Why do people not prepare for disasters? A national survey from China

The study addresses the persistent problem that despite evidence that pre-disaster mitigation and preparedness reduce losses, household preparedness remains suboptimal. In the context of increasing risks under climate change, understanding facilitators and barriers to preparedness is critical. The paper focuses on household-level preparedness actions and the reasons for non-adoption, aiming to answer: (1) What are the primary reasons for not adopting specific preparedness actions? (2) Does trust in government affect adoption of preparedness actions? The work situates preparedness within widely used frameworks and emphasizes the central role of households in community preparedness efforts.

Two conceptualizations of preparedness are discussed: a broad framework encompassing mitigation, prevention, protection, response, and recovery (FEMA), and a more traditional view of capacities and knowledge developed pre-disaster (e.g., planning, learning, drills). Household preparedness research commonly uses social-psychological models including PADM, HBM, SCM, ToPB, and PMT. Influencing factors typically fall into four groups: (1) socioeconomic and demographics (income, education, gender, dependents), (2) hazards/context (environmental cues, place attachment, disaster experience), (3) cognitive/psychological processes (risk perception, responsibility attribution, trust), and (4) social and structural factors. Prior studies show mixed effects of disaster experience and nuanced roles of trust in authorities. Response and self-efficacy often predict protective behaviors more than threat perceptions, and community engagement can facilitate preparedness. Nonetheless, barriers that prevent translation of intentions into actual behaviors are under-investigated.

Design and sampling: A national, cross-sectional, online survey was conducted in mainland China (August–September 2018) using stratified sampling. The capital and one additional prefecture-level city were selected per province (two extra prefectures in Yunnan and Jiangxi for comparison). Quota sampling targeted 100 respondents per city (200 in Beijing, Shanghai, Chongqing, Tianjin). Of 6611 returns, 81 were excluded; the final sample included 6530 respondents from 60 cities across 31 provinces. Ethical approval was obtained from the Institute of Geophysics, China Earthquake Administration; participation was voluntary and anonymous. Measures: Preparedness behaviors were measured as actual actions (yes/no) across seven items: (1) preparing emergency supplies (food/water for ~3 days), (2) paying attention to disaster-related information, (3) preparing a written family emergency plan, (4) knowing nearest emergency shelters, (5) knowing seismic building code requirements for the region, (6) participating in emergency-related training or drills, (7) volunteering for emergency-related activities. A summed index (0–7) captured overall preparedness (mean 5.09, SD 2.23). For each action answered “no,” respondents selected a primary reason from ten options: costly, needs knowledge/technology, time-consuming, need collaboration, energy-consuming, not aware, limited function/not necessary, don’t know where to buy/learn/reach, not my responsibility, other (open text). Covariates: Gender (male=1), age group, minority status (1), education, marital status (married=1), children at home (1), elders at home (1), rural (1), perceived socioeconomic status (1–5), disaster experience (natural disaster, H1N1/H1N5), relocation due to disaster (1), and trust in government. Trust in government summed trust in five levels (central, provincial, county, township, village/community; each 1–5), total 5–25 (Cronbach’s alpha=0.857). Analysis: Descriptive statistics summarized sample and preparedness prevalence. A two-level linear mixed model (province-level random effects) estimated associations with overall preparedness. Seven two-level logistic mixed models (melogit) estimated associations with each preparedness action. Stata 16.0 was used. Provinces were controlled as higher-level units.

Sample: N=6530; 59.95% male; age <30: 40.03%, 30–60: 59.0%, >60: 0.96%; 7.14% minority; 79.10% married; 20.47% with children at home; 24.61% with elders at home; 14.24% rural. Education: primary/below 0.63%, middle 3.86%, high 18.21%, college 72.54%, graduate 4.76%. Mean perceived SES 2.91 (SD 0.77; 1–5). Disaster experiences: natural disaster 14.75%, H1N1/H1N5 7.12%, relocated 14.75%. Trust in government mean 19.88 (range 5–25). Prevalence of preparedness actions: paying attention to disaster information 90.84%; prepared emergency supplies 76.60%; know building code 73.75%; know shelter 75.53%; family emergency plan 72.22%; participated in exercise/drill 66.26%; volunteer experience 54.10%. Primary reasons for non-adoption (selected examples, % among non-adopters):

• Not aware: material 44.83%; attention 38.46%; plan 33.74%; shelter 35.61%; building code 28.24%; drill 13.98%; volunteer 13.41%.
• Don’t know where to buy/learn/reach: plan 16.81%; shelter 36.92%; building code 32.85%; drill 40.26%; volunteer 35.92%; material 9.75%; attention 9.36%.
• Not necessary: material 20.68%; attention 18.56%; plan 9.21%; shelter 5.57%; building code 3.68%.
• Other notable barriers: need knowledge (building code 12.19%); not my responsibility (plan 12.51%); energy consuming (volunteer 14.94%); lack of time (drill 9.03%, volunteer 9.91%); need collaboration (drill 10.44%). Overall preparedness (multilevel linear model): Higher preparedness associated with being male, minority, married, higher SES, higher trust in government, and disaster-related relocation; lower preparedness associated with older age/elders, rural residence, and natural disaster experience (when not severely impacted). Education, children at home, and H1N1 experience were not significant predictors. Specific actions (multilevel logistic models; odds ratios [OR]):
• Trust in government positively associated with all seven actions (OR ~1.17–1.24 per unit increase).
• Relocation due to disaster strongly positive across all actions (OR 2.17–5.64).
• Higher SES positive for all actions (OR 1.22–1.75).
• Rural residence negatively associated with all actions (OR 0.45–0.70; volunteer 0.52).
• Education: positive for paying attention (OR 1.26), drills (OR 1.10), volunteering (OR 1.11); negative for knowing community plan (OR 0.89).
• Age: older groups less likely to have supplies (OR 0.81), plan (0.63), drill (0.75), volunteer (0.77).
• Gender: males more likely for supplies (1.27), shelter (1.29), building code (1.22), drill (1.13), volunteer (1.22); not significant for paying attention and knowing plan.
• Families with elders: lower odds for plan (0.79), shelter (0.74), building code (0.64), volunteer (0.78).
• Natural disaster experience: lower odds for plan (0.79), shelter (0.71), building code (0.57), volunteer (0.54); not significant for supplies, attention, drill.
• Children at home and H1N1 experience showed no significant associations across actions. Overall, the top barriers are lack of awareness, limited access/knowledge of where to obtain resources or information, and perceived lack of necessity; trust in government, urban residence, higher SES, and relocation experience increase the likelihood of preparedness across behaviors.

The findings directly address the research questions. First, the most common reasons for not engaging in preparedness are lack of awareness, not knowing where to obtain products or information, and beliefs that actions are unnecessary. This suggests that many households do not progress from intention to behavior due to information and access gaps and low response efficacy. Second, trust in government is consistently and positively associated with preparedness across all actions, indicating that confidence in authorities may facilitate acceptance of preparedness messaging and engagement with recommended actions. Structural and social determinants also matter: rural residence and lower SES are associated with lower preparedness, highlighting inequities in access and outreach. Disaster experience without severe impact can reduce future preparedness, whereas relocation due to disasters is associated with substantially higher preparedness, underscoring the importance of how experiences are operationalized. These results emphasize the need for targeted community outreach, risk communication, and accessible channels for acquiring preparedness resources and skills.

This national study of China provides comprehensive evidence on household disaster preparedness behaviors, the reasons for non-adoption, and the correlates of engagement. The principal barriers—lack of awareness, limited access to resources/information, and low perceived necessity—point to the critical role of community outreach and public education. Trust in government, urban residence, higher socioeconomic status, and disaster-related relocation are associated with higher adoption across seven preparedness actions. The study contributes rare national-level evidence from China and highlights the translation gap between intention and behavior. Future research should use longitudinal and survey-experimental designs to establish causal relationships, integrate geospatial data to contextualize preparedness, and examine population-specific barriers and facilitators across disaster types and countries.

The cross-sectional design limits causal inference. Preparedness measures were not linked to geospatial data, constraining contextual analyses. The study did not disaggregate the severity of disaster experiences beyond relocation, which may influence the observed mixed associations. Large-scale experimental designs are costly; the authors recommend survey-experimental approaches and deeper subgroup analyses of reasons for non-adoption in future work.