What Fifty Years of Research on the Misinformation Effect Can Tell Us: A Meta-Analytic Review

BackgroundThis meta-analysis investigates the impact of misleading or incorrect information on subsequent reporting of witnessed events, with a focus on identifying moderators that influence the misinformation effect.MethodsWe included experimental studies following the three-phase misinformation effect paradigm published between 1975 and 2022. Data were analyzed using a three-level mixed-effects meta-regression model, accounting for effect interdependence and key moderators such as retention intervals, warnings, overall test performance, and study characteristics. ResultsAnalyzing data from 433 studies and 1,764 effect sizes across 51,845 participants, we found the average size of the misinformation effect substantial, g = 0.739, 95% CI = [0.258, 1.220]), with considerable variation across different conditions, total σ2 = 0.777, I2 = 92.25%. Consistent with previous findings, accuracy for the control condition was strongly positively associated with the size of the misinformation effect. Post-misinformation warnings were found to reduce the misinformation effect. Additionally, post-event testing, contrary to previous findings on retrieval-enhanced suggestibility, did not significantly moderate the misinformation effect.DiscussionThe present analysis supports the conclusion that the misinformation effect is a reliable phenomenon, driven by a mixture of memory-related mechanisms as well as mechanism related to how participants strategize when responding to tests of different formats. Additionally, given the substantial heterogeneity we observed, we argue that the misinformation effect might be usefully conceptualized as a constellation of related phenomena, rather than a single coherent effect.