An illustration of meta-analytic structural equation modeling in health psychology

Abstract

Glass (1976) introduced Meta-Analysis (MA) as a way to combine results from previous studies to discover systematic patterns among the studies and to produce a more comprehensive summary of these studies. The goal in MA is to estimate the true effect by pooling effects from different studies. While conventional MA techniques provide a lot of information about mean differences and relationships, they have to be extended to be able to analyze more complex models. While structural equation modeling (SEM) has allowed researchers to test complex relationships among a multitude of variables, no single SEM study can provide complete evidence for the hypothesized relationships being examined. It is, therefore, important to produce multiple studies and to have the ability to synthesize these studies to be able to provide comprehensive answers to research questions. Meta-analytic structural equation modeling (MASEM) combines the techniques of MA and SEM in order to synthesize correlation or covariance matrices from primary studies and fit SEM models using the pooled matrix. The aim of the current presentation is to provide guidance on how to apply MASEM particularly in health psychology. In this study we employed the two-step MASEM approach proposed by Cheung and Chan (2005) to test the theory of planned behavior (TPB) framework proposed by Ajzen (1985) in the context of alcohol consumption. The results confirm the predictive validity of TPB in predicting health compromising behavior. By employing MASEM, the study provides more generalizable conclusion regarding the nature of the effects of TPB variables on alcohol consumption.