Interdependency of Risk Perceptions
How to cite
Qari, Salmai; Lohse, Tim; Bolouri, Armin. 2025. "Interdependency of Risk Perceptions." AEA RCT Registry. AEARCTR-0017240. https://doi.org/10.1257/rct.17240-1.0
Download: BibTeX · RIS — for reference managers without a browser translator. Zotero and most others pick up this page's citation metadata directly.
* Primary investigator
Abstract
This study analyzes the possible interdependence of individuals' risk perceptions across different domains. For example, information concerning personal risk in the health domain (e.g., a cancer diagnosis) may cause spillovers to other domains, for example the worry concerning crime. The experiment investigates whether these spillover-effects vary depending on the type of risk/event.
The survey experiment will employ a combination of between- and within-subject design. First, subjects are randomly allocated in three treatments concerning the information provision during the experiment. The information provision in the three treatments refers to 1) societal events (e.g. international tensions), 2) personal events, e.g. own health and 3) a neutral information. Experimental subjects are asked to state their subjective worries in various domains before and after the information provision which allows to measure within-subject differences induced by the information provision. The experiment further contains a budget allocation task concerning public spending and elicits risk aversion, ambiguity aversion, personality variables and demographic variables. The experiment will allow to measure to what extent risk perceptions are correlated across different domains, and whether the degree of correlation depends on the type of event/risk (e.g., societal vs. personal).
Intervention
The intervention is carried out by randomly assigning the subjects in the survey to different types of information provision, (i) societal events/risk, (ii) personal events/risk and (iii), a neutral text. Subjects state their subjective worries in various domains before and after the information provision which allows to measure within-subject differences induced by the information provision. Further, experimental subjects make choices in two versions of a budget allocation task concerning public spending.
The experimental design allows to test how the different information treatments affect the disribution of subjective worries and the budget allocations.
Experimental design
We conduct an online survey experiment with 1200 respondents/observations in Germany. The experiment examines whether spillover effects in risk perception depend on the type of the assessed risk. To this end, three treatments are developed which provide information either for (i) a societal event/risk, (ii) a personal event/risk and (iii) a neutral event/information (see the interventions above). In all three treatments, subjects are asked to state their subjective worries in the same set of domains (e.g., crime, economic situation, climate). These worries are measured before and after the information provision which allows to use differences for each subject, thereby eliminating individual fixed effects. The two budget allocation tasks measure the policy priorities concerning public spending for each experimental subject. In the first task, the budget is fixed and subjects can distribute 100 points (100 percent) to three policy areas (economy, health, security). In the second task, the budget is not fixed such that subjects can choose whether the budget for spending should be reduced / stay the same / expanded, and at the same time subjects distribute this budget to three policy areas. This part of the experiment allows to test the link between the type of risk / event and preferences for public spending. Finally, the remaining part of the survey contains standard tasks to elicit risk aversion, ambiguity aversion and questions for measuring personality and various attitudes. We hypothesize that there is considerable heterogeneity on the subject-level concerning the degree of spillover to other risk domains, and we will investigate whether this heterogeneity can be linked to personality.
The sample will be representative concerning age and gender (using quotas) for the German population, and randomization occurs at the subject/individual-level.
Design details
Cited by
No citation data available for this registration yet.