The academic evidence shows a multifaceted connection between disaster exposure and risk perception and sometimes the “risk perception paradox”, where people can incongruously undervalue their future vulnerability after a disaster. This is often clarified by the accessibility heuristic, which argues that someone adjusts their mental likelihood approximation according to the ease with which he or she can remember recent events ([15]; [72]; [42]; [34]). However, other pragmatic studies have shown that personal experience with a natural hazard event, particularly an evacuation, can expressively increase perceived risk for several natural hazards ([63]; [95]; [6]; [33]; [13]). However, the newer studies suggest that the potential for future vulnerability and community-level exposure may be more operational than individual evacuation history to assess risk ([5]; [24]; [17]; [99]; [27]). In addition, institutional trust can affect the impact of previous exposure, since households with low trust in the institutional emergency management processes may be more likely to experience risk concerns even if they do not have a personal experience of the hazard ([110]; [37]; [105]). Furthermore, researchers have suggested that past experience has an optimistic impact on future preparedness only when the disaster is relevant to the preparedness ([89]; [10]; [39]; [30]). Indeed, some studies have found that past disasters have urged households to take steps toward structural mitigation or to build up their stores of essential supplies, while others have shown that the degree of disaster losses can paradoxically lead to an upsurge in risk tolerance and not mitigation ([11]; [78]).

However, the capacity of populations to deal with environmental risks is inadequate due to their economic status, which influences individual preparedness and the community’s structural ability to protect itself ([70]; [25]; [92]; [73]; [101]). In addition to instant material resources, this gap is exacerbated by a malicious cycle in which disaster-induced losses further weaken the financial strength of disadvantaged households, making them vulnerable to disaster in a cruel circle ([47]; [41]; [4]; [77]; [23]). Moreover, the lack of firm needs in vulnerable groups is emphasized by traditional estimation methods that often fail to consider their specific needs, leading to inadequate emergency relief allocation ([53]; [75]; [7]; [54]). Furthermore, institutional failures are often replicated in the planning of disaster mitigation strategies, which are often created on an expected level of preparedness and literacy that demoted groups do not have ([116]; [58]; [100]; [97]; [52]).

Empirical evidence shows that people living in high-risk areas, recognized by hazard mapping, tend to be more aware of climate change than those who live in more stable areas ([3]; [76]; [66]; [60]; [49]). But this connection has been challenged by several longitudinal studies that have found that objective physical vulnerability does not necessarily associate with vulnerability insights or policy support ([29]; [84]; [14]; [59]). This “risk perception paradox” is often observed and leads to an unpredictable perception of the future probability of a disaster when people are exposed to extreme climatic events directly ([88]; [9]; [85]; [117]; [38]).

While important research has considered public concern towards natural disasters, there is not a large body of quantitative research that has addressed the question using ethnographic qualitative enquiry with a single type of disaster, mainly flood, earthquake and cyclone. While the present study has assisted in advancing knowledge about the cultural, religious and socioeconomic dimensions of risk perception, this work has not yet led to the development of a common and comparable index of risk concern for a wide range of locally relevant risks. Thus, there remains a lack of quantitative data concerning the relative importance and priorities communities assign to various environmental threats and the many perceptions of threats that have received relatively little study, but may be recurring and poorly understood, such as nor’westerly, cold wave, river bank erosion, driving rain, sand heat and lateral water pressure. There is little research that has followed a quantitative survey approach to measure the level of concern, expressed in intensity. In this study it has been tried to address the above gaps by employing a five-point Likert scale to obtain the interval level mean scores and standard deviation value regarding the level of concern towards 10 disasters.

The study area was Bakshiganj Upazila of Jamalpur District in Bangladesh. The area was chosen because it is situated in a district where disasters regularly occur due to its geographical vulnerability, socioeconomic features, and as an appropriate setting for studying community concern about natural disasters.

The study was conducted in 12 purposively selected wards from three unions of Bakshiganj Upazila, Jamalpur District, where the wards were selected on their relevance to the study objective. The sampling frame was developed with the help of the information of population and family information collected from relevant government offices, Union Parishad offices, and the Upazila Statistics Office. The household was the sampling unit, and one eligible household member from each sampled household was chosen. The household members aged 21 - 70 years who were physically and mentally capable of answering and were willing to participate were eligible. 390 questionnaires were completed. The approximate population size of 120,000, 95% level of confidence, and 5% margin of error were used to determine the required sample size. Repeat visits to households that did not respond were used to reduce nonresponse. If an individual selected respondent declined or could not be reached, another eligible household head was randomly selected from the population using the same sampling method. In the end, there were 390 completed questionnaires that were valid for analysis.

A questionnaire comprising ten items that measured the level of concern of the respondents about the selected disasters was used. The items were each presented as a statement, “To what extent are you concerned about…?” All participants were asked to answer all ten items on a five-point scale (Extremely Concerned = 5, Concerned to a Greater Extent = 4, Moderately Concerned = 3, Somewhat Concerned = 2, and Not Concerned = 1). Thus, higher scores showed greater concern. A pilot test was conducted on the questionnaire before the final survey. Selected items based on pilot feedback, unclear/vague wording, response options, item order, etc., were modified to increase clarity, comprehensibility, and relevance of items.

A structured survey questionnaire was used for data collection. The questionnaires were checked daily to reduce the amount of missing information and to ensure that the data collection was accurate. The surveys were conducted from June 2020 to June 2021 in the study area. All the data were analyzed using SPSS to calculate the mean and standard deviation. The data were pre-processed for missing data, outliers, and inconsistencies prior to statistical analysis. Mean scores of concern for each disaster were compared using bar charts, and a heat map and a radar chart (or spider chart) were drawn to show the multidimensional pattern of overall concerns. A radar chart was generated by averaging the concern score for each disaster in the chart. Perceived concern was greater with increasing distance from the centre. This visualization highlighted the comparative importance of each disaster and assisted in detecting the most important climate risks in the study area quickly.

Informed consent was obtained verbally from all respondents before data collection. Respondents were told that they were contributing to an unpaid study in which they may have withdrawn at any time without penalty.

The mean concern scores varied across the ten natural disasters examined, with driving rain (M = 3.85), tornado/cyclone (M = 3.84), drought (M = 3.73), and flood (M = 3.69) having a high level of concern. The moderate concern group comprised nor’westerly (M = 3.63), cold wave (M = 3.55), and river bank erosion (M = 3.50), all with scores above the center of the scale, while the low scores were for sand heat (M = 3.42), water logging (M = 3.40), and lateral water pressure (M = 3.18), and none were below the center of the scale. The bar chart (Figure 1) shows that driving rain (M = 3.85), tornado/cyclone (M = 3.84), drought (M = 3.73), and flood (M = 3.69) received a high level of concern. The low level of concern was given to lateral water pressure (M = 3.18). This figure presents a representation of these mean scores and shows that although important dissimilarities exist in concern across disasters, the overall pattern of scores falls in the moderately concerned to highly concerned range of the five-point scale, which indicates a high level of awareness of the risk associated with disasters among the people in the study area.

The standard deviations of the ten disasters ranged from 1.06 to 1.36, indicating moderate variation in the level of concern among respondents but not significant dissimilarities. A comparatively low difference (SD = 1.06) was observed for tornado/cyclone, suggesting that the majority of the respondents perceive high risk in a similar way, which could be explained by the high frequency and damage pattern of this incidence being familiar in the study area. Lateral water pressure, on the other hand, showed the highest variation (SD = 1.36), meaning that general awareness and exposure to this disaster had the highest variation among the respondents (Figure 2). Moderate variability was found in the remaining disasters, indicating a general trend of greater disaster concern throughout the sample.

Figure 1. Mean concern level by disaster type.

The level of concern regarding tornado/cyclone had the lowest level of variation (SD = 1.06), indicating that there was a high level of consensus with regard to the importance of the concern. The highest inconsistency (SD = 1.36) was observed for lateral water pressure, indicating either less awareness or a mix of perceptions. Figure 2 shows the reflection of differences in concern between all types of disaster.

Figure 2. Standard deviation of concern levels.

A heat map was created to display the mean concern and the standard deviation by disaster. A heat map can be helpful in discovering patterns in multivariate data sets. The lighter colors in the top row indicate areas of higher concern and the darker, lower colors, areas of lower agreement (Figure 3).

Figure 3. Heat map of concern & standard deviation.

The following radar chart reinforces previous observations, where driving rain, tornado/cyclone, flood, and drought were the most critical disasters (Figure 4).

Figure 4. Radar chart of disaster concern levels.

The results indicate that driving rain, tornado/cyclone, flood, and drought were the disasters that were considered the most serious. The public was well aware of these disasters, as they repeatedly occur and have a trend to be sudden. In many parts of Bangladesh, for instance, floods are effectively an annual event, and their devastating potential is well known. The high concern score for tornado/cyclone and driving rain was because of their sudden onset, which usually gives little warning and can result in serious injuries and property damage. Moreover, this prioritization is consistent with the wider academic discussion about the need to move beyond single-hazard assessments to a combined, multi-hazard approach to effectively tackle the multi-hazard nature of the country’s risk landscape ([87]; [82]; [43]). Current empirical studies support the use of individual hazard assessment, but there is a lack of studies examining concurrent risk perception for different types of disasters, including drought, earthquake, and flood ([62]; [28]; [90]). Moreover, the literature emphasized that connecting risk perception of households with experts’ views is a precondition for applying coherent disaster mitigation strategies that are context-specific ([50]; [31]; [83]). This synthesis is supported by research that highlights the importance of local knowledge gaps in disaster mechanisms, which often stifle the successful application of effective climate change adaptation policies ([32]; [2]; [48]).

Lateral water pressure, sand heat, and waterlogging were lesser concerns in terms of disasters, however. One reason could be a lack of awareness of the public. Technical terms like lateral water pressure might not always be linked to the first thought of danger, although they do have importance in regard to structural safety. A less talked-about phenomenon that can also seem less intimidating than flood or storm is sand heat. This gap in perceived threat matches the empirical evidence that the individual relevance of heat-related cues is far less than that of high-impact events such as flooding ([57]; [81]; [86]). The fluctuating nature of public concerns is further underlined by the fact that the impact of floods and media attention tends to focus on events that are noticeable and therefore received by the public, while other climate-related stressors, such as overheating, are less noticeable and less likely to be perceived by the public ([109]; [68]; [12]). The psychological distance of climate change is further enlarged when it is perceived as remote and abstract, meaning that instinctual emotional responses often required to catalyze individual mitigation actions and policy advocacy are not triggered ([35]; [113]; [94]). Moreover, this psychological distancing is compounded by the perception of risk being distant from one’s social networks or future self, which can lead to a feeling of being destitute and unable to act jointly ([93]; [111]; [18]).

This lack of personal engagement is reinforced by psychological research which proposes that climate change is seen as a distant danger both in time and space and so does not elicit the “gut feeling” necessary to build sustainable behavioral change ([51]; [103]; [94]). These perceptions essentially make the phenomenon abstract, because public concern decreases if it is perceived as a problem that is distant in space or time ([108]; [16]; [91]; [18]).