Methods: We followed a process used in industrial-organizational psychology and applied it to data from 11,413 youth in 55 randomized controlled anxiety and depression treatment and prevention trials. Broadly, pre to post, post to follow-up, and pre to follow-up effect sizes (ESs) for mediator and outcome variables were calculated using the independent groups pre-test post-test procedure and random effects models. Pre to post and post to follow-up ESs for each mediator and pre to follow-up ESs for outcomes were transformed into correlation coefficients. Using the resulting 3x3 correlation matrices, regression analyses were conducted to obtain the regression coefficients and standard errors for the intervention to mediator path and the mediator to outcome path, controlling for the intervention to outcome path. RMediation was then used to estimate indirect effects and 95% confidence intervals.
Results:Indirect effect estimates showed that changes in information-processing biases and coping strategies mediated changes in outcomes for anxiety and depression at both levels of intervention. Gains in social competence and reductions in physiological hyperarousal did not emerge as significant mediators. This approach also showed that most studies failed to report critical information for meta-analysis, including some descriptive statistics, correlations, and measurement reliabilities.
Discussion: This is the first study to use meta-analytic path analysis to evaluate support for theory-driven mediators of youth interventions targeting anxiety and depression. From this research, it is clear that combining meta-analytic procedures with path analysis provides a unique opportunity to identify promising mediators that guide future research efforts. To facilitate use of meta-analytic path analyses of other interventions, a step by step outline of procedures will be provided.