Flood adaptation: feedback effects of (non-)protective behaviour on risk perception and other cognitive factors

Individual protective responses, such as preparing emergency devices and retrofitting homes, can help reduce flood damage. Risk perception is often assumed to lead to individual protective behaviour in popular theories, e.g., the Protection Motivation Theory. Yet, empirical studies show weak positive, no, and even negative correlations between risk perception and protective behaviour. This inconsistent relationship is often attributed to a reverse feedback effect, assuming that the implementation of protective measures leads to a reduction of risk perception. The feedback effect has been used in various studies to explain the weak relation between flood risk perception and protective behaviour. However, the existence of the feedback has not been statistically proven in the flood risk domain, yet. Investigating this gap is challenging due to a lack of longitudinal studies, which observe individuals over time, thus being able to establish a temporal relationship between adopted behaviour and risk perception. This study thus explores the reverse feedback effects of protective responses on risk perception, using data from a three-wave longitudinal survey over 1.5 years in Vietnam. With the same reasoning, we further investigate the feedback effects of protective and non-protective responses, e.g., wishful thinking, on other cognitive factors, such as perceived social norms and coping appraisals.

Using structural equation modelling, we do not detect significant reverse feedback effects of protective responses on risk perceptions in our dataset. Neither do we find a feedback effect on social norms and coping appraisals, except for retrofitting homes: this measure increases perceived financial capacity and perceived expectation of the general society on doing so. Given the sample size and the very high retention rate of our dataset (401 initial respondents with a retention rate of 94%), it is highly unlikely that we would not detect a feedback effect with a high or medium effect size. We thus conclude that there is no medium or strong feedback effect on risk perception in our sample. Hence, other explanations should be tested, including the most basic of all: perhaps risk perception is not (always) an important driver of protective behaviour.

By contrast, the adoption of non-protective responses has a highly significant influence on many cognitive factors. Specifically, delaying and denial reduce perceptions of risk, social norms, and coping appraisals, in line with our hypotheses. Wishful thinking, by contrast, increases perceived flood consequences, social norms, and coping appraisals, but reduces perceived flood probability. Fatalism increases the perceived flood consequences and probability, but reduces perceived coping appraisals. We thus recommend that risk communication focuses also on other cognitive factors, such as coping appraisals and social norms, and not only on risk perceptions. More research on the role of non-protective responses in flood risk adaptation and communication is needed to eventually inform policy interventions. The predictive power of behavioural theories may be improved if the relationships between non-protective responses and cognitive factors are better understood. Future studies on other flooding contexts are highly recommended to contextualise our findings.