An Alternative Approach to Pooling Covariance Matrices in Meta-Analytic Confirmatory Factor Analysis

• In Event: 65.079 - Roundtable Session 23
  In Roundtable Session: 65.079-19 - Topics in Structural Modeling

Abstract

As instruments are created and validated, there is often controversy surrounding their latent factor structure. This controversy sometimes stems from the fact that exploratory or confirmatory factor analytic findings may vary across samples. One solution to this problem is to evaluate the latent factor structure of the instrument across numerous samples using meta-analytic techniques to better estimate the population covariance structure. However, such techniques can be unnecessarily complex when all items for an instrument are available. The present study is an empirical evaluation of an alternative approach for pooling covariance matrices in meta-analytic confirmatory factor analysis for cross-sample testing of latent structures using multiple published factor analyses using simulation studies.

Authors

• Kamden Strunk, Auburn University

• Forrest Lane, Sam Houston State University