Concealed Correlations Meta-Analysis Approach: A New Method for Synthesizing Standardized Regression Coefficients

• In Event: Marriott Roundtable Session 65
  In Roundtable Session: New Methods for the Meta-Analysis of Regression Coefficients

Abstract

The meta-analysis of standardized regression coefficients is still a challenging issue due to the difficulty of combining beta weights when regression models include different covariates. In this study, a new Concealed Correlations Meta-Analysis (CC-MA) approach is proposed. A simulation study compared the performance of the CC-MA approach with two other approaches: (1) doing separate meta-analyses for beta weights that belong to the same regression model and (2) coding an indicator variable indicating the presence of a covariate in the regression model, and then performing a meta-regression. Results showed that although the meta-regression approach led to a better standard error estimates when a random effects model was fitted, the CC-MA approach led to more accurate estimates of the combined beta weights.

Authors

• Belén Fernández-Castilla, University of Leuven

• Ariel M. Aloe, University of Iowa

• Susan Natasha Beretvas, The University of Texas - Austin

• Patrick Onghena, KU Leuven

• Wim Van den Noortgate, Katholieke Universiteit Leuven