Do intentions indicate actual behaviour? A comparison between scenario‐based experiments and real‐time observations of warning response

The sharp increase in the use of smartphones and rapid advances in mobile communication offer new ways to warn the public about developing natural or technological hazards. So far, the effectiveness of different warning types, namely standard and impact‐based warnings (SW and IBW), were mainly tested in scenario‐based experiments and not in real life. However, it can be questioned whether scenario‐based experiments are the appropriate methodology to test different warnings, for instance due to the absence of feelings catalysed by real events. Therefore, we collected information about warning responses in real time via the smartphone application ‘Wetter‐Alarm’ of a Swiss weather provider. In the first phase of the study, we conducted a field experiment to investigate actual responses for SW and IBW. In the second phase, we compared these results with behavioural intentions collected via a scenario‐based experiment with an almost identical set‐up. The comparison shows that warning messages were perceived and understood very similar in both experiments. Also, we did not observe any significant interactions between warning type (SW vs. IBW) and study (field vs. scenario‐based experiment). These findings indicate that the effects of warning type on behaviour can be studied on the basis of experimental (i.e. scenario‐based) research designs. The paper ends with some reflections on the potential of big data in the social sciences and on a research agenda for real‐time data collection to improve warning effectiveness and, ultimately, climate risk management.