Revisiting and updating the risk-benefit link : replication of Fischhoff et al. (1978) with extensions examining pandemic related factors

Abstract

The relationship between risk and benefit is complex and has been studied in a number of different ways. Fischhoff et al. (1978) approached the relationship in a novel way, using psychometric analysis to measure public attitudes regarding the perceived risks and perceived benefits of certain technologies and activities, finding a negative correlation between perceived risk and perceived benefit. In a well-powered, pre-registered study we set out to conduct a replication of Fischhoff et al. (1978) with several adjustments and extensions. Using a simplified survey design and improved statistical testing, we conducted an exploratory analysis indicating that (i) participants rated selected technologies and activities from the original study differently with respect to their perceived risks and perceived benefits and (ii) extending the study to Covid- 19 pandemic items, participants also rated such items differently with respect to their perceived risks and perceived benefits. In a second extension, we added a third condition to the study to allow within subjects analyses, which revealed empirical support for the negative correlation on the participant level between perceived risk and perceived benefit for Covid-19 pandemic items on an aggregated basis, r(347) = -0.46, p < .001, 95% CI [-0.54, -0.37], and for the majority of individual items, including Covid-19 vaccines, r(347) = -0.62, p < .001, 95% CI [-0.68, -0.55], general vaccinations, r(347) = -0.53, p < .001, 95% CI [-0.60, -0.45], lockdowns, r(347) = -0.48, p < .001, 95% CI [-0.56, -0.39], and handguns, r(347) = -0.46, p < .001, 95% CI [-0.54, -0.37]. Finally, an additional exploratory analysis of item level correlation revealed strong support for the negative correlation between perceived risk and perceived benefit for our between subjects analysis, r(16) = -0.63, p = .006, 95% CI [-0.85, -0.22], and our within subjects analysis, r(16) = - 0.69, p = .002, 95% CI [-0.87, -0.33].