Calculating Repeated-Measures Meta-Analytic Effects for Continuous Outcomes: A Tutorial on Pretest–Posttest-Controlled Designs

Abstract

Meta-analysis is a statistical technique that combines the results of multiple studies to arrive at a more robust and reliable estimate of an overall effect or estimate of the true effect. Within the context of experimental study designs, standard meta-analyses generally use between-groups differences at a single time point. This approach fails to adequately account for preexisting differences that are likely to threaten causal inference. Meta-analyses that take into account the repeated-measures nature of these data are uncommon, and so this article serves as an instructive methodology for increasing the precision of meta-analyses by attempting to estimate the repeated-measures effect sizes, with particular focus on contexts with two time points and two groups (a between-groups pretest–posttest design)—a common scenario for clinical trials and experiments. In this article, we summarize the concept of a between-groups pretest–posttest meta-analysis and its applications. We then explain the basic steps involved in conducting this meta-analysis, including the extraction of data and several alternative approaches for the calculation of effect sizes. We also highlight the importance of considering the presence of within-subjects correlations when conducting this form of meta-analysis.