. Hughes, K. Starbird and L. Palen (2010) “Microblogging During Two Natural Hazards Events: What Twitter May Contribute to Situational Awareness.” In: Proceedings of the Conference on Human Factors in Computing Systems (CHI). Atlanta, USA: ACM-Press, pp. 1079–1088. White, J. I., L. Palen and K. M. Anderson (2014) “Digital Mobilization in Disaster Response: The Work & Self – Organization of On-Line Pet Advocates in Response to Hurricane Sandy.” In: Proceedings of the Conference on Computer Supported Cooperative Work (CSCW). Baltimore, USA: ACM. Wittmann, S., M. Jurisch and H. Krcmar (2015) “Managing Network Based Governance Structures in Disasters: The Case of the Passau Flood in 2013,” Journal of Homeland Security and Emergency Management, 12(3):529–569. Yin, R. K. (2014) Case Study Research: Design and Methods. New York: SAGE Pub.

Angemeldet | [email protected] Autorenexemplar Heruntergeladen am | 05.03.16 09:51 TFS-18618; No of Pages 13 Technological Forecasting & Social Change xxx (2016) xxx–xxx

Contents lists available at ScienceDirect

Technological Forecasting & Social Change

Towards social resilience: A quantitative and qualitative survey on citizens' perception of social media in emergencies in Europe

Christian Reuter a,⁎, Thomas Spielhofer b a University of Siegen, Institute for Information Systems, Kohlbettstr. 15, 57072 Siegen, Germany b The Tavistock Institute of Human Relations, 30 Tabernacle Street, London, United Kingdom article info abstract

Article history: Social media is increasingly being used during emergencies. Most available studies are focused on how citizens Received 9 March 2016 and/or authorities use these technologies in concrete events. However, larger quantitative studies with signifi- Received in revised form 8 July 2016 cant results on attitudes, needs and future plans of citizens in such events are not available - especially such of Accepted 27 July 2016 a comparative nature related to emergency services. As part of the EU project ‘EmerGent’ this article presents Available online xxxx the findings of a survey of 1034 citizens across 30 European countries conducted between February and June Keywords: 2015 to explore citizens' attitudes towards the use of social media for private purposes and in emergency situa- fl fi Citizen tions. The article brie ycomparesthese ndings with a second survey conducted with 761 emergency service Europe staff across 32 European countries from September to December 2014. The aim of the overall study is to discuss Social media citizens' attitudes towards social media in emergencies in order to derive challenges and opportunities for social Survey resilience. Quantitative © 2016 Elsevier Inc. All rights reserved. Qualitative

1. Introduction public participation would become regarded as both normal and valuable. Fewer studies covering the situation in Europe exist (Reuter A well-known definition sees social media as a “group of Internet- et al., 2012). based applications that build on the ideological and technological In recent years, the use of social media has increased substantially foundations of Web 2.0, and that allows the creation and exchange of and at the same time the nature of that use has shifted towards a user-generated content” (Kaplan and Haenlein, 2010). In this context more collaborative model. Based on the broader definition of resilience user-generated content refers to “the sum of all ways in which peo- as the “ability of a system, community or society exposed to hazards plemakeuseofsocialmedia” and describes “the various forms of to resist, absorb, accommodate to and recover from the effects of a media content that are publicly available and created by end-users” hazard in a timely and efficient manner, including through the preser- (Kaplan and Haenlein, 2010). Currently the most common types of vation and restoration of its essential basic structures and functions” social media are Facebook with about 1.36 billion active users (United Nations, 2009), more specific terms with overlapping meanings monthly, and the microblogging platform, Twitter, counting approxi- emerged: cooperative resilience (Reuter et al., 2016a) as the ability to mately 284 million active users monthly. YouTube (1 billion), WhatsApp overcome crises of cooperation with the help of adaptability to modified (600 million), Google + (343 million), Instagram (300 million) and realities by means of cooperation technology, or social resilience as the Tumblr (230 million) (Statista, 2015) also all have N100 million active “capacity of social groups and communities to recover from, or respond users monthly and are widely distributed. positively to, crises” (Maguire and Hagan, 2007). Social media can be For at least one and a half decades social media has been used in understood as a key element in the accomplishing of social resilience. emergencies: after 9/11, by way of example, wikis, created by ordinary However, although we have a developing body of research which anal- citizens, were used to collect information about missing people (Palen yses use case with regard to the use of social media during emergencies, and Liu, 2007). Subsequently, analysis of social media in disaster there is less work which deals with attitudinal factors, especially with management, mainly in the USA, has become commonplace. Early stud- regard to the attitudes of citizens in such contexts. ies included Murphy and Jennex (2006) who, following hurricane In this article, then, we seek to explore the attitudes of European Katrina, looked at the use of PeopleFinder and ShelterFinder, and citizens towards the use of social media in emergency situations. Palen and Liu (2007), who anticipated a future where ICT-supported Based on the analysis of related work (Section 2), we describe the methodology of our survey (Section 3) and present its quantitative as fi ⁎ Corresponding author. well as qualitative results (Section 4). Following this, the ndings are E-mail address: [email protected] (C. Reuter). compared to a previously published survey on emergency service staff

The use of social media in emergencies has become a research agen- 2.2. Surveys on authorities' perception of social media da of its own to such an extent that the term, “Crisis informatics” is now commonplace. It “views emergency response as an expanded social Reuter et al. (2016b) analysed the state of the art and found that pre- system where information is disseminated within and between official vious research has shown that “volunteered individual reports, espe- and public channels and entities. Crisis informatics wrestles with cially pictures, are of particular value” to emergency services (see also methodological concerns as it strives to develop new theory and sup- Ludwig et al., 2015). However, at the same time, other studies have port sociologically informed development of both ICT and policy” shown that not all emergency responders make use of such data during (Palen et al., 2009, p. 3). This trend was predicted some years ago: disasters given the difficulties of receiving and filtering particularly “the role held by members of the public in disasters […] is becoming large amounts of data in emergencies (Hughes and Palen, 2012). more visible, active, and in possession of greater reach than ever seen There are a few quantitative studies which provide evidence on this before” (Palen and Liu, 2007). That is, the burgeoning research interest question, although most are from North America: has been allied with a greater acceptance of social media use by those Firstly, this includes a comparative study published by the American directly involved. National Emergency Management Association (NEMA) containing the In the recent past, a number of studies have been published on social results of a survey conducted in 2012 among members of emergency media in emergencies. Various special issues in international journals services from all 50 Federal States of the US (San et al., 2013). Although such as the Transactions on Computer-Human Interaction (Hiltz et al., the respondents indicated a positive attitude towards social media in 2011), Computer Supported Cooperative Work (Pipek et al., 2014)or general and valued its suitability for information dissemination, 75% of International Journal of Information Systems for Crisis Response and them highlighted the requirement of verifying citizen-generated con- Management (Reuter, 2015) as well as tracks at various conferences, tent, and questioned its credibility. The study also revealed that 85% of such as Information Systems for Crisis Response and Management deal US authorities already use social media. with this topic. Many studies focus on specific events such as the 2011 Secondly, a survey of 241 US emergency managers at the county London riots (Denef et al., 2013), the 2012 hurricane Sandy (Hughes level in 2014 shows that only about half of these agencies use social et al., 2014) or the 2013 European floods (Reuter et al., 2015a). These media (Plotnick et al., 2015). Most of them do not have any formal pol- studies demonstrate the specific ways in which social media have icies to guide their use. Of those who do have formal policies, about one responded to various crises. quarter actually forbid the use of social media. The range of different emergency situations, and responses to them, Thirdly, with over 500 participants, the annual study of the Interna- has also produced attempts to categorize the use of social media: Reuter tional Association of Chiefs of Police (IACP) about law enforcement's use et al. (2012) distinguish between different scenarios of social media use of social media to report on “the current state of practice and the issues in emergencies. They differentiate between two distinct groups, citizens agencies are facing in regard to social media” (International Association (C) and authorities (A) (such as emergency services) and describe the of Chiefs of Police, 2015). This found that, comparing the first resulting information flows in social media such as from authorities to (International Association of Chiefs of Police, 2010) and last surveys, citizens (A2C), from citizen to authorities (C2A) as well as the coordina- the agencies' use of social media increased from 81% (77% Facebook, tion of (voluntary) activities among citizens (C2C). Research regarding 37% Twitter, 16% YouTube) to 96% (94% Facebook, 71% Twitter, 40% types of users active on social media began by identifying individual YouTube) and the number of social media policies increased from 35% roles and proceeded with the development of role typologies. Hughes to 78%. and Palen (2009) initially identified information brokers who collect in- Finally, Reuter et al. (2016a,b) describe their findings of a survey formation from different sources to help affected citizens. For Starbird conducted as part of the EU project ‘EmerGent’ with 761 emergency ser- and Palen (2011), the second step was to recognize the actions of remote vice staff across 32 European countries from September to December operators as digital volunteers who progress from simple internet-based 2014. They found that the majority of emergency services had positive activities like retweeting or translating tweets to more complex ones, attitudes towards social media. Social media is more used to share infor- e.g. verifying or routing information. To further differentiate potential mation (44%) than to receive messages (19%). An increase in use is ex- user roles, Reuter et al. (2013) suggest distinguishing between activities pected (74%), even more for organizations already using it. However in the ‘real’ world as opposed to the ‘virtual’ world: real emergent groups there is a huge gap between rhetoric and reality; thus, while 66% of (Stallings and Quarantelli, 1985), whose involvement usually takes the emergency service staff indicated social media can be used to obtain form of neighbourly help and work on-site, and virtual digital volunteers an overview of the situation and to raise situational awareness, in fact (Starbird and Palen, 2011), who originate from the internet and work only 23% have often or sometimes used social media sites for this mainly online. Another study suggests that (real) volunteer groups in purpose. emergencies will in the future be challenged to mature and improve according to these enhanced possibilities, so that “professional responders 2.3. Surveys on citizens' perception on social media will begin to rely on data and products produced by digital volunteers” (Hughes and Tapia, 2015). Based on a timeline and qualitative analysis Very few quantitative studies have been conducted where citizens of information and help activities during the 2011 Super Outbreak, have been asked about their perception of using social media in

Please cite this article as: Reuter, C., Spielhofer, T., Towards social resilience: A quantitative and qualitative survey on citizens' perception ofsocial media in emergencies in Europe, Technol. Forecast. Soc. Change (2016), http://dx.doi.org/10.1016/j.techfore.2016.07.038 C. Reuter, T. Spielhofer / Technological Forecasting & Social Change xxx (2016) xxx–xxx 3 emergencies. Three in particular, however, are worth mentioning. This 3.1. Survey design includes a comparative study with over 1000 participants conducted by the Canadian Red Cross (2012), which aimed to identify to what ex- The survey aimed to identify the attitudes of citizens towards the use tent Canadian citizens use social media and mobile devices in crisis of social media and was conducted as part of the EU funded project communication and what they expect from the emergency services “EmerGent”. It was designed with the aim of collecting a mixture of both currently and in future. This study emphasizes the requirement quantitative and qualitative evidence. In some parts, we aimed to gain of trained social media personnel and pointed to the credibility issues statistical results, in some others we were interested in the reasons for of citizen-generated content. It also, however, shows the benefits of re- answers. Therefore, quantitative as well as qualitative methods were assurance for citizens, providing situational information and monitor- used. The survey consisted of two parts (see Appendix A for details), ing. Social media were seen as a support for existing channels, but not as follows: as a replacement for them. It is noteworthy that the Canadian Red Cross employs “trusted volunteers” to support official response via so- ▪ Part I: Demographic details of survey participants (age, gender, cial media. country of origin, role, type of organization) to explore any differ- Secondly, the American Red Cross (2012) also studied citizens' use of ences in responses depending on the characteristics of participants. social media during emergencies, with 1.017 online and 1.018 tele- ▪ Part II: Attitudes towards social media especially in emergencies – a phone survey respondents. According to the study, 12% of the gener- combination of closed questions (using Likert scales (Likert, 1932), al public, and respectively 22% of high school graduates, have used asking participants to rate on a scale of 1 to 5 how much they social media to share or obtain information during emergencies agree with a series of statements) and open-ended questions. and disasters or in severe weather conditions. Users were most likely to seek information about weather, traffic, damage caused and information on how other people were coping. Beyond that, users shared We designed the survey based on a strategy aimed at triangulation. not only weather information, safety reassurances and their feelings This methodological triangulation involved a combination of questions about the emergency but also their location, and eyewitness infor- that focus on more qualitative aspects of citizens' intentions towards so- mation. In terms of trustworthiness, friends, family, news media cial media and their usage before, during and after an emergency at (or reporters) and local emergency officials were the most trusted micro-level as well as more quantitative aspects to obtain a comprehen- sources, while unknown people in the general vicinity of the emer- sive picture of citizens' attitudes towards social media within emergen- gency were the least trusted. cies at a macro-level. Thirdly, Flizikowski et al. (2014) present a survey within Europe, In the beginning of February 2015, project partners sent out a link to conducted among citizens (317 respondents) and emergency services the online survey in English, Polish, Italian, German and Slovenian to (130 respondents). The study focuses on the identification of user friends, colleagues, professional and social contacts as well as via their needs concerning crisis management with the support of social media own social media channels and websites (snowball sample). This and mobile devices. The main goal of the study was to identify the pos- means that the sample of citizens responding to this survey cannot be sibilities and challenges of social media integration into crisis response assumed to be fully representative of citizens across Europe. management. Generally, the participants had a positive attitude towards social media. During the study, both citizens and emergency ser- 3.2. Characteristics of survey participants vices identified the same challenges, such as a lack of knowledge, personnel issues, uniform terms of use, credibility of citizen-generated The survey responses of 1.034 citizens (including 195 working or content, and accessibility for older generations. volunteering for an emergency service – excluded from the main analysis) were received from citizens across 30 countries, with the largest 2.4. Research gap number of respondents coming from Poland (306), Slovenia (169), Germany (164), the United Kingdom (146), Italy (72), Greece (43) Even though we know that many citizens use social media in emer- and Norway (39) (Q2) (Fig. 1). It has to be noted that the sample is gencies, there is very little evidence exploring what proportion and not representative for each country or for the whole of Europe. Respon- types of citizens currently do so. Most existing quantitative studies dents included roughly equal proportions of women and men (Q3), and focus on emergency services only or study the attitudes of citizens in a broad selection of citizens from different age groups – although the North America only (American Red Cross, 2012; Canadian Red Cross, largest proportion (33%) were aged 21–29 years old – only 4% were 2012). We still know relatively little about the situation in Europe, aged 60 or older (Q4) (Fig. 2). Around one-in-five (19%) of survey par- which might be very different. Flizikowski et al. (2014) did focus on ticipants were working or volunteering for an emergency service (Q5) Europe although the study is relatively small scale (based on 317 – these were excluded from the main findings reported on in this sum- respondents), and is primarily dependent on open-ended qualitative mary report as they were significantly more likely to use social media questions. Our study sought, therefore, to add to existing knowledge than other citizens and to express positive views about its use during by providing a combined analysis of qualitative and quantitative survey emergencies. questions with 1.034 respondents across Europe both in respect of the present situation and in terms of perspectives for the future. In particular, we intended to shed light on how social media can be used to foster social resilience to deal with disasters.

3. Methodology

This section presents our methodology. It has been adopted from a related study (Reuter et al., 2016b), which focused on the attitudes of emergency service personnel only. We ﬁrst present the survey design (Section 3.1), including questions, technical realization and channels of distribution. Then we present a characterization of our participants (Section 3.2), followed by a description of our quantitative (Section 3.3)andqualitative(Section 3.4) analysis design. Fig. 1. Countries

was added which contained only the qualitative results including the response identifier and original language identifier. As the survey had been distributed all over Europe, responses made by citizens were in different languages and there was a need for translation. For each question, two columns for translation and categorization were added. Thereafter each response was read manually and translated into English, if required. The translation was performed by native speakers of the respective languages. Where possible, translations were performed by translation services, such as Google Translate and supplemented with dictionaries, if single words could not be translated automatically or the translations needed manual adjustment for better Fig. 2. Age. intelligibility. The need for translating the results might be highlighted as a limitation of the study, however we aimed to ask people from dif- 3.3. Quantitative analysis ferent countries in their language, to lower the barrier to participate in our study. For the quantitative analysis, the survey data was extracted and To be able to use the grounded theory-oriented method, the open- analysed using SPSS, a software package for analysing quantitative ended questions were coded openly and participants' statements were data (IBM, 2014). Furthermore, Excel was used for qualitative coding. divided into categories. Each response was then assigned one or multi- The analysis consisted of three key steps: ple categories to achieve a quick overview of the interesting and relevant topics. The previously acquired knowledge from the literature 1. Exploring basic frequencies for each question and using cross- review and quantitative analysis was used to increase theoretical sensi- fi tabulations to explore any signi cant differences across different tivity. In the next section we only present those responses that show types of respondents. identifiably positive or negative perceptions of social media and its 2. Factor-analysis of the 12 Likert scale questions (see above) on partici- use by emergency services from an organizational as well as individual pants' attitudes towards social media. To measure respondents' atti- perspective. Each quotation is referenced with the participants' re- tudes towards the use of social media for both private and sponse identifier (e.g. EN146). organizational purposes, we used the statistical technique of factor analysis. Factor analysis is a technique used in research to identify 4. Empirical results groups or clusters of variables, which, taken together, represent an underlying construct or variable of interest in the study (Field, In the following sections we present the results of our survey. First 2009). The analysis identified two factors: the first measured partic- we present results regarding personal attitudes towards the use of so- ipants' use of social media in general, while the second measured cial media (Section 4.1). We then elaborate the results on searching in- their attitudes towards using social media in emergency situations. formation (Section 4.2), sharing information (Section 4.3), expectations Both of these factors had high reliability with Cronbach alpha scores, from emergency services (Section 4.4), as well as open question re- used to indicate reliability of the scale, of α = 0.725 and α = 0.774 sponses on what would encourage increased social media use in future (Cronbach, 1951) respectively. (Section 4.5). Finally we present results relating to participants' aware- 3. One-way Analysis of Variance - ANOVA (Field, 2009) was then used to ness of social media safety services (Section 4.6) and on their use of measure any significant differences between the types of respondent smartphone apps (4.7). in relation to these two factors. 4.1. Use of social media (Q8–10) 3.4. Qualitative analysis Initially, participants were asked about their use of social media in The analysis of our free-text survey questions was based on the in- general (Fig. 3, Q8). The results show that most participants use ductive approach of grounded theory (Strauss, 1987). We used open Facebook on a regular basis (73% answered “often” or “sometimes”). coding associated with grounded theory to derive categories from the Many participants also use YouTube at least sometimes (69%). However, more qualitative free-text answers by careful reading and aggregating the majority also stated that they never use Twitter (62%) or Instagram of categories. (73%). The first step was to extract the entire dataset from the survey Most participants agreed with the statement that they use social platform into an Excel (*.xls) output file. Accordingly, a second sheet media very often in their private lives (63%) and that they have many

Fig. 3. Current use of social media (Q8).

Fig. 4. Attitude towards social media (Q9). friends using social media to keep in touch (73%). 60% of the partic- 4.2. Searching information (Q14–16 + 21) ipants also stated that they regularly post messages on social media. While 73% of the participants thought, that it is important When asked whether they have used social media to find out infor- for emergency services to use social media, 38% remarked that they mation in an emergency, 43% of the people said they had used social would not trust messages on social media, apart from those from of- media for this purpose, while a similar proportion (49%) reported that ficial sources. Moreover, about one third (30%) answered that emer- they had not (Fig. 6). gency services should not trust information on social media (Fig. 4, Of those who had looked for information on social media relating Q9). to an emergency, most participants reported that they had looked for When asked, which communication channels they have used to get information about the weather (78%), road or traffic conditions information about an emergency, most participants indicated that they (70%) or damage caused by an event (63%). Many participants also had used TV (86%) and online news (80%), followed by local radio (54%) reported that they had used social media in the past to look for eye- and social media (42%). Furthermore, a smaller proportion of partici- witness videos or photographs (60%). Some also used it to find out pants said that they had used online sites (31%) and mobile apps and the location or status of friends or family (41%) and information text messages (22%). Only few people specified that they used other about how others were coping with the disaster (38%). Only a third channels or none of them (Fig. 5). (33%), said that they had looked for information about “what to do There are significant differences in the general use of social media to keep yourself safe” (Fig. 7). among different groups of citizens – young people (F (5, 813) = More than half (58%) of participants indicated that it was either 11.530, p b 0.001) and women (F (1, 813) = 26.527, p b 0.001) are far quite or very likely that they would use social media in the future to more likely to use it than other citizens. Overall, 13% of citizens currently look for information. In contrast, just under a quarter (23%) thought it do not use a smartphone – this rises to 29% of those aged 50 or above. was unlikely they would do so (Fig. 8). The level of social media use decreased with the age of participants in We furthermore asked participants questions about the possible rea- an almost linear fashion. Women displayed a significantly more positive sons for using social media as an information source. As can be seen in attitude towards the use of social media during emergencies – similarly, Figs. 9, 54% of citizens thought that information provided on social citizens with children under the age of 18 had a more positive attitude media sites during emergencies is more accessible than information towards this. provided via more traditional media channels such as TV, radio or

Fig. 5. Current communication channels in use (Q10).

Fig. 6. Current use of social media for information gathering in emergency situations (Q14).

Fig. 7. Current use of social media for information gathering in emergency situations (Q15).

Fig. 8. Future use of social media for information gathering in emergency situations (Q16). media websites. Similarly, 77% indicated that information provided on the proportion of those who had never shared information was consid- social media is made available faster during emergencies than via tradi- erably higher (67%). However, 27% stated that they had shared informational media channels. However, only 13% contended that information tion about emergencies in social media – this is likely to include sharing provided on social media is more accurate than information provided information with other citizens as well as with emergency services or via traditional media channels – in contrast, almost half (44%) disagreed authorities (C2A) (Fig. 10, Q17). Women were signiﬁcantly more likely with this statement. (χ2(1) = 17.926, p b 0.001) to have done so (33%) than men (only 20% had done so). 4.3. Information sharing (Q17–19) Of those who had shared information on social media, this was most likely to have involved information on weather conditions or While, as indicated above, about 50% of participants had never warnings (66%), road or trafﬁc conditions (64%) or uploaded eyewit- looked for information on social media as a result of an emergency, ness photographs (53%) (Fig. 10). In contrast, only 22% had shared an

Fig. 9. Attitude towards social media as information source (Q21).

Fig. 10. Current use of social media for sharing information regarding emergency situations (Q17).

Fig. 11. Current use of social media for sharing information regarding emergency situations (Q18).

Fig. 12. Future use of social media for sharing information regarding emergency situations (Q19). eyewitness video on social media. A complete overview is given in • The need for a clearer purpose for sharing information, in particular Fig. 11. that emergency services would definitely make use of such informa- Asked whether they might be inclined to use social media in future tion (185 responses fell into this category): “If I had something to tell to share information with others, 48% of the participants said that them that I thought was relevant.” (EN05); “To know in what ways it they thought it was likely they would do so; in contrast, as can be might be helpful. To be more aware of how the emergency services seen in Figs. 12, 28% indicated it was unlikely. would use this information” (EN10). • More confidence in the quality and data security of information shared on social media (157 responses): “Impossible to say, there's 4.4. Expectations from emergency services (Q22) so much guff on social media sites that you'd never could be 100% that a disaster is truly a disaster as opposed to ‘banter’.” (EN112); The survey also included a series of questions exploring citizens' “Including organizations (firefighters, police, civil protection) into expectations of how emergency response organizations would or social networks with beneficial up to date information, regular pub- should react to a citizen posting a request for help or information lishing on their web sites, also during peace times, when there are no on their social media site. It showed that 41% of citizens would natural or other accidents” (SL95, translated) expect a response within an hour if they posted such a request, • The provision of improved or more user-friendly applications to while 69% agreed that emergency services should regularly monitor share and access such information (64 responses): “Something their social media sites to be able to respond promptly to such a re- very easy to use and already integrated in the apps I currently use” quest. In contrast, 56% of participants thought that emergency ser- (EN08); “Validating Information. Performant selecting algorithms. vices were too busy during an emergency to monitor social media Geo referenced data supply.” (EN146) (Fig. 13). • Better guidelines and encouragement from authorities on the best ways of sharing information during emergencies (27 re- 4.5. Encouraging citizens to use social media more widely in future (Q20) sponses): “Didn't know where to look for information and advice. Felt the authorities should at least have had advice on the front of The survey contained an open question, asking participants what their websites.” (EN34); “Change in form, now social media is used would make them more likely to use social media to share informa- for any other purpose. The state would have to announce a new tion with others in future to which 485 participants provided at form of transmission of such information.” (PL71, translated). Un- least one response. The answers indicated that the main ways of surprisingly there are some guidelines for emergencies available, encouraging such more widespread use of social media in emergen- whichhaveemergedfromtherecentEUprojectsISAR+(Simão cies included: et al., 2015)andCOSMIC(Helsloot et al., 2015). However, there

Fig. 13. Perceived social media integration of emergency services (Q22).

is little or no evidence to suggest that awareness of such guide- Twitter on a regular basis – 32% of citizen who say they use Twitter lines is particularly high among citizens across Europe. ‘often’ are aware of this service. This contrasts with only 4% of regular Facebook users (that use it ‘often’) who say they are aware of Facebook Safety Checks. Other participants (98 responses) made comments which did not fall into speciﬁc categories. This included answers like “Nothing” (45 responses), or “I don't use social media (for this purpose)” (22 responses) 4.7. Use of emergency smartphone apps (Q11–13) as well as “I don't know” (21 responses). The majority of participants (71%) reported that they had never 4.6. Awareness about social media safety services (Q23) downloaded a smartphone app for emergencies or disasters, while 22% said that they had done so (Q11). 208 people answered the open There is generally low awareness among citizens of existing social question what kind of apps they had downloaded (Q12). The most pop- media safety services provided on Twitter and Facebook – thus, only ular apps were Warning apps (49 mentions), followed by Weather apps 6% of citizens said they were ‘very aware’ of Twitter Alerts, while only (28 mentions) and First Aid apps (16 mentions). Moreover, several peo- 3% were very aware of Facebook Safety Checks (Fig. 14). However, ple named Emergency Call apps (14 mentions), News apps (11 men- awareness of Twitter Alerts is considerably higher among those using tions) and Earthquake apps (10 mentions) (Table 1). Apropos of the

Fig. 14. Knowledge about social media services (Q23).

Table 1 App-categories and the frequencies of mentions (Q12).

App-categories Mentions App-categories Mentions App-categories Mentions

Warning app 49 Trafﬁc (Jam) app 8 Twitter app 4 Weather app 28 Location app 5 Safety app 4 First aid app 16 Red cross app 5 Breakdown app 3 Emergency call app 14 Maps app 5 Fire app 3 News app 11 Lifesaver app 5 Flashlight app 3 Earthquake app 10 Hazardous material app 4 Others 15

Please cite this article as: Reuter, C., Spielhofer, T., Towards social resilience: A quantitative and qualitative survey on citizens' perception ofsocial media in emergencies in Europe, Technol. Forecast. Soc. Change (2016), http://dx.doi.org/10.1016/j.techfore.2016.07.038 C. Reuter, T. Spielhofer / Technological Forecasting & Social Change xxx (2016) xxx–xxx 9 separate apps in the different countries, the app Regionalny System younger citizens and women are significantly more likely to use social Ostrzegania (RSO), which is named 21 times by Polish people, is eye- media both to look for and share information, while men and those catching. Furthermore, German survey participants mentioned Katwarn aged 50 or above are significantly less likely to use social media for 13 times. Both apps are warning apps developed by the government and this purpose (Fig. 16: Facts 2 and 4). As the results of the survey showed, are open to the whole population. No other apps were named by more this was exacerbated by the fact that almost a third (29%) of those aged than five respondents. 50 or above do not use a smartphone which is a necessary prerequisite While only a relatively small proportion of participants had previ- for using social media while not at home. The implications of this is that ously downloaded an app, most (60%) thought that it was either very while social media use is widespread and increasing, some groups are in or quite likely that they would download an app to share information danger of being excluded from any support, advice or instructions pro- with, or receive information from, emergency services in an appropriate vided via social media before, during or after emergencies from emer- situation. Only 21% stated that it was not likely they would do so (Q13) gency services or other citizens. This could mean that those most (Fig. 15). vulnerable in an emergency – older or disabled citizens – may be least likely to benefit from an increased use of social media by emergency 5. Discussion and conclusion services. The study revealed that citizens' awareness of Twitter Alerts and Recently “the role held by members of the public in disasters […]is Facebook Safety Checks was generally low – about 56% have never becoming more visible, active, and in possession of greater reach than heard of at least one of them (Fig. 16: Fact 5). Sixty percent of partici- ever seen before” (Palen and Liu, 2007). Social media has enabled pants were not at all aware of Twitter Alerts, and 68% were not at all many of these possibilities and may foster social resilience. Many aware of Facebook Safety Checks. Likewise, 71% have never downloaded studies are available that cover the use of social media during specific a smartphone app for emergencies – in contrast, most participants events, but some large studies also try to focus on citizens' perception (60%) indicated it was likely that they would download an app in future (American Red Cross, 2012; Canadian Red Cross, 2012). However, it is for an information exchange with emergency services in an appropriate doubtful whether findings from America can be assumed to be transfer- situation. It seems that the general awareness of such tools depends on able to Europe. Only one study was found which sheds some light on the the frequency of emergencies someone is confronted with – something comparative situation in Europe, as discussed above, but it included other studies also suggest (Reuter, 2014a). In other cases it is likely that only a relatively small number of respondents (Flizikowski et al., such tools are just used if they are integrated in daily used media, such 2014). This article has therefore sought to provide recent evidence of as Facebook. how European citizens are disposed towards the use of social media in The current study has also shown that accompanying many citizens' emergencies. increased use of social media in emergencies is a growing expectation Fig. 16 summarizes some of the main results and points to selected for emergency services to communicate with citizens via social media facts. and to make use of information shared by citizens via social media. This study has shown that many citizens across Europe are already Thus, the majority (69%) of citizens agreed that emergency services using social media to share and look for information during emergen- should regularly monitor their social media sites, and 41% expected a cies and that they expect their usage to increase in future. In particular, response within an hour (Fig. 16: Fact 6). This is higher than in the around a quarter (27%) of citizens said that they had used social media Canadian study where 63% of participants thought that emergency re- for information sharing (Fact 1 in Fig. 16, Q17) and 43% had used it to sponders should be prepared to respond to calls for help posted on so- look for information during an emergency. The most popular shared cial media. This could be explained by the fact that the Canadian study topics were weather conditions or warnings (66%) and road or traffic was conducted over three years ago and could suggest that our study re- conditions (64%) (Fig. 16: Fact 3). Furthermore, the survey suggests flects an increased awareness among citizens of social media and how it that citizens expected to increase their use of social media for such pur- can be used during emergencies – with greater demands made to emer- poses in future. This includes 48% of participants who thought it was gency services to use and be responsive to social media in disaster situ- likely that they would share emergency information on social media ations. However, a recent survey conducted in the same project (Reuter platforms in future (Fig. 16: Fact 7) and 58% who thought that they et al., 2016b) has shown that even though many emergency services would use social media to look for information. The main positive rea- sometimes use social media to share information with the public, only sons for using social media as an information source included that it very few often make use of data on social media during emergencies. was seen as faster (76%) and more accessible (54%) than conventional This reality was reflected in citizens' responses, as part of which 56% media. The levels of use of social media was considerably higher thought that emergency services were currently too busy during an than found in previous studies, including a study in the USA in 2012 emergency to respond to a request for help or information. Further- (American Red Cross, 2012) which found that 12% of the general public more, many of those responding to an open-ended question asking par- and 22% of high school graduates had used social media to share or ticipants to explain what would encourage them to share information obtain information during emergencies and disasters. The types of in- via social media said that this depended on a clearer purpose for sharing formation shared and looked for most frequently, however, (weather, information, in particular that emergency services would definitely traffic, damage caused and information on how other people were make use of such information. This suggests that citizens' perception coping) were very similar. of the behaviour of emergency services does not match their expecta- The survey also showed that use of social media for private purposes tions of how they would like them to behave in relation to social media. and in emergencies was not uniform demographically and that particu- The before mentioned gap between citizens' perception and emer- lar types of citizens are more likely to do so than others. This means that gency services behaviour means that the potential of social media aiding

Fig. 15. Future use of apps for information exchange in emergency situations (Q13).

Fig. 16. Infographic of Selected Survey Results (designed by our project partner OCC). the social resilience of citizens during emergencies is not yet fully real- recovery (to ﬁx damages quickly) and creativity (to learn from losses ized (Boin et al., 2010; Maguire and Hagan, 2007) and that both emer- and improve the system in future). Thus, the emergence of so called gency services and citizens need further support and encouragement emergent citizen groups (Stallings and Quarantelli, 1985) is today often to ﬁnd ways of using social media more effectively and cooperatively. initiated by social media, based on the activities of digital volunteers If supported in this way, such a concept of resilience could be seen as (Starbird and Palen, 2011). Technologies supporting collaboration not linking with other related concepts of resistance (to prevent damages), just within a planned frame, but that allow emergent collaboration,

Please cite this article as: Reuter, C., Spielhofer, T., Towards social resilience: A quantitative and qualitative survey on citizens' perception ofsocial media in emergencies in Europe, Technol. Forecast. Soc. Change (2016), http://dx.doi.org/10.1016/j.techfore.2016.07.038 C. Reuter, T. Spielhofer / Technological Forecasting & Social Change xxx (2016) xxx–xxx 11 the “need for spontaneous collaboration in novel and changing struc- ▪ Q4: *In what country do you currently live? (Australia, Austria, tures”, such as ad hoc participation, are needed (Reuter, 2014b, p. ix). Belgium, Bosnia & Herzegovina, Bulgaria, Canada, Croatia, Cyprus, This would suggest that increasing cooperation could increase social re- Czech Republic, Denmark, Estonia, Finland, France, Germany, Greece, silience via a complex web of (collaborating) actors in social media, Hungary, Ireland, Iceland, Italy, Latvia, Lithuania, Luxembourg, which is in line with the results on cooperative resilience and coopera- Macedonia, Malta, Moldova, Netherlands, Norway, Poland, Portugal, tion technologies (Reuter et al. 2016a). One contribution of this article Romania, Russia, Serbia, Slovakia, Slovenia, Spain, Sweden, is to outline the perception of citizen in Europe, which is a necessary Switzerland, Turkey, Ukraine, United Kingdome, USA, Other prerequisite for approaches addressing the issue. (please specify)) The survey has also revealed that citizens were more likely to use ▪ Q5: *Do you currently work or volunteer for an emergency ser- Facebook (73%) and YouTube (69%) rather than Twitter (21%) for pri- vice, such as the police, fire service or a medical service? (Yes, vate purposes. This is of particular interest given the fact that many in- no, other (please specify)) ternational studies on the role and use of social media in emergencies ▪ Q6: *Do you have any children under the age of 18? (Yes, no) rely on the analysis of Twitter only, as it is more accessible for research ▪ Q7: *What type of mobile telephone do you PRIMARLY use? (Regu- purposes (Kaufhold and Reuter, 2016). It means though that a lot of lar cell/mobile phone (not a smartphone), android (e.g. Samsung S5, citizen activity on social media during emergencies remains under- HTC One, Google Nexus 5, …), iPhone, Blackberry/RIM, Windows researched. This has been one of the limitations of the current study – mobile (e.g. Nokia Lumia 530, 820 or 930, HTC window), I don't that we have not explored differences in the use of different types of have a mobile phone, other (please specify)) social media platforms during emergencies. It is hoped that future stud- ▪ Q8: *To what extend do you currently use the following types of ies will fill this gap. It is also worth noting that almost all participants of social media to share or look for information? (Often, sometimes, the Canadian Red Cross (2012) study were Facebook users (97%). This seldom, never) might suggest that our sample was more representative of the whole - Facebook population, including older people who on average use social media -Twitter much less. Furthermore, about three quarters of the respondents - Instagram are from the countries Poland, Slovenia, Germany and the United -YouTube Kingdom. Therefore we cannot necessarily draw conclusions for the ▪ Q9: *Please indicate how strongly you agree or disagree with the fol- whole of Europe. A representative study among selected countries in lowing statements (Strongly Agree, agree, nether agree or disagree, Europe might help to fill this gap. However, the answers did not differ disagree, strongly disagree) by country a lot. The participants from Slovenia use Facebook more - In my private life, I use social media very often often (91% “very often”) than survey participants overall (60%); and in - It is important for emergency services to use social media to keep the UK, Twitter is used more often (30% “very often”) compared with in touch with the public during emergencies. survey participants overall (12%, Q8). Furthermore most people in - Most of my friends use social media to keep in touch. Germany said that they used the local radio as an information channel - I regularly post messages on social media such as Twitter, during an emergency (81%) while the average reported use across the Facebook or Instagram. whole sample was 54% (Q10). Otherwise, there were no significant dif- - I would not trust a message posted on social media during an ferences between countries or the number of responses from countries emergency unless it came from an official source such as the po- was too small to make any valid comparisons. lice or fire-service. Finally, our study identified very similar barriers to the increased use - Emergency services should not trust information on social media. of social media as found in the only other study of citizens' attitudes in ▪ Q10: *Which of the following communication channels have you Europe (Flizikowski et al., 2014). The previous study found that both cit- ever used to get information about an emergency, such as a power izens and emergency services identified the same challenges, such as a cut, severe weather, flood or earthquake? (TV News, local radio sta- lack of knowledge, personnel issues, uniform terms of use, credibility tion, online news, mobile apps, social media (Facebook, Twitter, of citizen-generated content, and accessibility for older generations. In Instagram, etc.), text message alerts, online sites for disaster agen- comparison, our study mainly identified ‘mistrust’ as well as the per- cies or emergency services, none of the these, other (please specify)) ceived lack of a clear purpose for using social media in emergencies ▪ Q11: *Have you ever downloaded a smartphone app that could help These points can be addressed if citizens gain awareness that the in a disaster or emergency? (Yes, no, don't know/not sure) value of the information being provided fits their expectations and ▪ Q12: *What kind of app did you download and for what purpose? needs. ▪ Q13: *How likely is it that you would download an app in future to use on a smartphone to share information with, or receive informa- Acknowledgements tion from, emergency services as a result of an emergency such as an accident, power cut, severe weather, flood or earthquake? (Very The research project EmerGent’ was funded by a grant of the likely, quite likely, neutral, not very likely, not at all likely) European Union (FP7 No. 608352). We would like to thank all members ▪ Q14: *Have you ever used social media such as Facebook, Twitter, of our project for distributing our survey. Instagram, etc. to find out information as a result of an emergency such as an accident, power cut, severe weather, flood or earthquake? Appendix A. Survey description and questions. (Yes, no, don't know/not sure) ▪ Q15: *What kind of information were you looking for? (Weather Welcome, the following questions will take only about 5–10 min to conditions or warnings, road or traffic conditions, damage caused answer. We thank you in advance for your participation. by the event, the location or status of friends/family, information For the completion of the questionnaire it is mandatory to answer all about how others are coping with the disaster, eyewitness photo- questions, unless the question specifies otherwise. graphs or videos, what to do to keep yourself safe, other (please specify)) ▪ Q1: *Please indicate that you agree to participate in this survey: (Yes ▪ Q16: *How likely are you to use social media in future to look for in- (I agree), No (I do not wish to participate)) formation as a result of an emergency such as an accident, power ▪ Q2: *What is your age? (under 21, 21–29, 30–39, 40–49, 50–59, 60 cut, severe weather, flood or earthquake? (Very likely, quite likely, or older) neutral, not very likely, not at all likely) ▪ Q3: *What gender are you? (Male, female, other) ▪ Q17: *Have you ever used social media to share information as a

result of an emergency such as an accident, power cut, severe Flizikowski, A., Hołubowicz,W.,Stachowicz,A.,Hokkanen,L.,Delavallade,T., – fl 2014. Social media in crisis management the iSAR + project survey. Pro- weather, ood or earthquake? (Yes, no, don't know/not sure) ceedings of the Information Systems for Crisis Response and Management ▪ Q18: *What types of information did you share? (Weather condi- (ISCRAM), pp. 707–711. tions or warnings, road or traffic conditions, reassurance that you Helsloot, I., de Vries, D., Groenendaal, J., Scholtens, A., 2015. In: ‘tVeld,M.,vanMelick,G., Blaha, M. (Eds.), Guidelines for the Use of New Media in Crisis Situations Retrieved are safe, your feelings or emotions about what was happening, from http://www.cosmic-project.eu/sites/default/files/Guidelines for the use of new your location, what actions you were taking to stay safe, an eyewit- media in crisis situations.pdf#page = 1&zoom = auto,-130,842. ness description of something you experienced, advice about what Hiltz, S.R., Diaz, P., Mark, G., 2011. Introduction: social media and collaborative systems – actions others should take to stay safe, an eyewitness photo, a for crisis management. ACM Trans. Comp. Human Inter. (ToCHI) 18 (4), 1 6. Hughes, A.L., Denis, L.A.S., Palen, L., Anderson, K.M., 2014. Online public communications video, other (please specify)) by police & fire services during the 2012 hurricane Sandy Proceedings of the Confer- ▪ Q19: *How likely are you to use social media in future to share infor- ence on Human Factors in Computing Systems (CHI). mation with others during or after an accident or emergency? (Very Hughes, A.L., Palen, L., 2009. Twitter adoption and use in mass convergence and emergency events. In: Landgren, J., Jul, S. (Eds.), Proceedings of the Information Systems for likely, quite likely, neutral, not very likely, not at all likely) Crisis Response and Management (ISCRAM). Gothenburg, Sweden. ▪ Q20: *What would make you (even) more likely to use social media Hughes, A.L., Palen, L., 2012. The evolving role of the public information officer: an exam- for this purpose in future? ination of social media in emergency management. J. Homeland Sec. Emerg. Manag. 9 ▪ (1). http://dx.doi.org/10.1515/1547-7355.1976. Q21: *Some people prefer now to look for information via social Hughes, A.L., Tapia, A.H., 2015. Social media in crisis: when professional responders meet media rather than via TV, radio or traditional websites for the digital volunteers. J. Homeland Secur. Emerg. Manag. 12 (3), 679–706. http://dx.doi. following reasons (Strongly agree, agree, nether agree or disagree, org/10.1515/jhsem-2014-0080. IBM, 2014. Stat. Packag. Soc. Sci. Retrieved from https://www14.software.ibm.com. disagree, strongly disagree) International Association of Chiefs of Police, 2010. 2010 Social Media Survey - The information via social media is available faster Results. Retrieved from http://www.iacpsocialmedia.org/Portals/1/documents/ - The information via social media is more accurate SurveyResultsDocument.pdf. - The information via social media is more accessible International Association of Chiefs of Police, 2015. 2015 Social Media Survey Re- sults. Retrieved from http://www.iacpsocialmedia.org/Portals/1/documents/ - The information via social media is more reliable FULL2015SocialMediaSurveyResults.pdf. - The information via social media is richer Kaplan, A.M., Haenlein, M., 2010. Users of the world, unite! The challenges and opportu- – ▪ Q22: *Imagine that you posted an urgent request for help or nities of social media. Bus. Hor. 53 (1), 59 68. http://dx.doi.org/10.1016/j.bushor. 2009.09.003. information on a social media site of a local emergency response or- Kaufhold, M.-A., Reuter, C., 2016. The self-organization of digital volunteers across social ganization, such as your local police, coastguard, fire or medical media: the case of the 2013 European floods in Germany. J. Homeland Sec. Emerg. – emergency service. To what extent do you agree with the following Manag. (HSEM) 13 (1), 137 166. Likert, R., 1932. A technique for the measurement of attitudes. Arch. Psychol. 140, 1–55. statements (Strongly agree, agree, nether agree or disagree, dis- Ludwig, T., Reuter, C., Siebigteroth, T., Pipek, V., 2015. CrowdMonitor: mobile crowd sens- agree, strongly disagree) ing for assessing physical and digital activities of citizens during emergencies. Pro- - Emergency services should regularly monitor their social media so ceedings of the Conference on Human Factors in Computing Systems (CHI). ACM Press, Seoul, Korea. they can promptly respond to such a request. Maguire, B., Hagan, P., 2007. Disasters and communities: understanding social resilience. - It is very likely that the emergency service will not know about Aust.J.Emerg.Manag.22(2),16–20. this request. Murphy, T., Jennex, M.E., 2006. Knowledge management, emergency response, and hurricane Katrina. Int. J. Intel. Control Syst. 11 (4), 199–208. - I would expect to get a response from them within an hour. Palen, L., Liu, S.B., 2007. Citizen communications in crisis: anticipating a future of ICT- - Emergency services are too busy to monitor social media during supported public participation. Proceedings of the Conference on Human Factors in an emergency. Computing Systems (CHI). ACM Press, San Jose, USA. ▪ Palen, L., Vieweg, S., Liu, S.B., Hughes, A.L., 2009. Crisis in a networked world: fea- Q23: *To what extent are you aware or have you heard of the follow- tures of computer-mediated communication in the April 16, 2007, Virginia ing services provided via social media? (Very aware, moderately Tech event. Soc. Sci. Comput. Rev. 27 (4), 467–480. http://dx.doi.org/10. aware, somewhat aware, slightly aware, not at all aware) 1177/0894439309332302. - Twitter alerts - used by key organizations to send alerts during Perng, S.-Y., Büscher, M., Wood, L., Halvorsrud, R., Stiso, M., Ramirez, L., Al-Akkad, A., 2013. Peripheral response: microblogging during the 22/7/2011 Norway attacks. In: emergencies Rothkrantz, L., Ristvej, J., Franco, Z. (Eds.), International Journal of Information Sys- - Facebook safety checks - allows you during an emergency to let tems for Crisis Response and Management (IJISCRAM). 5(1). Pipek, V., Liu, S.B., Kerne, A., 2014. Special issue: crisis informatics and collaboration. friends and family know you and friends are safe – ▪ Comp. Supported Coop. Work (CSCW) 23 (4 6). Q24: *Is there anything else you want to mention with regard to the Plotnick, L., Hiltz, S.R., Kushma, J.A., Tapia, A., 2015. Red tape : attitudes and issues related use of social media before, during or after an emergency? to use of social media by U.S. County - level emergency managers. Proceedings of the ▪ Q25: *Thank you very much for your help in completing this survey! Information Systems for Crisis Response and Management (ISCRAM). Kristiansand, fi Norway. The ndings will be made available on our project website (click Reuter, C., 2014a. Communication between power blackout and mobile network over- here to access the website). Please provide an email address if you load. Int. J. Inform. Syst. Cris. Resp. Manag. (IJISCRAM) 6 (2), 38–53. would like to receive a link to the summary findings from this re- Reuter, C., 2014b. Emergent Collaboration Infrastructures: Technology Design for Inter- Organizational Crisis Management (Ph.D. Thesis) Springer Gabler, Siegen, Germany Re- search when they are available. trieved from http://www.springer.com/springer+gabler/bwl/wirtschaftsinformatik/ ▪ Q26: *Would you be willing to take part in a similar survey in future? book/978-3-658-08585-8. (Yes, no) Reuter, C., 2015. Guest editorial preface: special issue on human computer interaction in critical systems I: citizen and volunteers. Int. J. Inf. Syst. Cris. Response Manage. (IJISCRAM) 7 (2). Reuter, C., Heger, O., Pipek, V., 2013. Combining real and virtual volunteers through social References media. Proceedings of the Information Systems for Crisis Response and Management (ISCRAM), pp. 1–10. American Red Cross, 2012. Social Media in Disasters and Emergencies. Retrieved from Reuter, C., Ludwig, T., Kaufhold, M.-A., Pipek, V., 2015a. XHELP: design of a cross-platform http://www.a1881.g.akamai.net/7/1881/26640/v0001/redcross.download.akamai. social-media application to support volunteer moderators in disasters. Proceedings of com/26640/PollData/Social. the Conference on Human Factors in Computing Systems (CHI). ACM Press, Seoul, Boin, A., Comfort, L.K., Demchak, C.C., 2010. The rise of resilience. Designing Resilience, Korea. pp. 1–12. Reuter, C., Ludwig, T., Kaufhold, M.-A., Spielhofer, T., 2016b. Emergency services attitudes Canadian Red Cross, 2012. Social Media during Emergencies. Retrieved from http://www. towards social media: a quantitative and qualitative survey across Europe. Int. redcross.ca/cmslib/general/pub_social_media_in_emergencies_survey_oct2012_en.pdf. J. Human Comp. Stud. (IJHCS). Cronbach, L.J., 1951. Coefficient alpha and the internal structure of tests. Psychometrika Reuter, C., Ludwig, T., Pipek, V., 2016a. Kooperative Resilienz – ein soziotechnischer ansatz 16 (3), 297–334. durch Kooperationstechnologien im Krisenmanagement. Gruppe. Interaktion. Orga- Denef, S., Bayerl, P.S., Kaptein, N., 2013. Social media and the police — tweeting practices nisation. Z. Angew. Org. Psychol. (GIO). of British police forces during the August 2011 riots. Proceedings of the Conference Reuter, C., Ludwig, T., Ritzkatis, M., Pipek, V., 2015b. Social-QAS: tailorable quality assess- on Human Factors in Computing Systems (CHI), pp. 3471–3480. ment service for social media content. In Proceedings of the International Symposium Field, A., 2009. Discovering Statistics Using SPSS-. third ed Sage, London, United Kingdom. on End-User Development (IS-EUD). Lect. Notes Comput. Sci.

Reuter, C., Marx, A., Pipek, V., 2012. Crisis management 2.0: towards a systematization of White, J.I., Palen, L., Anderson, K.M., 2014. digital mobilization in disaster response: the social software use in crisis situations. Int. J. Inf. Syst. Crisis Resp. Manage. 4 (1), 1–16 work & self-organization of on-line pet advocates in response to hurricane Sandy. (IJISCRAM). Proceedings of the Conference on Computer Supported Cooperative Work (CSCW). San, Y.S., Wardell III, C., Thorkildsen, Z., 2013. Social Media in the Emergency Management ACM, Baltimore, USA. Field: 2012 Survey Results (June). Simão, J., Luís, B., Schmidt, S., Rhode, D., Freitag, S., Lück, A., ... Villot, E., 2015. iSAR+ Guide- Dr. Christian Reuter is senior researcher at the University of Siegen. He studied Informa- lines: Online and Mobile Communications for Crisis Response and Search and Rescue. tion Systems at the University of Siegen, Germany and the École Supérieure de Commerce Retrieved from http://isar.i112.eu/downloads/ﬁles/D2271-iSARGuidelinesRoadmap.pdf de Dijon, France and received a PhD for his work on (inter-)organizational collaboration (there is a human perspective described based on quite comprehensive survey). technology design for crisis management (Dr. rer. pol.) with summa cum laude. He has Stallings, R.A., Quarantelli, E.L., 1985. Emergent citizen groups and emergency manage- worked as a web developer, consultant and researcher and has published about 85 scien- ment. Public Adm. Rev. 45 (Special Issue), 93–100. tiﬁc articles, many focusing on ICT and social media in emergencies. He is voluntary Starbird, K., Palen, L., 2011. Voluntweeters: Self-organizing by digital volunteers in times founding spokesman of the section "Human-Machine Interaction in Safety-Critical Sys- of crisis. Proceedings of the Conference on Human Factors in Computing Systems tems" of the German Informatics Society. He is work package leader within the EU- (CHI). ACM-Press, Vancouver, Canada. project EmerGent on social media in emergencies. Statista, 2015. Leading Social Networks Worldwide as of January 2015, Ranked by Num- ber of Active Users (in Millions). Retrieved from http://www.statista.com/statistics/ Dr. Thomas Spielhofer is a principal researcher working at the Tavistock Institute for Hu- 272014/global-social-networks-ranked-by-number-of-users/. man Relations (TIHR) with an expertise in leading qualitative and quantitative research Strauss, A.L., 1987. Qualitative Analysis for Social Scientists. Cambridge Press. projects in the UK and Europe. Thomas combines an expertise in educational research United Nations, 2009. 2009 UNISDR Terminology on Disaster Risk Reduction. Internation- with recent experience in policy-focussed research on mental health and family relation- al Stratergy for Disaster Reduction (ISDR). United Nations International Strategy for ships, as well as a strong interest in studying the use and impact of social media in emer- Disaster Reduction (UNISDR), Geneva. gency situations. He is work package leader within the EU-project EmerGent on social Valecha, R., Oh, O., Rao, R., 2013. An exploration of collaboration over time in collective media in emergencies. crisis response during the Haiti 2010 earthquake. Proceedings of the International Conference on Information Systems (ICIS). Milan, Italy , pp. 1–10 Retrieved from http://aisel.aisnet.org/icis2013/proceedings/ResearchInProgress/96/.

Abstract. Catastrophes like the storms in Europe in spring 2016, but also terrorist attacks regularly not only lead to extensive monetary damage but also threaten human lives. In such situations, the population desires information about the status of damage and safe behav- iours. Apps can address this potentially, but comparatively, record a low number of users. Based on Reuter et al. (2017), this article shows the importance of information in disaster situations and illustrates features from current mobile warning apps.

1. Motivation In recent years, there have been major disasters in various parts of Germany again and again. These were particularly often natural disasters such as storms and floods (Münchener Rückversicherungs- Gesellschaft, 2016). Their consequences show that the issue of disaster control is also relevant in Germany. Floods, storms or thereby resulting power failures can affect every person at any time. Moreover, the damage has been reinforced due to changing living conditions such as settlements on the coasts. Therefore, it is important to develop precautionary measures early. Here, the preparation of affected people especially matters. Since such catastrophes do not happen as frequently in Germany as in other countries, it is possible that the persons 1

are not sufficiently acquainted with protective measures (Bundesverwaltungsamt, 2006). In case of a catastrophe, the consequences could be greater due to the insufficient experience than if preparations had previously been made. Another problem during a disaster is that the civil protection, for example, through the civil protection law by the federal states is obsolete and should be adjusted. In a disaster situation, many volunteer helpers are available (Kaufhold & Reuter, 2016). These must, however, be well-informed and coordinated to help as many affected persons as possible. Furthermore, premature informing and warning of large parts of the population play an important role to make preparations as early as possible.

2. Informing the Population Informing the population plays a significant role to get a grip on the disaster situation, especially in the phase immediately after the catastrophe. The government should take the lead in all the different places where citizens can obtain information – “not only in the actual physical management of the event but also in the information of the affected population“ (Sulzberger & Glattbrugg, 2014), whereby private organizations can work supportively. However, they only contain a “partial overview of the happening and can therefore only give limited information”. In general, the government should use its current information advantage for the protection of the population. This protection of the citizens is the primary goal, financial aspects or the good reputation are only in second place (Coombs, 2009). Even from the citizens’ point

of view, the responsibility for the protection of all lies with the government. Self-protection stagnates as past events are quickly displaced or forgotten (Menski & Gardemann, 2008). Only citizens, who often deal with crisis situations, protect themselves better (Bundesverwaltungsamt, 2006). Another problem is that the broad population is not familiar with existing concepts of risk communication (Helsloot & Beerens, 2009; Menski & Gardemann, 2008). Learning from past crisis situations may also help to avert future crises or its consequences and reactions (Coombs, 2009). Since there have not been so many bigger crisis situations in Germany yet, the learning behaviour can be classified as low.

Besides, in crisis situations, there is a growing information demand and coordination effort on the part of the population and therefore also on the part of the media (Kaufhold & Reuter, 2016). It is problematical that the information cannot be confirmed as quickly as the media demand and disseminate it. Possible subsequent events, as well as actions and speculations, also spread very quickly (Sulzberger & Glattbrugg, 2014). However, studies show that rapid information is more useful than risky (Coombs, 2014). Therefore, it is important to provide the public with information from the official side as soon as possible before they spread their own speculations, misinformation or unnecessary fears. That is usually done via the mass media. A positive side effect is that the trust in the emergency organizations is strengthened. Because immediately after the occurrence of a catas-

trophe the need for protection and security in the population is particularly high (Bollmann, 1992). If the population is not provided with information, this might lead to general uncertainty and the feeling of being left alone, especially in long-running crises (Reuter, 2014). As many people are scared and helpless, simple instructions are most suitable to take people to action. Information can be divided into basic plans and instructions, including best practices, for example. In this context, Coombs (2009) coins the term “instructing information”. That is information how the affected people can protect themselves.

Before an occurring crisis, it is mandatory to give a warning (Volgger et al., 2006). In the case of predictable crises, specific warnings must be sent via every possible channel to enable everyone to prepare themselves individually. That is of importance because the self-protection stagnates at a low level. For instance, the supply of food and drinking water in private households decreases because everything is constantly available in the supermarket. Especially in cities, the sup- plies are low (Menski & Gardemann, 2008) since there is no more scarcity. Information is important not only for preparation but also as sensitization for uncertain situations. It is crucial to create an awareness of possible crises so that citizens take precautionary measures (Geenen, 2009). Therefore, citizens need to be informed of current threats or dangers, as well as their probability of occurrence and possible consequences. Using such information, citizens have a vague idea of their behaviour in an emergency (Volgger et al., 2006).

During a crisis, the need for information in the population should always be covered, and a consistent and transparent informing of the affected people is required (Nilges et al., 2009). For the orientation of the population, the assumed duration of a crisis is one of the most valuable information. If an exact time cannot be specified, an estimated time should be given so that citizens can rearrange themselves and adapt to the situation.

3. Warning Apps in Disaster Situations

A Red Cross study has dealt with the topic of how people in disaster situations use the possibilities of the Internet (Wade, 2012). According to this study, 55% of the respondents obtain information about emergencies online. However, only 20% use mobile applications to inform themselves. If no further investigations are done on this subject, and therefore no new developments arise, it seems highly improbable that this number will increase in the future. The information that is relevant to users is, particularly, weather conditions, weather warnings, and damage caused by disasters. Approximately 70% of users search for this. Another need of the persons is to inform others about their safety as well as finding out other people’s current situation (Wade, 2012).

When using mobile applications for crisis situations, a distinction must be made between whether the information is location-based or not (Vieweg et al., 2010). If this is the case, people can be supported to become aware of their current situation. For example, concrete information about the status can be given. Location-based information is 5

dependent on GPS (Global Positioning System) or Internet access, although valuable information can still be obtained from a mobile application for crisis situations without location detection. For instance, mobile crisis applications often provide general information (for action) for specific situations and support the exchange with others. Thereby, users can exchange experiences, but also share new information with other people (Karl et al., 2015). In the following, three mobile applications are presented and it is discussed particularly how the apps can help in crisis situations.

Figure 1: KATWARN, NINA and FEMA warning apps

Katastrophen-Warnung (KATWARN)

The core idea of KATWARN is to warn users in different emergency situations like „conflagrations, bomb finds or storms“ (Köllen, 2015). The app’s information is directly gathered from disaster control authorities, fire brigades or storm headquarters. These send concrete data via push messages to the app users. The information is location-based so that users are informed about near emergencies or crises. Accord- ing to Köllen (2015), the goal of KATWARN is to „quickly clarify about the imminent danger and provide advice on the ideal behaviour“.

In case of a relevant emergency, the user receives a push notification and, depending on the settings, an alarm sound or vibration is trig- gered. Each warning contains a short message with an overview of the sender of the message, the type of emergency, and the recommended behaviour in the current situation. The app allows to person- alize warnings regarding your current position, e.g. via GPS, or predefined locations. Users may deactivate the GPS functionality, but the option to save up to seven areas allows the user to stay informed without location detection. The attribution of a place to a certain area is realized using postcodes or place marks.

The warning may be shared, e.g. on social media, to inform further people about the emergency. Finally, the app contains a test alarm functionality, which triggers a fictional warning so that the user can test and customize the warning behaviour of KATWARN. For future ver-

sions, it is planned to establish a function that alerts emergency responders. That comprises the registration of personnel like doctors and paramedics to request to inform them and ask for their help during emergencies. Notfall-Informations- und Nachrichten-App (NINA) NINA is an app which is supposed to warn against floods or other crises (BBK, 2015). Besides actual warnings, NINA also provides general advice and tips on how to behave during different emergency or crisis situations. The information is obtained from a modular warning system as well as from the German Weather Service and the Water- way and Ship Management of the Federation. The website of BBK (2015) summarizes the following functions of NINA: Warning, emergency contacts, emergency advice, data protection and additional functions.

In the warnings section, all official warnings are enlisted as well as information regarding flood levels or storm warnings. These warnings may be visualized in a simple list view or in a map to illustrate the locality. The map view also allows a colour-coded and symbolic sorting of the entries. For instance, blue warnings indicate a warning from the modular warning system. The flood levels of rivers are illustrated as dots on the map. The user may receive warnings via push messages.

The function for emergency contacts was designed to support users who are affected by an emergency. It allows to deposit various con-

tacts that are informed quickly via e-mail or SMS in case of emergency. If the emergency is over or the user wants to state that he is not affected, the implemented all-clear function may be used. Con- cerning general guidance and preparatory measures, a function with emergency tips is available. This general information is relevant before, during and after an emergency. Data protection is also an essential topic of the mobile app of BBK. Thus, the location and other information are directly queried from BBK and data is only processed on the system locally.

Under the term of additional functions, BBK summarizes the geolocation and the offline mode of NINA. If the geolocation is activated via GPS, WLAN or other services, the user may receive location-based warnings. But NINA may also be used in offline mode. In this case, the user’s last location with a connection is used. In summary, the main function of NINA is to „bridge the last mile to all people who are in a place of danger and who are not reachable via common warning messages like sirens or radio announcements“ (Schwan, 2015). Federal Emergency Management Agency App (FEMA) FEMA is a mobile application of the U.S. Government Federal Emer- gency Management Agency (FEMA) (Goncalves et al., 2014). Accord- ing to the website of the Department of Homeland Security (2015), the major function of the app is the provision of warnings from the national weather service. It allows to save up to five areas within the U.S. about which the user receives weather warnings.

Another function is the Disaster Reporter. It allows users to upload pictures and enrich them with location indicators and a short message to share their current situation and exchange with other people (Goncalves et al., 2014). Moreover, personal security information may be saved and aggregated. Within an emergency box, the user can save a list of objects or different locations where users can meet, e.g. with their family, or how they communicate in emergencies.

For information during an emergency, the app contains a map which shows assembly points and how the user reaches them. Moreover, general security advice is enlisted about preparations for an emergency, correct behaviour during an emergency and the follow-up tasks. These tips are available for 20 different types of emergencies like floods, hurricanes or earthquakes. Lastly, the function of request- ing help is enlisted. The app allows the users quick access to different support websites where users can ask for help (Department of Homeland Security, 2015).

4. Discussion Requirements for warning apps comprise general information (e.g. information for action), location-relevant information (e.g. the estimated time of a blackout or contact addresses) and setting-specific information (e.g. for persons with special needs) (Reuter & Ludwig, 2013). Often mentioned prerequisites for the successful use of an app in crises are location detection to transmit location-specific information in

real-time, tailorability, the display of emergency numbers, the facilita- tion of feedback and current contents (Karl et al., 2015). On behalf of the users, within app reviews, the functionality and stability are criti- cized mainly (Kotthaus et al., 2016). The core functionalities of the presented warning apps are (1) area-based and (2) location-based warnings featuring a (3) warning map, (4) general disaster information, as well as capabilities for (5) information sharing and (6) disaster reporting.

Features KATWARN NINA FEMA Area-based warnings    Location-based warnings    Warning map    General disaster information    Information sharing    Disaster reporting    Table 1: Core features of KATWARN, NINA and FEMA

However, the motivation of citizens to inform themselves proactively seems to be low (Reuter & Ludwig, 2013). Reuter et al. (2017) examined user behaviour and motivation of citizens regarding crisis apps and concrete recommendations for the design of such apps, based on a comparison of KATWARN, NINA and FEMA. The quantitative study (n=1.034) revealed a low interest among citizens to install such an app. Most participants (71%) indicated that they never downloaded a smartphone app for emergencies or crises, 7% were not sure and only

16% said they did it (Reuter et al., 2017; Reuter & Spielhofer, 2016). Promoting the acceptance of and motivation of using disaster-purpose apps seems to be an important research topic to ensure the success and utility of these apps in terms of social resilience. References

BBK. (2015). Warn-App NINA. Retrieved from www.bbk.bund.de/DE/NINA/Warn-App_NINA.html Bollmann, U. (1992). Seminar SFU 92. In Führung in außerordentlichen Lagen, Überlegungen zur Kimmunikation als Element der Gesamtstrategie. Bern. Bundesverwaltungsamt. (2006). Zweiter Gefahrenbericht der Schutzkommission beim Bundesminister des Innern. Schriftenreihe der Schutzkommision beim Bundesminister des Innern. Bundesverwaltungsamt. Coombs, W. T. (2009). Conceptualizing Crisis Communication. In R. L. Heath & D. O’Hair (Eds.), Handbook of Risk and Crisis Communication (pp. 99–118). Routledge, New York. Coombs, W. T. (2014). Ongoing crisis communication: planning, managing, and responding. Sage Publications Ltd. Department of Homeland Security (DoHS). (2015). Mobile App. Retrieved from http://www.fema.gov/mobile-app Geenen, E. M. (2009). Warnung der Bevölkerung. In Schutzkommission beim Bundesminister des Inneren (Ed.), 12

Gefahren und Warnung (pp. 61–102). Bonn. Goncalves, A., Silva, C., & Morreale, P. (2014). Design of a Mobile Ad Hoc Network Communication App for Disaster Recovery. In 2014 28th International Conference on Advanced Information Networking and Applications Workshops (pp. 121–126). IEEE. https://doi.org/10.1109/WAINA.2014.26 Helsloot, I., & Beerens, R. (2009). Citizens’ response to a large electrical power out-age in the Netherlands in 2007. Journal of Contingencies and Crisis Management, 17(1), 64–68. Karl, I., Rother, K., & Nestler, S. (2015). Crisis-related Apps: Assistance for Critical and Emergency Situations. International Journal of Information Systems for Crisis Response and Management (IJISCRAM), 7(2). Kaufhold, M.-A., & Reuter, C. (2016). The Self-Organization of Digital Volunteers across Social Media: The Case of the 2013 European Floods in Germany. Journal of Homeland Security and Emergency Management (HSEM), 13(1), 137–166. Köllen, K. (2015). Smartphone So wird das Handy zum Lebensretter. Kotthaus, C., Ludwig, T., & Pipek, V. (2016). Persuasive System Design Analysis of Mobile Warning Apps for Citizens. In Adjunct Proceedings of the 11th International Conference on Persuasive Technology. Salzburg, Austria. Menski, U., & Gardemann, J. (2008). Auswirkungen des Ausfalls Kritischer Infrastrukturen auf den Ernährungssektor am Beispiel 13

des Stromausfalls im Münsterland im Herbst 2005. Münster: Fachhochschule Münster. Münchener Rückversicherungs-Gesellschaft. (2016). Schadenereignisse in Deutschland 1980 – 2015. Nilges, J., Balduin, N., & Dierich, B. (2009). Information and Communication Platform for Crisis Management (IKK). In Proceedings of the International Conference and Exhibition on Electricity Distribution (CIRED). Prague, Czech Republic. Reuter, C. (2014). Communication between Power Blackout and Mobile Network Overload. International Journal of Information Systems for Crisis Response and Management (IJISCRAM), 6(2), 38–53. Reuter, C., Kaufhold, M.-A., Leopold, I., & Knipp, H. (2017). Katwarn, NINA or FEMA? Mixed-Method Study on Distribution, Use and Public Views on Crisis Apps. In European Conference on Information Systems (ECIS). Guimarães, Portugal. Reuter, C., & Ludwig, T. (2013). Anforderungen und technische Konzepte der Krisenkommunikation bei Stromausfall. In M. Hornbach (Ed.), Informatik 2013 - Informatik angepasst an Mensch, Organisation und Umwelt (pp. 1604–1618). Koblenz, Germany: GI-Edition-Lecture Notes in Informatics (LNI). Reuter, C., & Spielhofer, T. (2016). Towards Social Resilience: A Quantitative and Qualitative Survey on Citizens’ Perception of Social Media in Emergencies in Europe. Journal Technological Forecasting and Social Change (TFSC), onlinefirs. 14

Schwan, B. (2015). NINA-App soll vor örtlichen Gefahren warnen. Sulzberger, M., & Glattbrugg, S. (2014). Katastrophenmanagement. (O. Grün & A. Schenker-Wicki, Eds.), Springer. Wiesbaden: Springer Fachmedien. https://doi.org/10.1007/978-3-658-06173- 9 Vieweg, S., Hughes, A. L., Starbird, K., & Palen, L. (2010). Microblogging During Two Natural Hazards Events: What Twitter May Contribute to Situational Awareness. In Proceedings of the Conference on Human Factors in Computing Systems (CHI) (pp. 1079–1088). Atlanta, USA: ACM. Volgger, S., Walch, S., Kumnig, M., & Penz, B. (2006). Kommunikation vor, während und nach der Krise - Leitfaden für Kommunikationsmanagement anhand der Erfahrungen des Hochwasserereignisses Tirol 2005. Innsbruck: Amt der Tiroler Landesregierung, Abteilung Öffentlichkeitsarbeit in Zusammenarbeit mit SVWP Kommunikationsmanagement und dem Management Center Innsbruck (MCI). Wade, J. (2012). Using mobile apps in disasters. Risk Management, 59(9), 6–8.