Abstract
Comparative optimism (believing one’s own risk is lower than others’) has been widely studied. Nonetheless, important methodological features have largely remained unrecognized. Focusing on a single event (contracting COVID-19), we estimated the average effect, tested moderators with attention to measurement artifacts, and appraised methodological quality. We synthesized 135 effect sizes from 50 studies (38 articles; sample sizes summing to 88,935, not necessarily unique participants) using a multilevel model with random effects for study, article, and nationality. We then examined moderators at both aggregated and individual-participant levels. Although, on average, comparative optimism was observed (d = 0.334, 95% CI [0.225, 0.444]), true effects varied substantially, and the meta-analytic prediction interval ranged from −0.371 to 1.040. Strikingly, this substantial heterogeneity was observed despite our efforts to control for various potential sources of heterogeneity during data collection. Measurement format was the most consistent moderator, with indirect assessments yielding larger adjusted effects than direct assessments, whereas longer response scales were not consistently associated with smaller effects. Effects were larger in samples collected in 2021 than in 2020, but this exploratory finding was sensitive to one article. The association with national collectivism was unclear in the primary model. Individual-participant analyses showed no clear overall associations for age, gender, or education. Response distributions indicated that nearly half of participants reported lower personal risk than others, more than one third reported equal risk, and about one sixth reported higher risk. Small-study and selection-model analyses suggested possible selective reporting. In sum, although we observed an average comparative optimism effect for contracting COVID-19, the generalizability of the effect and its measured correlates is limited by substantial heterogeneity and the influence of measurement choices and study context.