The literature shows that there are a lot of factors influencing flood vulnerability, such as socioeconomic, demographic, environmental, institutional, and spatial factors that interact with each other. Repetitive and regular socioeconomic factors are always identified as factors of vulnerability. Factors that make people more vulnerable to the effects of flooding and limit their capacity to adapt are poverty, low income, poor housing, limited access to services, poor infrastructure, and low education ([83]; [89]; [27]; [47]; [74]; [28]). In the same way, [80], [85], [7], [102], [35], and [113] reported that having lower income, fewer assets, and inadequate adaptation options make people more vulnerable, while higher income and owning assets and adaptation options reduce risk. [51], [62], [5], [94] also noted that poverty and inadequate livelihood assets and coping mechanisms increase livelihood vulnerability, and that better assets and coping mechanisms increase livelihood resilience.

Other factors related to space and the environment are also important drivers of vulnerability. Some prominent studies have established that there are significant geographical vulnerabilities to flooding, with low-lying areas, riverine regions, and compact population areas being most vulnerable ([76]; [41]; [98]; [4]).

Environmental factors like flood frequency, elevation, river proximity, landscape, drainage density, rain spreading, and land-use characteristics have a significant impact on exposure to flooding ([79]; [76]; [109]; [49]). Similarly, [75], [96] found that areas near groundwater, flood-prone areas, and compact populations have greater vulnerability because they are more exposed to flooding and have less adaptive capacity.

The literature specifies that vulnerability is not only a result of physical exposure to flooding, but rather a product of the interaction of environmental factors, socioeconomic disparities, demographic features, institutional capacity, and governance effectiveness. Poor communities with inadequate resources, poor infrastructure, social demotion, and high exposure to environmental hazards steadily show high vulnerability, whereas strong livelihood assets, adaptive capacities, and access to resources are constantly associated with reduced risks and increased resilience ([61]; [51]; [85]).

The literature has shown that community and household resilience to climate-related hazards in Bangladesh is a multidimensional phenomenon which is influenced by social, economic, institutional, infrastructural, environmental, and governance scopes. Research indicates that there are noteworthy geographic and socioeconomic differences in resilience across regions and hazard settings. Socio-economic variances, infrastructure, services, and social capital influence the resilience of communities to riverine floods ([84]; [38]; [60]; [63]; [89]).

The same was found: social unity and local adaptation practices played a positive role in increasing resilience; however, there was also a need to address the limits of infrastructure, financial resources, and institutional capacity building ([66]; [64]; [9]; [12]; [29]).

Poverty, environmental dilapidation, limited resources, and their associated factors have been found to be key challenges that hamper resilience, while effective early warning systems, cyclone shelters, native knowledge, and the participation of communities have been identified as building blocks for improving adaptive capacity and recovery ([44]; [1]; [55]; [19]; [86]; [95]).

[21], [45], [110], and [116] identified livelihoods divergence, financial resources, education, institutional support, and access to climate information as factors influencing the resilience of coastal fishing communities. Similarly, [99] have revealed that community assets, such as human, social, financial, natural, physical, and institutional assets, are highly connected with climate-induced drought resilience in Barind Tract. The results indicate that improving livelihoods, resources, and access to them continue to be key to growing adaptive capacity and lessening vulnerability. In addition, policy interference, governance, and community involvement have been highlighted as key factors in achieving resilience ([32]; [105]; [114]).

It is shown in the literature that flood vulnerability is the result of the combination of complex socioeconomic, demographic, environmental, institutional, and spatial factors. In practice, the following factors have been increasingly identified as protective: wealth, housing quality, infrastructure, education, and adaptive assets, and in contrast, dense low-lying riverine populations, weak drainage, low income, and limited livelihood assets are consistently identified as amplifying vulnerability. In turn, resilience has proven to be multi-dimensional, including physical, social, financial, and institutional capitals. However, this study attempted to fill the gap in the literature by measuring the community resilience outcomes of flood-vulnerable people in Bangladesh by using a 5-point Likert scale.

Bagar Char, Merur Char, and Nilakshmia union of Bakshiganj Upazila of Jamalpur district were the study areas for conducting this study. The study population was around 120,000, who were spread across the three unions of Bakshiganj Upazila of Jamalpur district.

This study applied a mix of purposive and random sampling techniques. The study area was determined through non-probability sampling (purposive), and on the other hand, probability sampling (random) was used to select the respondents. The study population comprised males and females of households of all the selected unions. With a 95% level of confidence and a 5% margin of error, the sample size is 383 households, but 390 households were sampled. Further buffer was added for any possible problems that could occur, such as not submitting the appropriate information or important information being lost or overwritten. 130 respondents from each union have been selected for the survey. The following criteria were applied for choosing the respondents: (a) male and female; (b) age not less than 21 years; (c) physical and mental condition to take part in the survey; and (d) a desire to take part in the survey.

The people of households in the selected unions were interviewed using a structured survey questionnaire to collect data on DRM awareness; risk reduction; government support to coping; NGO support to coping; vulnerability mapped; community cooperation; loss of life reduced; property damage reduced; community leaders’ roles; precautionary measures; stress reduction; post-disaster health safety; sustainable livelihoods; hazard-resistant cropping; eco-friendly practices; and hazard-resilient construction. It used a 5-point Likert scale, which extended from strongly disagree (1) to strongly agree (5), to assess the respondents’ attitudes. The middle alternative (3 = Neutral) was left in place to enable respondents to show true impartiality in answers to the items where neither agreement nor disagreement was expressed. The format of the questionnaire was structured, which resulted in consistency in data collection from the sample, reducing interviewer bias and discrepancies in data recording. The field interviewers were trained to conduct in-person interviews in a standard manner and received information about the goals of the study and the proper administration of each item. The questionnaire was planned in the language that is easily understood by the respondents to ensure clarity and comprehension. The questionnaire was tested by enumerators to ensure that they were complete and then coded for analysis. The data were collected in June 2020 to June 2021.

The collected data were coded, entered, and analyzed using SPSS. The responses from the 390 respondents were summarized using descriptive statistics (percentages and means) and considered by the 16 five-point Likert scale items. Responses were received through a 5-point Likert scale to specify their agreement with each statement, with 1 representing strongly disagree and 5 representing strongly agree. This assessment was followed by a visual representation in the form of two corresponding charts. The diverging stacked bar chart gives a breakdown by responses of where community members placed themselves in the 16 outcomes on the five scale points. The bubble mean score chart summarizes the same data in one easily comparable tool, the arithmetic mean score, which can be used to rapidly rank order the 16 statements based on the level of community ratification.

The research was conducted in accordance with the guidelines of voluntary participation, informed consent, privacy, and protection from harm. There was no pressure for participation, and oral consent was received from all respondents who were informed about the purpose of the study. The respondents were free to decline to answer any question and to stop replying at any point without penalty. The data were anonymized and stored on password-protected devices, which were only accessible by the researcher.

By studying the figure in detail (Figure 1), we can see that the most dominant place in the collective authorization of the community was in the forms of protective measures taken. This was the highest level of agreement (76%) and strongly agree (64%) by any proportion of respondents to the 16 statements. This is a noteworthy finding and reveals a deeper level of meaning, as residents believe the precautionary measures put in place in their community, whether at the community level (such as early-warning protocols or structural mitigation measures) or at the household level (such as preparedness measures), are relevant and that they are truly being applied on a day-to-day basis. Three other outcomes received good levels of firm authorization, although at slightly lower strengths. A majority (60%) of respondents strongly agreed that vulnerability mapping was significant to have hazard levels, vulnerable areas, and evacuation routes formally mapped and cartographically documented. Sustainable livelihoods received 53% strongly agree responses, with only just over half of the community members directly linking developments in their long-term economic resilience to community-based DRR activities. Risks reduced by community contribution did not get as few strongly agree responses as any other foremost cluster, but 51% of respondents strongly agreed, marking a clear majority that community contribution itself is a risk reduction mechanism.

Figure 1. Community resilience outcomes—Diverging Likert chart.

As can be seen in Figure 2, the mean scores for all 16 statements of outcomes were in a fairly limited range of 3.79 - 4.29. Outstandingly, the majority of items were at or above “agree” (4.0 or higher on a five-point Likert scale), whereas none of the items was below “neutral” (3.0 or lower). This pattern suggests that the community was not just standing or accepting interventions as such, but showed a clear and reliable leaning to affirm tangible benefits that have been delivered by the interventions. Upon closer review of the mean scores, a clear order of the perceived outcomes becomes deceptive. Vulnerability mapping was the most popular outcome as it acknowledged the highest mean score of 4.29. This finding is especially important as it designates a very high level of community agreement that there is a systematic process of identifying, documenting, and mapping hazard levels, geographic areas of vulnerability, and selected evacuation ways within the community. Four other statements assembled closely together in a narrow range of agreement were second on the list of leading outcomes. First, the mean score for risk reduction due to community participation was 4.20, indicating that residents feel that there has been a positive impact on their communities’ security directly as a result of them being part of the risk management activities. Second, the mean score for precautions taken was 4.19, reflecting, generally, that there has been recognition that the necessary defensive action has been taken at the community level. Third, sustainable livelihoods had a mean score of 4.17, representing that the participants recognize the linking between disaster risk reduction activities and sustainable economic livelihoods of their households and community. Four, this highly similar mix of outcomes closed with property damage reduction, at around 4.16, thus representing that interventions have produced real perceptible protection of physical assets. The close alignment of these four outcomes—with a difference of only 0.04 points—indicates that disaster risk reduction, preventive behavior, livelihoods sustainability, and asset protection are seen as intensely consistent outcomes and not separate outcomes. The findings, taken together, suggest that the level of involvement in disaster risk management in the local community has led to multidimensional welfares and these benefit the normal lives of community members.

Figure 2. Mean agreement scores for community resilience outcomes.

A diverging stacked bar chart provides a far more detailed view of these same data, not only showing the central tendencies but also the full range of community responses for each of the five Likert responses. The diverging chart reveals the consistency of community perception, going beyond the single numeric summary provided by the bubble chart and showing the number of respondents who expressed ambiguity, mild reservation, or disagreement. This richer visualization allows for a more nuanced interpretation of disaster risk management experiences for the intended recipients of a community-based DRM system, as opposed to average ratings. Coping and DRM awareness (Mean 3.88 and 3.79, respectively) had the lowest mean ratings of the study. While both scores are above the neutral verge of 3.0 and thus not considered to be a sign of program failure, these lower scores are best interpreted as representing comparatively weak levels of community confidence in two areas: community support mechanisms provided outwardly to the program, and the extent to which awareness-raising activities have reached the community. Lower mean scores in these specific indicators are particularly important, as low scores can be due to visibility or accessibility problems, or lack of trust in actors outside the disaster risk management outline and not to a lack of the concept itself. A second insight related to health safety, which was the outcome of post-disaster health safety, scored a mean score of 3.90. This score is just below the common “agree” score of 4.0, but the difference is, however, a meaningful one, although numerically small. The outcome shows that even if communities were able to cope with flood events, they were struggling to maintain health and hygiene, indicating the ongoing challenges faced by communities. From a practical perspective, this means that in the community, preparedness for disaster and risk reduction is seen to be effective, whereas the post-disaster phase is a comparatively sensitive space in the community’s resilience profile that may not have seen the same level of capacity building in the other outcomes measured as per preparedness and risk reduction. Disaster risk management activities at the community level have generally shown perceived improvements across most of the 16 outcome domains, with the highest levels of endorsement being given to the core activities of precautionary action and vulnerability mapping.