Statistical Association, Impact, and Relevance of Effect-Size Measures in Meta-Analyses
(A2026-132255)

AUTHORS

Sönke Albers, Kühne Logistics University

ABSTRACT

Correlation coefficients are the most commonly applied effect size measure in meta-analyses when examining the strength of association between variables. If marketing researchers strive to determine the effectiveness of marketing actions on performance outcomes, effect size measures should be made comparable via relativization, as is the case for elasticities. A review of meta-analyses published in top marketing journals reveals that researchers are mostly not concerned with the fit of their effect size metrics. Correlation coefficients differ substantially from elasticities when the coefficients of variation for the dependent and independent variables are not equal or when elasticities are based on coefficients of a multivariate regression. Calculating elasticities requires ratio-scaled variables, while variables found in most marketing studies are interval-scaled constructs. Therefore, non-dispersion-based effect size measures are proposed, utilizing relativization and normalization.