A Meta-analysis of the Peak-End Rule and its Boundary Conditions

Balca Alaybek

Major Professor: Reeshad S. Dalal, PhD, Department of Psychology

Committee Members: Lauren Kuykendall, Seth Kaplan

Online Location, Online
August 03, 2020, 12:00 PM to 02:00 PM


According to the peak-end rule (Kahneman, 1999; 2000), when evaluating past experiences retrospectively, instead of weighting the intensities of all episodes in the experience equally, people give more weight to the episode with the peak intensity and to the final episode. I conducted a meta-analysis of the peak-end rule and its boundary conditions. In addition, I compared the peak-end effect to the duration effect, the effects of “Gestalt characteristics” (i.e., beginning, trough, trend, and variability; Ariely & Carmon, 2000), and the overall average effect. Based on the analyses using effect sizes from 85 studies reporting a peak effect and 95 studies reporting an end effect, I found that the peak-end effect was large (r = .57) and robust across most demographic, conceptual, and methodological moderators examined. In addition, the peak-end effect was stronger than the effect of the duration, which was essentially null (supporting the idea of duration neglect; Fredrickson & Kahneman, 1993). The peak-end effect was also stronger than the effects of the other Gestalt characteristics, and indeed was comparable to the effect of the overall average score across the entire experience. This meta-analysis demonstrates that the peak-end effect is an important and robust heuristic when evaluating past experiences. I end with an agenda for future research on the peak-end rule and a description of the important practical implications of this rule.