Reproducibility of individual effect sizes in meta-analyses in psychology

• DOI: 10.1371/journal.pone.0233107
• Publication Date: 2020-05-27T17:25:35Z
• Authors: Esther Maassen; Marcel A. L. M. v...; Michèle B. Nuijten; Anton Olsson-Coll...; Jelte M. Wicherts
• Keywords: Science; General Biochemistry,Genetics and Molecular Biology; Q; 05 social sciences; R; Reproducibility of Results; reporting errors; meta-analysis; Meta-Analysis as Topic; Confidence Intervals; Medicine; Psychology; 0501 psychology and cognitive sciences; General Agricultural and Biological Sciences; General; reproducibility; effect sizes; Research Article

To determine the reproducibility of psychological meta-analyses, we investigated whether could reproduce 500 primary study effect sizes drawn from 33 published meta-analyses based on information given in and recomputations altered overall results meta-analysis. Results showed that almost half (k = 224) all sampled not be reproduced reported meta-analysis, mostly because incomplete or missing how studies were selected computed. Overall, this led to small discrepancies computation mean sizes, confidence intervals heterogeneity estimates 13 out meta-analyses. We provide recommendations improve transparency reporting entire meta-analytic process, including use preregistration, data workflow sharing, explicit coding practices.