To act or not to act: Warning communication and decision-making in response to weather-related hazards

Abstract

Many severe hydro-meteorological events cause unnecessary losses of life and property despite being well forecasted with accurate and timely disseminated warning information. Apparently, warnings have no inherent value, but only become valuable once they influence people’s decisions. A good warning message (i.e. in terms of accuracy and timeliness) that is not conveyed to the end-user in a way that leads to desired behaviour is useless. In future, the expected losses of bad weather are projected to further increase: climate change causes more intense, frequent and long-lasting extremes, and social vulnerability grows with population growth and spatial expansion into areas impacted by hazards. Thus, there is an urgent need to better understand why some people do not respond to warning messages and to identify effective communication strategies, which provide practical insights for policy- and decision-makers, and help people at risk take better informed decisions. The four contributions that comprise this thesis identify and answer open research questions, which aim at advancing the effectiveness of warning communication and decision-making by increasing the understanding of how message content influence people’s response to warnings. Contribution I shows how better targeted communication strategies can increase individual preparedness levels. With natural hazard management shifting towards more integrated approaches that address the role of private households in reducing risks, there is a need to investigate the readiness of people in adopting risk reduction behaviours. I build on protection motivation theory that I combine with decision-stage theory to develop a dynamic protection motivation framework, which helps to understand the motivations of different individuals to undertake risk reduction measures. Based on a household survey conducted in an Italian community that was recently affected by flash floods, I show that not all people are motivated by the same threat or coping appraisals, but that these differ depending on people’s risk reduction stage, i.e. the type of measures already undertaken. Findings contribute on better targeting long-term communication strategies to individuals at risk, which ought to motivate them to undertake risk reduction measures. Contribution II analyses the effects of inconsistent warning messages on the general public’s response. This inconsistency is the result of the duplication of public weather warnings, which are disseminated by different public and private providers. Therefore, I developed a simple matrix to categorize the level of inconsistency across multiple warnings, addressing both visual and textual inconsistencies. Following a severe rainfall event in Switzerland, I conducted an online experiment with different warning pairs that provided participants with more or less consistent information. Results highlight that inconsistent messages were perceived of lower quality and negatively impacted warning response. People were less likely to engage in self-protective behaviours for different types of inconsistent warnings compared to consistent warnings; the higher the level of inconsistency the poorer was the behavioural response. Contribution III examines how changes in hazard description can help people at risk to translate the warning information into an appraisal of risk. With increasing interests of National Meteorological and Hydrological Services to provide impact-based warnings that focus on the impacts that result from the weather, there is a need to better understand whether these warnings would result in greater self-protective behaviours. I also investigate the individual effect of behavioural recommendations in the warning messages. I conducted the research in Switzerland using an online survey experiment that included a hypothetical decision-scenario. Results show that participants who received impact-based warnings - with and without behavioural recommendations - were more likely to engage in self-protective behaviours than people who received the standard warnings. Messages with impact text and recommendations were also reported to be clearer, easier to understand, and more credible, and people reported to be more concerned for their safety, and better understood the threats and behaviours to engage in. Contribution IV draws on insights from Contribution III, but the research was conducted with a different method. Theory from cognitive science suggest that people can use a fast decision pathway that is largely driven by feelings and/or a slower pathway that is based on rational deliberation to process risk information. While rational deliberation must be the dominant pathway in an imagined situation, the presence of fear that emerges in a real-crisis situation could influence the effectiveness of impact-based warnings and people’s actual decisions. I conduced a field experiment, in which wind warnings were disseminated through an existing smartphone application of a Swiss weather provider (Wetter-Alarm), and real-time data was collected on people’s responses. Results show a domination of the affective model of decision-making when people are under an imminent threat of life. The influence of fear resulted in no effect of impact-based warnings. However, stronger event severity and shorter lead times generated a greater behavioural response. The thesis makes four important contributions to the literature. First, it makes a theoretical contribution as it applies a novel protection motivation framework to identify individual motivations for risk preparedness. Second, the thesis performs the first empirical analysis of the influence of inconsistent weather warnings on people’s behaviours. Third, it adds to the growing literature on impact-based warnings by investigating the individual effects of impact text and behavioural recommendations. Fourth, for the first time the dissertation investigates behaviours in real-world crisis in response to actual weather warnings. Altogether, the findings from the four contributions have implications for practitioners, policy-makers and researchers. First, the protection motivation framework ought to help practitioners to develop more effective strategies that link patterns of perception or capacity to effective patterns of communication and avoid to target large communities at risk with a single communication strategy. Second, public and private weather providers should enhance cooperation and deliver more consistent warning information. If cooperation fails, then policy-makers may perform modifications in the warning process and clarify the stakeholder’s roles and responsibilities. Third, decision-makers should develop and use impact-based warnings with behavioural recommendations only in the case of very severe weather. At the same time, when communicating messages, they should be aware of the importance of affect in making decisions when people are under an imminent threat. Fourth, researchers that investigate responses to threats should use research designs that capture real-world conditions.