Synthesizing data from pretest-posttest-control-group designs in mediation meta-analysis

This study aims to address the existing concerns for mediation meta-analysis (MMA) conducted under pretest-posttest-control-group (PPCG) designs. These issues include the validity of meta-analytic structural equation modeling with a binary independent variable, heterogeneously defined treatment doses, and potential violations of the homogeneous variance assumption. Moreover, an open question remains regarding whether effect sizes in MMA over PPCG designs should be computed from posttest scores (PSMMA) or from pretest-posttest change scores (CSMMA). This study employs theoretical discussions and Monte Carlo simulations to demonstrate the feasibility of MMA under PPCG designs and to compare CSMMA and PSMMA. Furthermore, the procedures for computing effect sizes are summarized, and an empirical example is presented.