A meta-analysis on study and sample characteristics modulating mock earwitness performance.

Abstract

Mock earwitness performance mainly addresses cognitive functions like free recall and recognition of auditory information. Based on primary studies a-priori experimental moderator effects on mock earwitness performance have been hypothesized. Including 46 articles with k = 66 experimental studies, a bare-bones and a random-effects, artefact-corrected meta-analysis have been performed. The results show a substantial ratio of the population effect size and the standard deviation of the population effect size (δ/SDδ) for the a-priori moderators bimodal compared to unimodal stimuli and for gender of listener. These results indicate that bimodal stimuli compared to unimodal stimuli yield substantially better mock earwitness performance. Women outperform men in mock earwitness performance. The fail-safe number demonstrates robust population effects for both a-priori moderators. As a post-hoc moderator, line-ups with target-present revealed substantial δ/SDδ ratios for stimulus length and gender of listener. These results for post-hoc moderators suggest that longer compared to shorter stimuli reveal better mock earwitness performance. The gender of listener effect is in target-present line-ups comparabale to the before mentioned gender of listener effect. G*Power calculations for future primary studies suggest that sample sizes in several primary studies were too small. Single-person earwitness assessments should be substantiated by reality monitoring and disentangled from hearsay evidence (i.e., situations in that individuals have heard information from other persons). The data recommend that best-practice options for earwitnesses in court settings should be derived from meta-analytic results, corresponding to the jurisdictions of the countries and exclusively for results that were substantial (δ/SDδ ratios) and robust (fail-safe number).