Should everyone in need be treated equally? A European survey of expert judgment on social vulnerability to floods and pandemics to validate multi-hazard vulnerability factors

Several European countries were affected by severe floods in 2021. At the same time, despite the deployment of vaccines, Europe was the COVID-19 pandemic's epicenter several times during 2021. One research aim of this study is to identify socio-demographic groups vulnerable to floods and whether the groups vulnerable to floods and pandemics overlap or are disjoint. We ran a survey in four languages (English, French, German, and Spanish) and collected the judgment of 366 experts in disaster risk management and first-responders to find out how those people caring for “people in need” (be it operational or administratively) think about which persons are more at risk than others. Another research aim is to validate multi-hazard vulnerability factors by comparing judgment on groups vulnerable to the COVID-19 pandemic and to floods. The main findings are that experts think that socially vulnerable groups should be rescued or treated first. Treating everyone equally is less favored by comparison. Infrastructure losses, followed by economic losses, reveal better than deaths or psychological issues whether vulnerability played a role in a disaster. Regarding vulnerability characteristics, older, homeless people, and immigrants rank highest, and most factors can be used to explain both flood and COVID-19 vulnerability, while some differ; for example, mobility impairment is less important for COVID-19. There are major discrepancies between what respondents think should be done to prioritize help to certain groups and what they have experienced is being done on the field.


Introduction
Several European countries were affected by severe floods in the summer of 2021, including flash floods in the United Kingdom, Belgium, Germany, the Netherlands, and Switzerland [1]. Pluvial and riverine floods affected many areas worldwide in 2021; in Europe, more than 200 people died in the floods [2]. Extreme weather events with heavy rainfall and flood damages are expected to continue to rise due to climate change [3]. At the same time, despite the deployment of vaccines, Europe was the COVID-19 pandemic's epicenter several times in 2021 [4]. With the pandemic, some socio-demographic groups such as the elderly or those with pre-existing health conditions have come into focus as 'risk groups' to be prioritized due to their health-related susceptibility [5][6][7]. Such vulnerable groups have been the core focus of the pandemic management, with preferential measures such as early vaccination and targeted policies, such as partial lockdowns. Studies explicitly focusing on vulnerable groups have also focused on migrants, unemployment, psychological and other impacts [8][9][10].
A. Fekete and S. Rufat Vulnerability is defined as "The conditions determined by physical, social, economic and environmental factors or processes which increase the susceptibility of an individual, a community, assets or systems to the impacts of hazards." [11]. Complementary, sometimes alternative concepts to vulnerability in a disaster context are capacity or resilience [12,13]. Vulnerability is a key explanatory dimension of disaster risk [14]. Common methods for analyzing social vulnerability are indicators or interviews/surveys [15]. Since the 1980s, a growing number of studies have successfully used such approaches [16,17]. The limitations of these approaches have also been debated, especially in social vulnerability indices and spatial assessments [18]. These limitations include data accessibility, scale, and resolution, representation of the people, and social contexts. However, both semi-quantitative as well as qualitative approaches are based on conceptual notions of vulnerability characteristics. Whether interviews or statistics capture these characteristics, they are all based and limited to the original question items of who is vulnerable and why.
Meanwhile, social vulnerability indicator assessments have become common research [15,19]. The disaster risk reduction policy also has adopted the usage of vulnerability measurements, for example, in reporting to the Sendai framework of the United Nations [20,21]. However, there is still a paucity of understanding of what researchers actually have in mind when analyzing social vulnerability. While social vulnerability assessment has been established as its line of research, it seems to have lost the connection to its origins emerging from the field of aid and needs research [22][23][24][25] or from the context of shelter and preparedness planning [26,27]. This study, therefore, asks specific research questions to researchers and experts in the field of disaster risk. The questions should enlighten the understanding of the basic connection between vulnerability and the needs of people. We assume that it is too easily taken for granted that social vulnerability identification is carried out in disaster assessments following real events. We investigate this by analyzing the connection between recognition of social vulnerability on the one hand and actual observation of help received in aid operations. While this study, therefore, analyzes social vulnerability by its general understanding, it can contribute to a better understanding of social vulnerability assessments in disaster risk reduction for multiple hazards and impacts. For example, when everyone seems to agree that social vulnerability is important, but it is not operationalized or observed in real cases, the notion and usage of social vulnerability should be questioned. This study then selects two distinct hazards that play a certain prominence and, therefore, recognition amongst multiple groups of researchers worldwide to enable such a broad study. Floods are selected as one type of hazard since they are one of the most common natural hazards in many areas of the world [28,29]. The other hazard selected is COVID-19 since it is also currently observed worldwide and offers a substantial difference to the nature of floods as a natural hazard. The pandemic mainly affects people through human interaction in spreading the disease, while floods affect people mainly due to location exposure close to a water body.
For policy-making and disaster risk reduction, this study can also help to allocate resources better and foster institutional monitoring of social vulnerability when a clearer connection to the needs of people is identified. Policy-making, as well as the science-policynexus, is currently in demand and explaining how to address better climate change-induced extreme events, including floods [21]. But with the COVID-19 pandemic especially, additional importance on disaster risk reduction has created a new awareness about special needs groups in disasters [5,8,10,30,31]. In this pandemic, in many areas of the world, it has been observed that socially deprived or marginalized groups were affected more and had fewer resources to recover from the impacts of the pandemic.
The rationale of the study is additionally embedded in our experience as researchers in the field conducting social vulnerability assessments. We have observed challenges in applying social vulnerability indicator assessments regarding technical and institutional issues and user needs [32,33]. We also have identified a paucity of validation approaches in social vulnerability assessments [34,35]. Many social vulnerability indicator assessments seem to follow methodology lines that had once been established by certain research groups or are currently used by renowned institutions such as the Center for Disease Control in the USA. While using those methods and approaches is fine, we believe that more validation is needed in terms of understanding the nature of social vulnerability and recognizing the different perceptions and, therefore, notions that may exist by different researchers in different parts of the world. Having floods and pandemics happening concurrently in many parts of the world simultaneously at the moment allows us to better juxtapose the notions of social vulnerability by comparing them to each other for two different hazards. This can elicit underlying confirmatory or contradicting notions of researchers that can be helpful in understanding the true nature of social vulnerability better. Understanding risk and vulnerability as a core component is important, as expressed in the Sendai framework [21]. But to better enact and use the social vulnerability to improve the conditions of people, a connection to understanding peoples' needs is also crucial. Many researchers attempt this by emphasizing more local studies that capture people's local context and needs [36,37]. We assume it would be beneficial to add more context notions and connections between social vulnerability understanding and needs perceived and observed on the one hand.
Therefore, this study wants to determine what experts in the field of vulnerability and hazards research and practice consider valuable items in identifying "the vulnerable". An online survey collects opinions on how experts researching and caring for 'people in need' think about which persons are more vulnerable than others. The core research interest of our study is: do people think that vulnerable groups should be prioritized? If so, based on which criteria? To better understand this question, we investigate the perceptions of researchers and experts regarding specific survey questions around the main research question: 'Is it possible to differentiate and prioritize vulnerable groups and/or should everyone be treated equally?'

State of the art
Using specified search terms, we conduct a literature analysis of the latest studies on social vulnerability of floods in Europe and the COVID-19 pandemic in Scopus, ScienceDirect, and Google Scholar. In addition, seminal studies and original papers are selected based on the authors' experience.

Vulnerability and needs groups
Social vulnerability as a field emerged out of different contexts and origins. One stream often cited is a reaction towards a predominance of quantitative and technically oriented research, for example, in flood hazard and disaster management research with a prevailing focus on hazard and not on the impact side in a more nuanced way [36,37]. Another often related stream is research influenced by a politicized cultural environment in the 1960-1970s, which transformed research theories and selection of methods specifically in risk and disaster studies and came from broader fields such as political ecology and many others. A range of researchers is now imbued with a certain political perspective, also influenced by post-colonization theories and many more. This can help to understand how people-centered approaches and focus on the diversity of human beings became more prominent, alongside also new security and environmental views, for instance, human security [38][39][40], influenced by social, economic, and political inequality [41] or environmental sustainability awareness [42]. Another stream of research follows more quantitative methods such as statistics, indicators, and semi-quantitative surveys, and often irritations exist between these and the previously mentioned more qualitative research streams [43].
Disaster aid recipients' needs are identified mostly within international development aid practice. As an example, the work of Frederick Cuny is cited in social vulnerability studies [22], who wrote a book based on 30 years of practical first-hand experience as an international disaster relief specialist [23,44]. The basic argument he produces is that "aid should support local coping mechanisms (internal social structures used to adapt and respond to a crisis) instead of rapid delivery of large quantities of supplies" [44]. The context of disasters and social vulnerability is established here by linking the poorest populations with potential improvements within the process of aid distribution. Cuny also suggests to rather focus on the process of aid delivery rather than the simple delivery of goods.
The topic of needs points mainly to external help, which is sometimes a discussion topic within vulnerability literature, whether vulnerability only captures internal capabilities or aspects of people or external dependencies [45,46]. However, even when vulnerability is defined as a purely internal aspect of people or other subjects of research, the connection to needs still exists. Needs can be internal or external as well, and studies also reflect the struggle in separating external or internal factors or capacities from susceptibilities which also often intermix internal and external factors [47]. Needs assessments, especially post-disaster needs assessments, are a common approach [48] but are not commonly connected to social vulnerability assessments, except for more practically oriented or qualitative studies [49,50].
'Natural disasters and Vulnerability Analysis' by the United Nations in the 1980s summarises experience with composite vulnerability analysis carried out in the 1970s on a range of natural hazards [51]. The term "vulnerability analysis" by UNDRO is used for a "method of risk identification and evaluation" [51], which from a modern perspective resembles a probabilistic hazard and damage model. Struggles with defining vulnerability are documented in this document, too. Indeed, the main part of the document is about hazard parameters or disaster management, but it is void of differentiation of people or people's capabilities and susceptibilities.
"VULNERABILITY meaning the degree of loss to a given element at risk or set of such elements resulting from the occurrence of a natural phenomenon of a given magnitude and expressed on a scale from (no damage) to 1 (total loss)." Anderson differentiates material, organizational and socio-psychological vulnerability, including 'people's capacities to cope' [52]. The importance of investigating the psychological aspects of disaster was brought to attention by the 'Conference on Field Studies of Reactions to Disasters', held at the University of Chicago, January 29-30, 1952 [53], which was influenced by chemical warfare experiences and research. Terms such as vulnerability, resilience, or differentiated capacities of people are not documented verbatim, but the relation to socio-psychological vulnerability of community resilience can be compared.
"Social-psychological research into disasters should be oriented toward the practical objectives of understanding how people respond to disasters in order to be able to supply information as to the probable effect of a disaster on the efficiency of functioning of a community and its inhabitants and as to means of minimizing any impairment of functioning that may occur." In "Assessment of Victim Needs", Alan Taylor reviews the "physical and cultural distance" and "the difficulties which rich-world relief organizations encounter in attempting to assess the needs of disaster victims in poorer countries of the world." [25] He asserts that "Assessing and interpreting need are steps prerequisite to any relief operation." ( [25]: 2) and provides the following caveats: "Yet these processes are ill-defined and in practice often haphazard. Sometimes need is described as a simple lack of a commodity or facility thought to be required by victimsfood, housing, medical care, etc. But more usually the need is expressed in terms of what the defining agency believes it can do in response to the situation." Many more of his observations seem to be valid today, too, and this can set a good background for investigations in this article: "In any case, the needs which each relief organization understands and is willing to identify depend very much on the history and tradition of that particular agency." [45]. Taylor also perpetuates a community-centered perspective that argues against the need for external help: "Within the family unit, all sorts of help are available." (…) "Only if all of these services of assistance fail will impersonal, formal organizations be resorted to." Taylor also warns that foreign aid should be aware of family patronage systems hindering direct access to individuals but rather having to tend towards a household head. It is stated as a reality to accept. However, reflection about limitations of family or local systems or further advice on how to differentiate the needs of people are missing or stop here in the plea about external door aid shortcomings.
Concluding, early studies on social vulnerability already show a broader picture of social vulnerability as currently used in many social vulnerability assessments. Moreover, a connection between needs assessments and social groups or social traits can be identi-fied from the background of humanitarian aid studies which largely influence social vulnerability conceptualizations. However, connections between analyzing social vulnerability and needs assessments appear to be more theoretical, and fewer studies are operationalizing it recently.

Social vulnerability indicators and questionnaire items
Research on social vulnerability has matured, which is reflected by over one hundred review articles with the term' social vulnerability' in their title. Adding the term 'survey' into the title word search, only one scientific poster abstract is found. In this source, survey data is reviewed and tested in an assessment of tornado impacts on residents of mobile homes. Typical social vulnerability attributes are used, such as gender, housing ownership, ethnicity, education, family, employment, and age. Findings from an online survey are statistically analyzed using ANOVA. The paper discusses the correlation of variables and how well they can represent or explain vulnerability. This study [54] reflects a good example of a typical semi-quantitative social vulnerability assessment.
Adding the term 'interview' into the title search instead, revealed no direct hits. When adding the term indicator into the title search, the number of findings sinks to only two papers, and none can be found when adding the term index instead. Using the term 'assessment' in the title search, five papers were found, but only three were accessible, and excluding our own papers, it is two papers. A study analyzes climate change-related topics such as heat and flooding and investigates the relation of vulnerability to resilience [55]. It mainly conducts a literature review on the common consent about vulnerable groups and their characteristics to design focus group interviews and construct an index. It provides a good overview of how many studies use certain categories such as "the elderly" and also cites critique on such categories such as "false positives" when not all older persons are vulnerable [56]. The article thereby raises the question of which variables represent vulnerability. Indirectly this calls for more analyzes to justify the use of common variables in social vulnerability assessments. In some studies, limitations of aggregated indices of social vulnerability are addressed [18,[56][57][58][59]. This further underlines the need for a better understanding of the nature and usage of such indicators.
In many studies, a literature review is used as a justification basis to conduct case study assessments, for example, recently for landslides in Malaysia [60]. Recurringly, this justification is based mainly on the experience of other studies following a similar approach. While this is common in science, it lacks an alternative investigation of how and which variables can or should be used [61].
Approaches to better balance shortcomings of indicator and index approaches analyze, for instance, proxy indicators that serve as "greatest common variables" and can represent many other variables [57]. This avoids statistical confounding and also saves costs [62]. In some areas, such as lifeline vulnerability, broadening the number of variables but also developing new sets of vulnerability is necessary [63]. Other studies investigate possibilities to analyze uncertainty in the data and methods and ways to evaluate and validate the methods and results. This includes validation by user communities [15,64], statistical sensitivity analyses [58,65], or testing of results against a second independent data set [34,35], or by triangulation of methods [56,66,67].
There is a convergence of variables and question items used to identify social vulnerability based on people's characteristics. Limitations of such approaches are recognized, and methods are applied to validate and test such assessments. However, the validation approaches are still few by comparison. Furthermore, studies using independent additional data sets are rare, too. This indicates a need to investigate the usefulness of social vulnerability question items or variables through additional studies and alternative ways.

Social vulnerability studies on floods, pandemics, or both
Numerous studies have internationally analyzed social vulnerability in context to one hazard, such as COVID-19 recently, for example, in the USA [68]. Social vulnerability studies are also increasingly common for analyzing floods in recent years. There are few studies analyzing both floods and the COVID-19 pandemic internationally. No paper was found by a title word search covering both floods and pandemics or COVID. Many studies mention floods and the background of studies on this hazard in recent pandemic studies on social vulnerability. However, excluding "social", some studies are found, on hurricane evacuation, for example [69]. More studies seem to emerge around the term cascading effects, compounding event or hazard, multi-hazards or concurrent hazards, that analyze interrelations of pandemics with floods or hurricanes [70][71][72][73]. Systemic risk is another upcoming topic [74], but no papers were found with a direct connection to the title word terms of vulnerability or floods and pandemics.
However, studies are trying to define vulnerability in disasters of various types sensitive to place-based contexts or situationspecific vulnerabilities [22,36,37,75]. Overall, these studies are few compared to social vulnerability assessments for either floods or pandemics.
Our concluding presumptions from this section are that needs assessments are one driver of social vulnerability assessments, but connections are missing in both research and integration of methods such as social vulnerability questionnaires or indicators. The question of needs relates to questions of differential needs and to which people are to prioritized. A second conclusion is the persisting gap of validation studies within social vulnerability assessments. It seems to be important to contribute to this field not only by more validation studies per se, but exploring also new approaches. And finally, there appear to exist many single hazard studies but few combining social vulnerability to e.g. floods and pandemics.

Method
The survey centers around the question, which people are more vulnerable to hazard impacts such as river floods or pandemics. The study centers on Europe, but the sample includes international participation of survey respondents, too, to enable comparisons of whether the European experiences fall out of place or are similar. This can help better embed and compare the study later in follow-up studies. The survey respondents were recruited based on a systematic literature review of key authors publishing social vulnerability assessments in the past decades. Using keyword search in different databases, individual names and email addresses have been ex-tracted from published and accessible peer-reviewed journal papers. In addition, experts and researchers on social vulnerability have been selected based on their visibility internationally and from past conferences and networks of the authors. Experts in the field were addressed by individual email and asked to forward it, too. Expert names were collected over the past years by the authors, and an additional 489 authors were identified that have published on social vulnerability and relevant approaches for this study in peerreviewed journals by a systematic search of individual articles. But also, practitioners were addressed. In addition to personal contacts, the survey was distributed in the aftermath of the floods in Europe 2021 to operational forces, too. The respondents were also encouraged to distribute the survey further, and the survey was openly available online.
The survey themes were selected based on the overall rationale of the research as outlined in the introduction. Specific themes derive from the main streams in social vulnerability research. The first specific theme of prioritization in context to hazards (see 4.2) is connected to the main notion of social vulnerability and differentiating certain personality or group characteristics. Such differentiation can be used in policy contexts to prioritize help. Therefore, the specific theme is also related to the question of who is vulnerable (4.3), and we aim to better understand how to analyze this. Therefore, another specific theme looks into the perception of help to verify a connection between a disaster need and certain population groups that receive such help. Another specific theme is experience with vulnerability (4.4) which also follows up on the first two specific themes in understanding who is vulnerable. While the first two questions were rather hypothetical in understanding better what respondents think about who should receive help or attention, this new specific theme of experience tries to assess whether people have actually seen it happen. Based on these basic and general notions of vulnerability, in the next specific theme-specific characteristics of variables (4.5) are investigated. The respondents received a list of variables traditionally used in social vulnerability indicator assessments. Comparing floods against COVID-19, the main ambition is to identify key variables and convergence as well as contradictions in ranking them between the two hazards, as explained in the rationale of the introduction. The next special theme is living conditions (4.6) and local context, as this is a common demand for many social vulnerability assessments and research. Livelihoods, (social) capital, and similar indicator variables used in traditional social vulnerability assessments have informed the selection of the subcategories of this specific set of questions. In this line, the capacities (4.7) have been analyzed as another special theme since most traditional social ability assessments complement the topic of coping or adaptive capacities to social vulnerability assessments. The final specific theme is revealed vulnerability (4.8). This includes the view of the respondents on categories to best define vulnerability as an outcome or end of a process of a disaster.
An online survey was set up using an open platform, soscisurvey. de, to enable broad participation from many countries and a flexible option for experts to reply to it. The survey was distributed in four languages (English, French, German and Spanish) to experts in disaster risk management and emergency management between September and November 2021. The survey was designed and tested to take around 10 min. A pretest was run with 21 experts from all five continents to ensure the survey could be understood in principle by multiple nationalities. Those pretest experts have a background in natural hazards or disaster risks. Their response was overall positive and included improvement suggestions regarding terminology, the phrasing of questions, missing items, and clarifications of cultural and ethical context.
The anonymous survey consists of nine topic questions, both selection, multiple options and open questions, and additional questions on the respondents background. The survey contains questions that at first glance appear obvious to reply to such as "Should the most vulnerable receive help first". However, such seemingly obvious questions are included here on purpose to check, whether indeed all respondents agree with it. It is also a cross question to check the previous set of questions. And the results of the previous questions already reveal that there is no full agreement and therefore common and unique understanding that all population groups or some couple population groups should receive aid first. This is in line with the discussions in the pretests and respective feedback, which revealed that indeed, even experts are uncertain how to position themselves when comparing calls to help everyone equally versus a selection of specific persons or those vulnerable. It is also the core research aim of this paper to unravel who are more or most vulnerable, in comparison to others. Therefore, additional sets of questions and reply options are provided to better unravel, in which direction people really think when conceptualizing "the vulnerable". Questions items, therefore, contain medical indications, accessibility, the idea to treat everyone equally, social role or demand. Several of those response options can overlap, as do other social vulnerability variables, quite commonly, such as population density or certain age groups. We do not know how respondents define vulnerability, and do not pre-occupy them with specific definitions. The main purpose is to reanalyze pre-existing conceptualizations of vulnerability by gaining empirical replies, offering different directions to further explore. It serves to question or confirm existing notions of vulnerability.
The survey answers were analyzed statistically while controlling for respondents background and experience to check for biases and disentangle the various possible explanations.
Answers are compared for floods with answers for pandemics according to the respondents background with bivariate statistical test and logistic regressions with a backward stepwise approach to systematically test for country of residence, gender, age, context (rural or urban), occupation (professionals), experience in hazards research and rescue training or experience. We only report significant differences at the 0.05 level.

Results
The total number of respondents is 366, with 286 complete or valid answers. Respondents have mainly used the English version (n = 151), followed by German (86), French (32), and Spanish (17).

Respondents' profile
There are some imbalances in the respondents' backgrounds (Table 1). There is an overrepresentation of respondents from Germany, probably due to the heavy flood losses in 2021 and high interest amongst practitioners and researchers. The age range sees quite an even distribution, fitting to the working ages of the respondents. There is a male overrepresentation, probably due to the number of practitioners involved. Most respondents have a background in research or studies or are practitioners in different fields. Most responses come from urban residents, but rural areas and towns are also represented. A high share of the respondents has either research or practical experience or training. As a result of these imbalances, we systematically control all answers by respondent's background and the experience of respondents (research vs. rescue), which turns out to be far more often significant than their personal characteristics (country, age, gender) (see Table 2).

Prioritization in context to COVID-19 and natural hazards
"During COVID-19, people at higher risk have received priority (protection, vaccinations, etc.). In other areas such as natural hazards, should also some population groups receive aid or care first?" This first survey question is designed to identify whether the more vulnerableor other groups -should be prioritized. It focuses on the broad context of natural hazards to capture answers independent of the context of disasters and emergencies.
There is strong support among the respondents to offer help first to some groups (n = 231, 81%) as opposed to treating everyone equally (n = 31, 11%) during and after a disaster, whereas less than 10% did not know (Fig. 1). As a result of the imbalance in the sample, we controlled all answers by the respondent's background and experience -only significant differences are reported (see Table 2).
Respondents with a background in hazards research are two times more likely to answer yes (1.1-4.1 95% CI), those with training or experience in rescue are three times less likely to answer yes (0.2-0.6 95% CI) (all other things being equal).
There seems to be some confidence that some groups should be prioritized. Interestingly, the result is backed not only by researchers selected for this study but also by practitioners.

Who should receive help first?
The second question -"Who should receive help first?" -intends to narrow down the previous question on a decision whether or not vulnerability plays a role. It focuses on the context of the two hazards of the survey title and cross-checks the previous question. Several options are provided to prioritize vulnerable people. It includes the key hypothesis that the most vulnerable should be in the focus of assessments [76], and those with specific socio-economic vulnerability characteristics, but also questions of exposure and spatial setting [22]. More options are added to rescue and treatment situations when questions of prioritization of medical treatment (also triage) play a role in emergency or disaster management [77]. An option is added for comparing the prioritization with an alternative to treat everyone equally, a common denominator in humanitarian assistance (Sphere Association 2018).
Most respondents declare that the most vulnerable should be treated or rescued first (Fig. 2). There is also a majority declaring that those with medical indications (illness, wounds etc.) should be treated first. The other possible prioritizations (easiest to reach, Fig. 2. Responses to the question "Who should receive help first?" most important social or economic role, asking for help the earliest) are supported by less than 10% of respondents. There is also little support for treating everyone equally.
Respondents with a background in hazards research are two times more likely to prioritize people with medical indications (1.2-3.3 95% CI); this is also the case for respondents based in Germany (1.1-3.2 95% CI). Respondents based in Germany are two times more likely to advocate treating everyone equally (1.0-5.5 95% CI), whereas respondents with a background in hazards research are two times less likely to do so (0.2-0.9 95% CI).
Interpreting the results, prioritizing the vulnerable seems to be preferred over treating everyone equally. It is interesting to note that treating everyone equally is in a similar range to socio-economic conditions. Socio-economic conditions are the predominant theme in most social vulnerability indicator or index approaches [15]. Also, rescuing those with medical conditions seems to dominate over socio-economic conditions. This can indicate the needs to broaden the range of social vulnerability question items and variables on asking for help, distance to reach people, and especially, medical conditions.

Experience with vulnerability
"Have you ever seen yourself that people were treated differently, according to being in a risk group or being vulnerable?" This question is based on the recognition that (social) vulnerability mainly is a conceptual construct [19,78] but based upon real-life experiences of practitioners reporting from humanitarian aid and development research [22]. Therefore, it is a survey question to test whether experts have experienced vulnerability in reality or base their opinion on theoretical knowledge. The question also is rarely asked in studies, so it is also of empirical interest whether people think they have actually experienced vulnerability. Experience is a broad term and has been narrowed down to visual observation.
The most common situation is respondents declaring that they did not experience having seen people being treated differently during an emergency or after a disaster (n = 84, 29%) (Fig. 3). Just below, almost one-third did see that vulnerable people were treated differently (n = 81, 28%). The other situations that were often observed are that some people needed longer to recover than others (n = 75, 26%) and that some people received donor aid or aid support before others after an incident (n = 69, 24%). Conversely, fewer respondents observed that preparedness programs were directed at helping certain people first (n = 45, 16%), whereas less than 13% did not know if they had observed people being treated differently during an emergency or after a disaster.
The only significant differences are that respondents living in a rural background are two times more likely to have observed that some people received aid support before others (1.0-3.8 95% CI), whereas respondents based in Germany are two times less likely to have observed this (0.2-0.8 95% CI).
Interestingly, rural people have more experience with receiving aid support. For future studies, this may indicate a need for separating vulnerability assessments of cities versus rural areas, as often suggested [79]. It can also direct considerations on how to validate social vulnerability, rescue, aid distribution, recovery, and specific preparedness programs seem all possible categories.
There are major discrepancies between what respondents think that should be done to prioritize help to certain groups and what they have experienced is being done on the field (Fig. 4). For example, while a majority of respondents declared that vulnerable groups should receive help first, among them, only one in ten also declared that they had personally seen vulnerable groups treated differently. Conversely, more than three-quarters of the respondents declaring that they had seen vulnerable groups being treated differently had previously declared that they should be helped first. But this is not a unanimous situation as for every other answer to the first question. A minority also stated that -contrary to their expectations expressed in the previous question -they had seen vulnerable groups being treated differently. Concluding, while an overwhelming majority consider that the most vulnerable should be treated or rescued first, most of them declare that they did not witness any prioritization. The others balance their responses among the other possibilities. While there is convergence on what should be done in principle, the practical experiences of respondents in each position vary widely. Overall, there is a major gap between their goals and what is being done on the ground.

Vulnerability characteristics or variables in floods versus pandemics
Based on rising use of social vulnerability assessments using socio-demographic characteristics, and related traits for qualitative assessments or variables in indicators [58], it is of interest to learn from experts, which of them are useful for floods and pandemics. Therefore, the following survey question 4 was asked: "In your view which personal characteristics result in vulnerability when floods or pandemics occur?" Comparing both hazards is of interest since their nature and impacts are quite different, but the same social characteristics are used to assess both hazards in recent studies (see state of the art above).
Respondents did not rank the personal characteristics resulting in vulnerability in the same way for floods and pandemics (Fig. 5). In the case of floods, the most cited characteristics are mobility impaired people (n = 264, 92%), older people (n = 248, 87%) and poor people (n = 199, 69%). In the case of pandemics, the most cited characteristics are older people (n = 260, 91%), poor people (n = 182, 64%), single people 65 years or older (n = 149, 52%) and people that are not trusting in authorities (n = 139, 49%). While the order is not the same for floods and COVID, there is some convergence among the five to eight most cited characteristics.
In the case of floods, respondents with a background in hazards research are five times more likely to mention mobility-impaired people (1.8-13.8 95% CI), and those with training or experience in rescue are six times more likely (1.9-25.6 95% CI) to mention this. Respondents with experience in research are also two and half times more likely to mention older people (1.3-5.6 95% CI). Respondents with experience in research are also three times more likely to mention poor people (1.6-5.0 95% CI), and those living in a rural context are two times more likely (1.1-5.7 95% CI). In contrast, older respondents (0.2-0.7 95% CI) and those with rescue experience (0.3-0.9 95% CI) are two times less likely to do so.
In the case of pandemics, respondents with experience in research are five times more likely to mention older people (1.9-12.9 95% CI). In contrast, those based in Germany are two times less likely to mention mobility-impaired people (0.3-0.9 95% CI). Respondents with experience in research (1.1-2.7 95% CI) and younger ones (1.2-3.6 95% CI) are two times more likely to mention poor people. Respondents living in a rural context are two times more likely (1.1-3.3 95% CI), and younger respondents are three times more likely (1.8-5.0 95% CI) to mention trust in authorities.
It is interesting that a background in research again influences the replies towards convergence with existing literature on social vulnerability characteristics. It is just as interesting how opinions from other groups, such as practitioners and young or rural people, differ. Mobility impairment is very interesting to consider to be included more in future studies on floods, as compared to other socioeconomic factors. Unsurprisingly, some factors differ between floods and COVID, such as mobility impairment or single-parent households, due to the different impacts of both hazards on people. However, it is striking how many characteristics are thought to serve in assessing both hazards.
Conducting another comparison of the question 4 items (Fig. 6), the lines are proportional to the share of respondents moving from one answer to the other. They are blue when answering "yes" in the case of floods and red when the answer is "no". For example, 248 respondents (87%) declared old age an increasing vulnerability to floods, and 231 (93%) considered old age an increased vulnerability to pandemics. Conversely, most respondents declared that old age is not increasing vulnerability to floods yet replied that it does increase vulnerability to Covid-19. 'Poor people' is an answer in a similar position, but with more noticeable differences, more respondents answer yes for floods and no for pandemics (15%) and vice versa (9%). Conversely, while 264 respondents declared mobility impaired people are vulnerable to floods, only 68 (26%) consider mobility also to increase vulnerability to pandemics. Single-  parent household is an answer in a similar position, but with less pronounced differences; fewer respondents answer yes for floods and no for pandemics (44% of them). A third situation is to have the same share of responses for floods and for pandemics, but with a minority of respondents switching from yes to no, balancing the shift from no to yes. This is the case for the single people, 65 years or older, and the lack of trust in authorities replies.

Living conditions and context
Many social vulnerability studies emphasize the dependency of studies on the local context (Wisner et al., 2004). We, therefore, asked question 5; "Which living conditions contribute to people's increased vulnerability when floods or pandemics occur?" Respondents mostly agreed on the living conditions contributing to vulnerability to floods and pandemics (Fig. 7). In the case of floods, the most cited living conditions are homelessness (n = 187, 65%), illegal immigrants (n = 156, 54%), living in mobile homes (n = 154, 54%), unemployment (n = 148, 52%) and rural communities (n = 140, 49%). In the case of pandemics, the most cited living conditions are homelessness (n = 205, 72%), illegal immigrants (n = 186, 65%), densely populated areas (n = 184, 64%), unemployment (n = 155, 54%) and urban communities (n = 137, 48%). Respondents show that vulnerability arising from living conditions is not context dependent as they are quite similar in the case of floods and pandemics.
In the case of floods, female respondents are two and half times more likely (1.5-4.9 95% CI), and those with a background in hazards research are one and half times more likely (1.0-2.9 95% CI) to mention homelessness, whereas those based in Germany are two times less likely to do so (0.3-0.9 95% CI). Researchers are also two and half times more likely to mention illegal immigrants (1.6-4.3 95% CI), whereas respondents with rescue experience or training are two times less likely to do so (0.2-0.8 95% CI). Researchers (1.2-3.2 95% CI) and respondents living in a rural context (1.0-4.0 95% CI) are also two times more likely to mention mobile homes. No significant difference for density. Older respondents (0.3-0.9 95% CI) and those based in Germany (0.2-0.7 95% CI) are two times less likely to mention unemployment. Respondents with rescue experience (1.0-3.0 95% CI) and those living in a rural context (1.0-3.9 95% CI) are two times more likely to mention urban communities, whereas younger respondents are two times more likely to mention rural communities (1.1-3.1 95% CI).
In the case of pandemics, researchers (1.0-2.6 95% CI) and respondents living in an urban context (1.1-3.6 95% CI) are two times more likely to mention illegal immigrants. In contrast, younger respondents are two times more likely to mention mobile homes (1.1-3.4 95% CI). No significant difference for homelessness. Younger respondents are also two and half times more likely to mention density (1.3-4.5 95% CI), whereas those based in Germany are two times less likely to do so (0.2-0.8 95% CI). Older (0.3-0.8 95% CI) and Germany-based respondents (0.3-0.8 95% CI) are two times less likely to mention unemployment. Younger respondents are also four times more likely to mention urban communities (1.3-4.5 95% CI), whereas older ones are three times less likely to mention rural communities (0.2-0.6 95% CI), and those based in Germany are five times less likely to do so (0.1-0.4 95% CI).
As in the previous question on vulnerability characteristics, it is interesting to observe that, indeed, a ranking of items emerges. Certain characteristics of special groups and their local conditions and context clearly are homelessness, illegal immigration, mobile homes, but also urban density, and others. This seems to support the characteristics commonly used in social vulnerability assessments. However, there are also differences between floods and COVID, such as in mobile homes or rural areas, that can be informative to analyze in future studies. Comparing the living conditions resulting in vulnerability for floods and pandemics also reveals three situations (Fig. 8). First, a similar share of responses for floods and pandemics. This is the case for homelessness or unemployment. This results from a balance between respondents switching from yes to no and those switching from no to yes. Secondly, a there is a larger share of positive answers for floods than for pandemics. This is the case for living in mobile homes or rural communities. This second situation derives from the majority of those saying yes for floods shifting to no for pandemics. Third, the opposite situation, a lower share of positive answers for floods than for pandemics. This is the case for densely populated areas or illegal immigrants. This third situation derives from roughly half the respondents saying no for floods and shifting to yes for pandemics. This means that despite the apparent overall convergence, there is less agreement among respondents and contrasting views on what increases vulnerability to floods or pandemics.

Lack of capacities
Capacities are either important elements of vulnerability [47] or counterparts to compensate vulnerability or susceptibility [24,46]. Alternative and additional terms include adaptational capacities, abilities, resources, or resilience [80,81]. In order to break this down into a question easily understood by both researchers and practitioners, the following survey question 6 is phrased as; "Which lack of capacities makes people more vulnerable to floods or pandemics?" Respondents did not rank in the same way the lack of capacities making people more vulnerable to floods and pandemics (Fig. 9). In the case of floods, most respondents declared that people without education about hazards (n = 235, 82%), people without experience with hazards (n = 227, 79%), people who do not speak the language of the authorities (n = 224, 78%) and having no technical measures installed against the hazard (n = 218, 76%) are the main situations producing vulnerability to floods. In the case of pandemics, the most cited gaps in capacities are chronic illness (n = 237, 83%), people without education about hazards (n = 217, 76%), and people who do not speak the language of the authorities (n = 203, 71%), medicine-dependent people (n = 199, 69%) and people with no health insurance (n = 181, 63%). Respondents make it clear that they consider the conditions that increase vulnerability to be context-dependent and different in the case of floods or pandemics.
In the case of floods, respondents with a background in hazards research are three times more likely to mention a lack of education about hazards (1.7-6.4 95% CI) and also two times more likely to mention a lack of hazards experience (1.1-3.7 95% CI), just like those with rescue training (1.3-4.9 95% CI). Researchers are three times more likely to mention language issues (1.6-5.5 95% CI), whereas older respondents are two times less likely to do so (0.2-0.7 95% CI). Respondents living in a rural context (1.1-8.8 95% CI) and those based in Germany (1.4-5.7 95% CI) are three times more likely to mention a lack of technical measures, while researchers (1.0-3.5 95% CI) and females (1.2-4.5 95% CI) are two times more likely to do so. Researchers are one and half times (1.0-2.4 95% CI). Females are two times more likely (1.4-4.0 95% CI) to mention chronic illness, while researchers (1.0-2.9 95% CI) and females (1.3-3.8 95% CI) are two times more likely to mention medicine-dependent people. In contrast, older respondents are three times less likely to do so (0.2-0.6 95% CI). Respondents living in an urban context (1.0-3.0 95% CI) and females (1.3-3.8 95% CI) are two times more likely to mention a lack of health insurance, whereas older respondents are three times less likely to do so (0.2-0.6 95% CI).
In the case of pandemics, respondents with rescue experience (1.2-4.3 95% CI) and younger ones (1.2-4.9 95% CI) are two times more likely to mention a lack of education about hazards. In comparison, those with rescue experience (1.1-4.6 95% CI) and those liv- Fig. 8. Comparing the living conditions for floods and pandemics. The interpretation is similar to Fig. 6. For example, while the overwhelming majority of respondents declaring that density is increasing vulnerability to floods also declared that this was also the case for vulnerability to Covid-19, among the respondents declaring that density has no impact on vulnerability to floods, the answers were almost evenly split for the same question relative to Covid-19. ing in a rural context (1.1-3.4 95% CI) are two times more likely to mention lack of hazards experience. Researchers (1.4-4.6 95% CI) and younger respondents (1.1-4.0 95% CI) are two times more likely to mention language issues, while researchers are also two times more likely to mention a lack of technical measures (1.1-3.5 95% CI). Researchers (1.0-3.7 95% CI) and younger respondents (1.2-5.6 95% CI) are also two times more likely to mention chronic illness, while they are also two and half times more likely to mention medicine-dependent people (1.4-4.2 95% CI) (1.5-5.0 95% CI). Researchers are twice as likely to mention a lack of health insurance (1.0-2.8 95% CI), whereas older respondents are three times less likely to do so (0.2-0.6 95% CI).

A. Fekete and S. Rufat
Interestingly, people with a research background seem to favor education over experience with floods. This could represent a general understanding that education is a key instrument to improving risk literacy [19,82]. But why real experience is less regarded by direct comparison would have to be analyzed in future research. There is some indication that singular experiences of floods often are not enough to change behavior [83]. However, experience seems to play a greater role in floods as compared to pandemics.
Comparing the lack of capabilities resulting in vulnerability for floods and pandemics also reveals three situations (Fig. 10). First, a similar share of responses for floods and pandemics, such as for lack of education about hazards or languages issues, results from a balance between respondents switching from yes to no and those switching from no to yes. Second, a larger share of positive answers for floods than for pandemics, such as for previous experience or technical measures, derives from a large sharethe vast majority in the case of technical measuresof these saying yes for floods shifting to no for pandemics. Third, the opposite situation, a lower share of positive answers for floods than for pandemics, such as chronic illness or medicine-dependent people, derives from the majority of those saying no for floods shifting to yes for pandemics. This means that despite the apparent agreement that the conditions that increase vulnerability are context-dependent, the specific choice of a criterion adapted to floods or pandemics is linked to the respondents' training and experiences.

Revealed vulnerability
A known problem in risk and vulnerability studies is that in most cases, it remains a hypothetical assumption, a potential risk or vulnerability [15,84] until a real event reveals whether it played a role or not. But how to prove or assess whether an assumed social vulnerability turned into a "revealed vulnerability" is assessed less often. Even less is known by which categories to prove a vulnerability. Loss and damage assessments only sometimes capture social characteristics equivalent to the multiple variables used in social vulnerability assessments [85,86]. And it is even less understood whether death or any other type of impairment explains which characteristic of (social) vulnerability. Therefore, the question is rather exploratory and asks, "Which category defines best whether people were vulnerable after a flood happened?" Most respondents declare that the category best defining after a flood happened whether people were vulnerable (Fig. 11) is the loss of infrastructure (n = 83, 30%) to support humans (water, food, emergency services, flood defense, etc.). There is no significant difference in any of the respondents' profiles (age, gender, language, training, etc.) on this answer. The other situations that were often mentioned were economic loss or poverty (n = 52, 18%) and deaths (n = 51, 18%). Psychological impacts or trauma was only considered by 10% of respondents, whereas wounds or severe health problems requiring treatment and loss of family members, friends, or pets were only mentioned by less than 5% of respondents each. Respondents tend to assert that economic loss, loss of infrastructure, and death are the best outcome indicators to capture vulnerability to floods.
While there are no significant differences in the answers on the loss of infrastructure (see Table 2), researchers are two times more likely to mention economic loss or poverty (1.0-4.2 95% CI), whereas respondents based in Germany are two times less likely to do so (0.1-0.9 95% CI). Researchers are two times less likely (0.2-0.8 95% CI), while respondents living in a rural context are three times less likely (0.1-0.9 95% CI) to mention deaths, whereas respondents based in Germany are two and half times more likely to mention psychological impacts or trauma (1.1-5.7 95% CI) while younger ones are two times less likely to do so (0.1-0.9 95% CI).
It is quite a surprise that infrastructure loss is often selected. Most social vulnerability assessments neglect infrastructure, only a few studies work on lifelines and their specific social vulnerability, and the area of critical infrastructure also only starts investigating social vulnerability (Holand 2015). It could be that respondents were under the impression of the latest flood damage in Europe, which included heavy infrastructure losses and have hindered rescue operations. Researchers seem to rather follow the state-of-theart literature and theory, as in the other questions. Rural people have another separate opinion again, which underlines the importance of separate studies in the future.

Discussion
An ethical dilemma? The research question is: 'Is it possible to differentiate and prioritize vulnerable groups, and/or should everyone be treated equally?' This is an ethical dilemma known for many risk assessment decisions, be it in floods, COVID-19 triage, or in other fields such as automatic safety systems [30,87,88]. Also, overarching principles of humanitarian assistance have long been established, for example, by the Red Cross and Red Crescent societies and in the Geneva humanitarian conventions. Fig. 11. Categories defining whether people were vulnerable after a flood happened.

A. Fekete and S. Rufat
Based on the experience of human conflict and war, the notion of treating all humans like brothers has become known [89]. This idea of treating everyone equally is also fostered by ambitions of the 'leaving no one behind' actions [90].
On the other hand, it has been long recognized that certain social groups suffer more from disasters than others [26,53]. Diversity in terms of differences in personality traits as well as the local context of social groups is embedded in the brunt of disaster risk reduction research and practice. But we put forward the question in the study whether the actual connection has been made between treating everyone equally on the one hand and social differentiation of certain groups on the other hand. It appears logical to treat certain persons first, but how is this actually carried out in different disaster situations? On a sinking ship, it is a common notion that women and children should be rescued first. In a local pandemic, the elderly and those with preconditions of detrimental health need more attention. But how can this be put into planning and preparedness? There needs to be a better notion of whom to rescue in which hazard first. And, of course, based on a holistic understanding of risk as a process and not as a static picture, different phases of disaster must also be differentiated according to such prioritizations. This limited study is not able to capture all the dimensions before, during, and after a hazard and is also limited in focusing only on two hazards. It is, therefore, intended to be an exploratory study that investigates which questions could be useful to differentiate notions of social vulnerability between disaster types. And the other main idea is to better understand the connection between a notion of social vulnerability on the one hand and its connection to perceptions of needs on the other hand. This study cannot solve the question in the title of whether everyone should be treated first. This is a rather ethical and even philosophical question. But investigating the actual practices in prioritizing certain groups based on needs assessments or social vulnerability assessments could also reveal interesting results of either conversion or contradiction in other studies. Analyzing the results of our survey already indicates some contradiction. While on the one hand, most respondents agree that the vulnerable should be treated first; on the other hand, most people have not actually seen that the vulnerable were treated first.
The results of our survey show there is little support for treating everyone equally. This is cross-checked by the second question and confirmed by the replies to both questions. Few studies seem to have analyzed the ethical dilemma of deciding between treating everyone equally and prioritizing. Most social vulnerability studies follow the paradigm that there needs to be a prioritization of 'the vulnerable'. And humanitarian aid follows the principle of treating everyone equally. How do both fit together? Certainly, this study can only indicate a further need to investigate this question. The main purpose was to make respondents reconsider whom to prioritize among the vulnerable, understand the difference between hazards, and assess the categories best defining vulnerability after a disaster. Additional research is required to understand how and in which situations people need to be treated equally and how or whether this can be integrated with an understanding of different social vulnerabilities.
Validating vulnerability: The main objective of the study was to understand better which of the existing characteristics assumed to explain social vulnerability are regarded more useful than others by experts in the field. The results not only indicate that a ranking of variables or question items can be conducted and that there is an agreement on some characteristics being more relevant than others. This shows that prioritization of certain characteristics is possible. This can help to select certain variables for further research. It can also help to inform which variables or characteristics to select for validation studies. Validation studies are still lacking in vulnerability research and practice [34]. Another challenge emerged when asking how to prove or validate whether vulnerability played a role or not. The results are constrained to visual observation only. Certainly, other forms of observation need to be analyzed in addition. That such proofing of vulnerability is necessary can also be derived from the results that many more respondents think that certain groups should be prioritized but have not observed or experienced such prioritization in real situations. We do not wish to argue that all researchers necessarily have to be in real situations of loss and damage to be able to draw up helpful assessments. On the contrary, theoretical developments in the field of social vulnerability have advanced overall risk assessments. Since previous risk formulas and models were largely based on a hazard perspective only. But it is noteworthy to see that seminal studies on social vulnerability are based on practitioner observations on vulnerability. This connection between practitioners and researchers in validating vulnerability and risk assessments should be continued. Validation can also help to unravel limitations and caveats in analyzing vulnerability, but interestingly, recent publications do not address these topics explicitly [14].
Context dependency: This study has also analyzed whether vulnerability factors and outcome indicators are generic or contextdependent. The statistical analysis indicates that context dependency related to age, gender, language, and cultural context play a role, but even more, a background in hazard research or studies or training. The latter two are hardly addressed within the usual set of context variables that are used in many social vulnerability assessments [59]. Learning from the insights of this study can therefore help to improve future studies. It is interesting to observe that many of the responses correspond between the researchers in the respondents with the existing literature on social vulnerability theory and assessments. The responses of practitioners sometimes differ. It also appears that certain questions related to rescue or medical conditions have hardly been addressed in previous social vulnerability studies. Stemming from an interdisciplinary background of different fields of studies on social vulnerability in natural hazards and pandemic contexts, civil protection and critical infrastructure studies, as well as emergency disaster management and rescue management, we have brought together additional questions to test for vulnerability. In many established fields of social vulnerability, the argumentation is rather one-sided on power and politics, focusing on socio-economic aspects and governance, which are very important, but they do not address aspects such as rescue [14]. The findings in our study also indicate areas to advance existing social vulnerability assessments, and also, the results indicate that the socio-economic conditions alone do not explain the whole range of social vulnerability.
Limitations of this study: Regarding the focus of the two selected hazards, other hazard types are missing that have different ramifications for the needs and viability of certain people in groups. In a heat wave and drought, elderly citizens and rural people are affected differently than in a flood. Also, the specific pandemic selected, COVID-19, may not represent other variants of epidemics. Since other epidemics or even pandemics may affect different types of people and local contexts. The main limitation may be, how-ever, to miss out on other types of hazards with very different means of impact. For example, criminal acts or terrorism could add a dimension of intelligent attacks on certain groups of people. However, the main idea of this study was to explore how social vulnerability can be better validated. While some studies have tried to validate social vulnerability by statistical sensitivity analysis and others have tried to verify it by capturing more context on the ground, the study tried to show ways to better understand social vulnerability variables to be used in broad-scale assessments. This includes several other limitations. For example, capturing detail as in more local social vulnerability assessments. And there are known limitations of analyzing pre-fixed categories of vulnerable groups and some research also focuses rather on the intersecting nature of individuals and groups [91].
There is some inherent limitation in the selection of questions and selection response options. For example, replying "no" to the question about real experience could be further differentiated in follow-up studies. It could be interesting to analyze whether people have not been in a disaster situation or have been in such a situation but have not observed vulnerability. Other studies used more variables and characteristics of social vulnerability, but we had to limit the already very long list of variables for the survey. The survey was developed to make the questionnaire not too long and to focus on certain selected questions of main interest. Certainly, the concept of vulnerability and the related concepts of capacities and resilience would enable many additional questions to be asked. We understand the study, therefore, only as a starting point to raise awareness about certain gaps in the connection between ability and prioritization decisions.
The sample group and responses from Germany stand out in the studies. However, we systematically control the answers of the respondents based in Germany along with the other background characteristics. It is worth highlighting that the professional experience of respondentsall other things being equalintroduce twice as many significant differences as being based in Germany (see Table  2). In other words, most of the time it is their age, training and/or research vs. rescue experience that impacts their answers rather than the fact that they are based in Germany or not.
The respondents' awareness may also be correlated to the recent flood events in Germany, which have shed new light on the importance of more detailed risk assessments. It may be important to mention that the respondents' cultural background could shape their judgments in this case. In other studies, we have analyzed this in more depth and have found out that the differences in the German respondents to another survey looking more in detail into the perceptions and motivations as compared to other samples in Romania or neighboring countries can be related to certain characteristics of social, or institutional background in Germany [92]. For example, coordination and organizational weaknesses were revealed in the 2021 flood operations, which may lead to a more critical perception. In addition, in Germany, the discussion around the involvement of volunteers has become an important topic in the past decade. Compared to other countries such as Romania, the responses could be different from observing top-down regulations and may include more perception about bottom-up approaches and needs for integration. In principle, this could also influence the results of the study presented in this paper. For example, there is a tendency to overrepresent certain groups that are not sufficiently included in the processes.
But on the other hand, we assume that the backgrounds in different countries are more important than the singular flood event in Germany. We assume a certain convergence of opinions from Western and northern hemisphere countries towards an agreement to prioritize certain social groups. But while the agreement on social vulnerability variables could be consistent, the underlying cultural background could result in different conclusions from such assessments. This is beyond the scope of the study, but certain research has shown that cultural grid and other factors related to institutional and social structure influence who is prioritized and which social groups are treated differently [93].
Practical wider implications: Relating to the topic of the Virtual Special Issue "What makes individuals fall through the safety nets during disasters?" this study has mainly centered on people to be rescued and people and their vulnerability characteristics. Especially the latter as a general recognition of certain attributes that make certain population groups of people more vulnerable than others is tightly connected to the overall question of human security and thereby addressing safety nets. Safety nets can be understood as an overall perception of a society of who is to be supported in different situations. Resources may stand out as the most extreme of situations where people in society need a safety net. This study, therefore, contributes to disaster risk reduction by better addressing the needs of individuals in a specific severe life-threatening situation such as a flood or health impairment through a pandemic. Preventing people from being killed or experiencing problems in recovery from floods or pandemics can be guided by specific measures in disaster risk reduction. Most directly, it would be in the immediate situation of saving lives in rescue when certain population groups would be prioritized. This, of course, has implications for other population groups that will be saved later. This indicates a need for a broader political and societal discussion on whether it is fair or important to treat certain people first while putting other people in a waiting line.

Conclusion
Are the same people vulnerable to floods and pandemics? This study has looked into vulnerability characteristics commonly analyzed in social vulnerability assessments. While there has been a convergence in social vulnerability research on several socioeconomic characteristics, a lack of studies to actually validate vulnerability has been noted. The survey has asked experts and practitioners to prioritize and rank vulnerability characteristics. The results reveal that hazard types as different as floods and pandemics can be assessed using the same characteristics, although a few differences exist in the importance of the variables. This study has looked into the main research question of whether experts and researchers think that vulnerable groups should be prioritized and, if so -based on which criteria. The results show some convergence on certain criteria and an overall agreement on prioritization. However, the criteria ranking differs between the two hazards, floods and pandemics. In order to understand a selection preference of experts on social vulnerability, the results of specific research themes provide more detail. The study reveals that prioritization differs according to a different context of hazards but also regarding the perception of who is vulnerable. It is also shown that it is a difference between an assumption of vulnerability and vulnerability really experienced. Less than half of the experts have had some concrete experience with a visible or revealed vulnerability. The criteria or variables vary between the two hazards as well as between living conditions and capacities.
The study's results can help inspire further research on connecting social vulnerability assessments with validation studies and needs assessments. While this connection has been strong in the origin of social vulnerability indicator studies, it is much less addressed in recent quantitative indicator studies on social vulnerability. But the results show that not only is this connection important, but also there is still a gap in benchmarking social vulnerability. Knowing who was vulnerable due to which characteristics is still lacking in disaster research. Benchmarking social vulnerability could benefit from experience in having seen social vulnerability connected to disaster outcomes. The findings indirectly revealed that identifying social vulnerability is an ongoing challenge. The study has narrowed down this question on whether differentiation or prioritization has taken place in rescue situations. Further research in this direction is encouraged.
The survey also indicates future research avenues on how to validate vulnerability by which criteria. For instance, it is still unknown how vulnerability can be revealed, and existing loss and damage studies need to integrate further characteristics of social vulnerability to enable such validation. The results also reveal how hard it is to prioritize certain vulnerable groups, on the one hand, while trying to treat everyone equally, on the other hand. This ethical dilemma between humanitarian principles and social ideas warrants further research. Should everyone in need be treated equally? Answering this difficult question also seems to depend on how much research background or practical insights experts have. Differences also emerge between rural populations, different countries, and situational experiences, for example, the latest floods or the COVID-19 pandemic. It will be informative to conduct a follow-up longitudinal study in a decade to check whether or not the assumptions and rankings have changed.

Ethics approval statement
An ethics committee at TH Köln has examined and approved this study.

Declaration of competing interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Data availability
Data will be made available on request.