Beyond monetary outcomes: the probability-range reflection effect across decision domains

The reflection effect, a key component of prospect theory, highlights the reversal of risk preferences when decision outcomes change from gains to losses. The effect’s potential pervasiveness and impact on everyday decision-making necessitates a nuanced characterization. As emerging evidence demonstrates the domain-specificity of risk preferences, this study extends the investigation of the reflection effect beyond monetary outcomes to multiple life domains and proposes a probability-range approach to studying the reflection effect. In a between-subjects stated-preference experiment, we tested whether the probability-range reflection effect exists in both financial (gambling, investment) and non-financial (social, recreational, health, ethical) domains. By varying probability levels, we estimated the shapes of choice-probability curves using logistic regression models. Participants (525 adult US residents, representative in age and sex; 51.6% women, mean age 46 years, SD = 16, range = [18, 82]) saw vignettes corresponding to six risk domains of the Domain-Specific Risk-Taking scale. Each vignette was randomly assigned to a gain or loss prospect and one of nine probability levels. Based on each vignette, participants chose either a safe (certain) outcome or a risky lottery. Comparing the proportions of selected risky lotteries across all probability levels in gain and loss prospects allowed us to characterize the reflection effect on a domain-by-domain basis. We detected the effect in all except the ethical domain, where participants exhibited strong risk aversion across all probability levels and prospects. Notably, we observed an “opposite reflection effect” in the social domain, with higher risk-taking for gains than losses. These findings reveal important domain-specific variations in the reflection effect. Our probability-range approach provides a more nuanced characterization of risk preferences, challenging the universality of the reflection effect and suggesting the need for domain-specific considerations in decision-making models. These results have significant implications for decision theory and the design of interventions targeting risk-related behaviors across various life domains.