Keywords

1 Introduction

The earthquake in Haiti 2010 can be seen as a starting point of digital volunteering in the context of disaster management and humanitarian assistance. The first digital volunteers were committed to helping affected populations by processing and providing Volunteered Geographic Information (VGI) for disaster response. Over time, they organized themselves and formed virtual groups. These emerging so-called Volunteer and Technical Communities (V&TC) opened the view for multiple new fields of research. Many research approaches paid attention to technical topics like data mining and creating better analytical tools for the volunteers. In contrast, less research has been conducted to determine which motivational factors drive people to volunteer digitally in disaster management and which organizational requirements exist to collaborate with established Emergency Management Agencies (EMA). Furthermore, questions about the impact of analytical results in this time-critical context arise. The underlying research focused on the volunteers themselves and organizational requirements for collaboration, more specifically their motivational, participative, and analytical factors. This chapter is based on research carried out in the project “Active Participation and Motivation of Professionalized Digital Volunteer Communities: Distributed Decision-Making and its Impact on Disaster Management Organizations.”

After this brief introduction, the second section of this chapter presents a comprehensive study of the motivational factors of operationally active digital volunteers (Fathi and Fiedrich 2020). In a cross-organizational online survey, possible motives, individual organizational commitment, and potential incentive options were analyzed. In addition, two experienced digital team leaders of V&TC were interviewed in guided expert interviews about methods and measures for increasing motivational and organizational commitment. Based on the findings generated in this way, explanatory patterns for the motivation factors of digital volunteers can be derived on the one hand; on the other hand, beneficial and identification-generating measures can be identified.

Contrary to the V&TC, digital volunteers institutionalized Virtual Operations Support Teams (VOST), which are closely linked to established EMA. This development led to more in-depth research questions concerning the professionalized digital volunteers and their integration in decision-making processes using VGI in Emergency Operations Centers (EOC), which are discussed in Sects. 13.3 and 13.4. The organizational structure and technical requirements for succeeding and their decision-making processes in a time-critical environment were of interest. A research gap between the VGI created by VOST and decision-makers needs in the established EMA was acknowledged, as the digital volunteers started to collaborate with EMA. Therefore, the topic of voluntary digital participation for collaborative emergency management was explored in collaboration with the University of Stuttgart, whose research efforts mainly focused on visual analysis of VGI. In Sect. 13.2, we present a case study which was conducted with the project “VA4VGI-2” (Chap. 6), where structural, procedural, and technical requirements of integrating VOST in EOC structures were investigated, applying a mixed-method approach (Fathi et al. 2020).

Overall, little attention was paid to VOST, who act as groups of data analysts with direct integration to EOC. Specific tasks of VOST include filtering, verifying, and analyzing social media data from various platforms and creating information products for decision-makers in EOC. These information products can contribute to the situational awareness of EOC members and to the decision-making processes by integrating actionable information. In a case study following the 2021 flooding in Germany, the aspects of analyzing social media by digital volunteers in VOST and the impact of the information products on situational awareness and decision-making were examined and are presented in Section 4 (Fathi and Fiedrich 2022).

Analytical and decision-making processes in the time-critical environment of EOC in disaster management are challenging due to the numerous disaster-related conditions like time-pressure and uncertainty. To examine the interplay of the conditions in disaster management, VGI, and biases in Crisis Information Management (CIM), a workshop experiment was conducted with digital volunteers and decision-makers (Paulus et al. 2022). A three-stage experiment on epidemic response was developed as the underlying scenario to analyze how biases can be mitigated by observing digital volunteers and decision-makers in the analytical and decision-making processes and is presented in Sect. 13.5. The findings of this case study suggest that debiasing efforts are strongly undervalued, and external analysts fail to debias data successfully in favor of rapid results. The biased data was then passed on to decision-makers in the form of information products, who make decisions based on biased data.

Section 13.6 addresses the challenge of privacy-aware data analytics in disaster management and discusses a collaborative work between this underlying project and “Privacy Aspects” (Chap. 14).

2 Motivational Factors of Digital Volunteers in Disaster Management

With the emergence of social media and VGI, various new research areas and new opportunities to use this open-access data increased, also in disaster management. Anyhow, limited resources characterize disaster, and EMA do not have enough staff to analyze big amounts of data to integrate VGI in their situational awareness and decision-making processes. Digital volunteers analyze social media data from, e.g., Twitter and proceed disaster-related VGI. The volunteers can work dislocated from the actual operational site and thus can be deployed almost instantly. Over time, the volunteers organized themselves and formed V&TC. These communities fostered research interest, especially in the fields of organization and technology. An existing research gap in the context of volunteering on a digital basis in disaster management are the motivational factors of V&TC members. Questions regarding the motivation of the digital volunteer, the barriers to participation, and the commitment to the V&TC have not yet been extensively answered. In the paper “Digital Volunteers in Disaster Management—Motivational Factors and Barriers of Participation” (Fathi and Fiedrich 2020), we aimed to understand what motivates the digital volunteers and which incentive options there are to further motivate them. Additionally, we looked into measures and methods that can be implemented by digital team leaders to motivate their team members. Lastly, the differences and correlations between the needs of the digital volunteers and the motives of the digital team leaders were examined.

Therefore, we used a mixed-methods approach of quantitative and qualitative social science methods. It was of special interest to capture and query the digital volunteers in their social contexts as well as their individuality. An online survey was conducted among different V&TC, to explore motivational factors and incentive options. The survey was designed under the use of the Volunteer Functions Inventory (VFI) introduced by Clary and Snyder (1999), which is a widely used questionnaire on volunteer motivation. In order to understand measures and methods to foster motivation of digital volunteers by leaders of V&TC, guideline-based expert interviews were carried out. The guidelines were designed based on the online survey, but the content was transferred to the perspective of team leaders.

It was found that digital volunteers are mostly motivated by their values, but also by the experience, they are gaining paired with fun-based intrinsic motivation. In contrast, having the prospect of a “career” within V&TC was less motivating. The participants strongly agreed to statements of organizational commitment and identification with their organization. More than 70% stated that they fully agreed to be proud to be part of their V&TC. The main barriers of digital volunteering were named as time, trust in one’s own abilities, and Internet access. Especially during crises, time allocation becomes a challenge. Collaborating with other V&TC or a pool of digital volunteers, who can be acquired ad hoc, seems to be an appropriate method to allocate work of digital volunteers. The queried digital volunteers see potential for motivational enhancement rather in non-crisis times, for example, more feedback and additional online and offline community activities without a disaster context. Accordingly, the volunteers see incentive options in digital or analog exercises or events and appreciative measures. It became clear that feedback is very important to the digital volunteers, who, due to their dislocation, can only guess what impact and use the resulting information products and VGI have. Negative impact on the motivation of digital volunteers were identified as a lack of feedback and a lack of identification with the work or the V&TC. Feedback cannot only be given by the tasking EMA but also by the public or the digital team leaders (Fathi and Fiedrich 2020).

3 Virtual Operations Support Teams in Disaster Management

Virtual Operations Support Teams (VOST) are groups of professionalized digital volunteers, who are closely linked to EMA. During a VOST operation, the common tasks include monitoring and analyzing social media, verifying and geolocating information and developing crisis maps, recognizing and analyzing trends and sentiment in social media, and ad hoc tasks as assigned by the EOC. The actionable information identified and verified can then be provided to the EOC decision-maker to expand situational awareness and support decision-making processes. Furthermore, VOST integrate a liaison officer in the EOC structures. This enables to ensure effective communication and distribution of tasks between VOST and EOC. VOST pursue the goal of effectively integrating information products and VGI into decision-making processes through close organizational integration in the time-critical context of disaster management. This in turn led to multiple research questions concerning structural, procedural, and technical requirements for an effective collaboration between a virtual team of analysts and decision-maker in an EOC. These questions were to be explored in collaboration with the project “VA4VGI-2” (Chap. 6). As described in the paper “VOST: A case study in voluntary digital participation for collaborative emergency management” (Fathi and Fiedrich 2020), the main goal was to understand the decision-making processes, which emerged by integrating a VOST into the structures of an EOC. An exploratory case study was conducted as field research during the start (Grand Départ) of the Tour de France in Düsseldorf, Germany in 2017. Especially of interest were the requirements of structure and procedure for a successful collaboration, technical requirements and the evaluation of existing technical tools for social media analytics, the identification of the actual tasks that needed performing during the operation, and structural, organizational, and technical implications for future decision-making systems in EOC.

The VOST operation at the Grand Départ in Düsseldorf was in the scope of a pilot project with the German Federal Agency for Technical Relief (“Technisches Hilfswerk” – THW). The THW VOST consists of 20 digital volunteers which were appointed as THW members and thus act as team members in a governmental EMA rather than a loosely coupled group of digital volunteers in V&TC. To make use of the insights provided by the case study, multiple methods for data collection and analysis were conducted. These methods comprised participant observation during the two-day operation, focus group discussions and informal interviews with decision-makers and VOST members at different stages of the operation, analysis of the tasks performed by the VOST during the operation, analysis of the organizational setup, and technology use and decision-making processes of the VOST. For this particular operation, the following VOST working priorities were identified as a result of the focus group discussions: identification of critical crowd densities and flows; detection of unusual events; image analysis of social media; developing of a crisis map for spatial analysis; identification of false information, rumors, and fake news; and scenario-dependent tasks.

The structural and procedural requirements for a successful collaboration were identified as a division in small VOST working groups to simplify the distribution of tasks. This allows specialized subgroups to be formed to respond to dynamic operational situations, e.g., verification groups, and to ensure information exchange on the level of team and group leaders, individual group briefings to implement adjustments quicker, and, most importantly, the implementation of a liaison officer. The technical requirements are especially a reliable user experience and custom-tailored tools for the use during an operation to alleviate the high mental workload. Putting new tools to the test in real-world operations seems to be a beneficial way to ensure advanced algorithmic tools. The most time-consuming and mentally challenging work at the same time poses collection, filtering, and documentation of user-generated content from social media platforms. Advanced mining tools are crucial to verify social media data in a time-critical environment. Situation monitoring is a highly repetitive and demanding task, which can only be carried out by a digital volunteer for a certain amount of time. Nonetheless, the biggest challenge seems to be the velocity and the volume of user-generated social media data. Additional disaster-related data becomes available all the time during the analysis, which presents a challenge for real-time social media analytics. To address the questions of what disaster-related information is processed by a VOST during a disaster management and what impact it has on members of an EOC, another case study was conducted.

4 Social Media Analytics by Virtual Operations Support Teams in Disaster Management

Climate change poses numerous challenges and risks, including a significant increase in extreme weather events such as flooding (IPCC 2021). With this, the need for crisis communication and social media analytics in times of disaster rises accordingly. VOST address this need, integrated into EOC structures in times of crises, and their goals are to increase the situational awareness of decision-makers and to provide actionable information to improve decision-making in a time-critical environment. To examine these efforts, a case study was carried out, using the data collected by 22 VOST analysts during the 2021 flood in Wuppertal, Germany (Fathi and Fiedrich 2022). The city was severely flooded in July 2021; parts of the city had to be evacuated, and warning sirens were set off (Zander 2021). The EOC operated in cooperation with the VOST of the German Federal Agency for Technical Relief (THW VOST), which was deployed virtually but was directly connected to the EOC through a liaison officer, who was physically present in the EOC. This operation thus raised the research question, how VOST can support situational awareness and generate actionable information for EOC decision-making processes by integrating social media analytics practices. The research question was explored by analyzing the data generated during the THW VOST operation and by a survey among EOC decision-makers on the impact of the information provided by VOST on their decisions and situational awareness.

Unlike other research, which focuses, e.g., mainly on social media big data analysis, decision-making processes, or developing machine learning approaches, this study aims for examining closely the real-world VOST integration. Case studies, such as the underlying, can provide valuable insights of virtual teams in the disaster management context. In order to analyze the data generated by the digital volunteers during the operation, the following tasks were performed: data cleaning, summarizing categories, visualization of the data, and subsequently a comparative quantitative analysis and contextualization of the data. Furthermore, three different parameters, the format, the source, and the mean value of the prioritization, were used for an in-depth analysis. The survey of decision-makers was conducted among the EOC members, which collaborated with the THW VOST and used their information products during the flood response in Wuppertal. The prerequisite was that the interviewee had worked with VOST information products during the operation. Nine decision-makers from the EOC met these criteria, and all of them participated in the survey.

To classify information categories in the VOST dataset, which was identified by VOST volunteers during the flood response, 536 social media posts from eight different social media platforms were analyzed. Additionally, 42 posts from websites (e.g., traditional media) were collected. The dataset was classified in 23 different categories. The largest category was found to have emerged after the flood, namely, spontaneous community engagement (see Fig. 13.1). The earlier phases of the flood were dominated by categories like level of the river, warning, or flooded roads.

Fig. 13.1
A bar chart plots categories versus percentages. The highest percentage is of spontaneous community engagement at 11.9%, followed by the level of the river at 10.3%. The lowest percentage is of counterstatement false information at 0.2%.

Percentage distribution of social media information during the flood response 2021. (Source: Fathi and Fiedrich 2022)

Full size image

Posts from categories that could have had a direct impact were forwarded by VOST during the operation as actionable information to the EOC decision-maker. These social media posts were prioritized as “highly-relevant.” The information in the format of videos was found to have a higher priority than information in the form of texts or images. In a category analysis over time, it was found that real-time disaster events, such as the activation of the warning siren, are simultaneously apparent in social media data.

VOST impact on situational awareness was explored by querying EOC directors and executives who collaborated with the THW VOST during the flood. All statements in the survey were rated with a strong overall agreement. The statement with the highest degree of agreement was: “Information from VOST contributes to expanded situational awareness.” The necessity of a liaison officer in the EOC was also strongly agreed to. The lowest level of agreement was given the statement that VOST information can forecast developments of future situations. The second category of statements concerned the VOST impact on decision-making, e.g., to ensure people-centered risk and crisis communications and contributed to confidence in decision-making.

The results underline that situational awareness and decision-making is supported by VOST information and VGI on three different levels: perception, comprehension, and projection. Based on this distinction of situational awareness from Endsley (1988), it can be concluded that statements that can be assigned to the first two levels (perception and comprehension) receive high agreement. However, VOST information during the dynamic flood situation supports less to project the future flood-situation in a more precise way. Statements in this category received less agreement. This condition can be explained by the observation that VOST information and VGI provide new and complementary information that must be processed, comprehended, and projected by EOC decision-makers onto future scenarios in addition to that provided by other sources (feedback from responders, emergency calls, etc.). Due to the accompanying conditions in disaster management (e.g., time pressure, uncertainty), the cognitive load on decision-makers is very high so that the use of VGI can create biases in data analysis and decision-making.

5 Data Bias in Crisis Information Management

Due to various conditions, humanitarian crises and disaster management especially challenge digital volunteers’ data analysis. Crisis Information Management (CIM) is characterized, e.g., by time criticality and uncertainty. Additionally, resources are limited, and the cognitive load is high. This makes analysts and decision-makers prone to inducing biases in the data and cognitive processes. When undetected, biases remain untreated and lead to decisions based on biased information, which in turn can lead to an inefficient response. To find out more about the interplay of data and cognitive biases, an exploratory three-stage experiment on epidemic response was conducted (Paulus et al. 2022). A scenario-based workshop was held in The Hague in 2020 with experienced crisis decision-makers and digital volunteers from various V&TC and VOST, which entailed stage 1 and 2. For stage 3, the same participants were additionally addressed in an online survey. The experimental scenario was an epidemic outbreak in three countries. The first stage included an observation of digital volunteers, who were provided with different datasets with biased data, e.g., in the infection spread. The observation was set out to be in a fictional but realistic setting to avoid interference in a real epidemic response. In stage 2 of the experiment, the decision-makers were provided with the VGI and information products (e.g., maps) created in stage 1 and had to make decisions on treatment center placements. Stage 3 of the experiment was an online survey onsite of the workshop. It aimed to explore whether confirmation bias leads to path dependencies of former decisions based on biased information. In the survey, they were able to select the information they viewed as most important for future decision-making from a list of datasets.

The results of the experiment show that in the first stage, the participants failed to debias data, even though biases were detected. Debiasing efforts were undervalued in favor of immediate results. The information products created based on biased data in stage 1 were then forwarded to the decision-makers in stage 2, who made their decisions based on biased information. Even though the decision-makers in all three groups put enormous pressure on the digital volunteers to find out on which datasets the information products were generated, they did not succeed in ensuring that the decisions were not based on biased information. Confirmation bias was detected in stage 3; the reliance on conclusion reached with biased data was reinforced by it. Thus, biased assumptions remained undetected. The main causes for biased data remaining untreated are the described conditions of data analysis and decision-making in the context of disaster management. The realistic scenario design made it possible to recreate the general conditions, e.g., by simulating time pressure and uncertainty. Mindfulness debiasing efforts have been found to be effective to counteract these conditions and therefore pose a promising strategy to mitigate data and cognitive biases in future disaster management (Paulus et al. 2022).

6 Privacy-Aware Social Media Data Processing in Disaster Management

Social media analytics by Emergency Management Agencies for relief purposes has become a common practice in the last few years. In the time-critical environment of disaster and life-threatening situations, privacy is often perceived as a secondary problem. Nonetheless, avoiding unnecessary data retention is important to protect the social media users’ privacy by, e.g., preventing subsequent abuse. In times of crisis, social media users are especially vulnerable, e.g., by sharing names, addresses, or other personal information, for example, searching for missing relatives or friends.

In a cross-project effort, expertise from the present project and “Privacy Aspects” (Chap. 14) were combined for a joint case study (Löchner et al. 2020). The study examined the extent to which VOST can integrate privacy-aware methods and algorithms, in particular HyperLogLog (HLL), in their operational work in disaster management. To investigate the practicality of privacy-aware methods and HLL, a case study with digital volunteers from two VOST has been conducted. For this, a focus group discussion addressing opportunities, challenges, and implementation barriers was held. The focus group discussion with participants from THW VOST and VOST Baden-Württemberg aimed to document the expertise of the participants, who are experienced in the field of social media analytics.

The most important finding was that the focus group discussion revealed no disadvantage against using privacy-friendly methods and HLL for VOST. The algorithm does not distract the data analysis process, since the VOST work starts after the data processing via HLL. Overall, HLL was found to be an appropriate technology to ensure privacy-aware social media data processing. Opposing to the initial assumptions that HLL use might be conflicting with gathering data for creating information products, several benefits of the algorithm were come up upon. These benefits include improved working with big datasets, which might lead to a more widespread use of HLL and thus improved privacy-awareness among digital volunteers.

7 Conclusion and Outlook

In this chapter, various facets of digital volunteering using and processing VGI in the disaster management sector were highlighted. In particular, social, organizational and analytical factors were examined and discussed in five different papers. It was shown that the motivation of V&TC is particularly value-based and that the commitment to the virtual team is pronounced. However, measures can be derived, especially for future developments, in order to sustainably establish digital volunteering. Professionalized teams in the structures of established EMA have been institutionalized worldwide, but detailed research on the requirements for collaboration between a virtual team and an EOC has been lacking. In Sect. 13.3, a joint study with the project VA4VGI-2 (Chap. 6) was presented. From the results, it can be concluded that an effective integration of digital volunteers and their analytical skills into established disaster management structures is achievable, although specific requirements have to be considered.

The analytical value for situational awareness and decision-making of EOC members during a specific disaster management situation could be highlighted in Sect. 13.4. In conclusion, the section shows that VOST analysts were able to identify, verify, and categorize a large amount of disaster-related information from eight different social media platforms in various formats. During the 2021 flood, this information contributed to an expanded situational awareness of the decision-makers in an Emergency Operation Center and made it possible, for example, to conduct crisis communication in a more people-centered manner. Thus, it could be shown that during dynamic disaster situations, the involvement of digital volunteers was of major relevance for the operation management. Furthermore, it could be shown that the virtual structures of a VOST are able to effectively support an EOC even in an acute disaster situation. Nevertheless, analysts and decision-makers in such situations are accompanied by conditions that can bias results and decisions. In order to investigate data and cognitive biases in the work of digital volunteers and decision-makers with VGI, a paper was presented in Chap. 5 in which this aspect was investigated in a three-level experiment. It could be shown that debiasing efforts were not pronounced enough, and thus biased information was considered in decision-making. It can be deduced that in future exercises and trainings, debiasing efforts need to receive more attention in order to ensure the integration of digital volunteers and VGI in the future. This is also accompanied by privacy-aware analytics of social media in disaster situations, where the affected population is particularly vulnerable. To investigate possible methods and an algorithm developed in the project “Privacy Aspects” (Chap. 14) in the use of VOST, a case study was conducted with two VOST (Sect. 13.6). Thereby, it could be examined that the use of privacy-aware methods and algorithms in operational use is reasonable and possible.

Overall, it could be shown that digital volunteers make a significant contribution during disaster management, in which they effectively process their analytical results and VGI for the management of disaster situations. However, human limitations and privacy-aware methods need to receive greater attention in the future, both in research and in practice. In addition, questions remain about how situational pictures will need to be designed by digital volunteers in the future. In a project funded by the “German Federal Office of Civil Protection and Disaster Assistance” called “#sosmap” (Fiedrich 2022), it will be investigated from August 2022 to what extent psychosocial situation pictures can be created for EOC by analyzing social media.