Heterogeneity Coefficients for Mahalanobis' D as a Multivariate Effect Size. (4th March 2017)

• Record Type: Journal Article
• Authors: Del Giudice, Marco
• Is Part Of: Multivariate behavioral research. Volume 52:Number 2(2017)
• Page Start: 216
• Page End: 221
• Publication Date: 2017-03-04
• Subjects: Effect size -- group differences -- heterogeneity -- Mahalanobis distance -- multivariate; Psychometrics -- Periodicals; Psychology, Experimental -- Periodicals; Psychology, Experimental; Psychometrics; Periodicals; 150.15195
• DOI: 10.1080/00273171.2016.1262237
• Languages: English
• ISSNs: 0027-3171

Abstract

ABSTRACT: The Mahalanobis distance D is the multivariate generalization of Cohen's d and can be used as a standardized effect size for multivariate differences between groups. An important issue in the interpretation of D is heterogeneity, that is, the extent to which contributions to the overall effect size are concentrated in a small subset of variables rather than evenly distributed across the whole set. Here I present two heterogeneity coefficients for D based on the Gini coefficient, a well-known index of inequality among values of a distribution. I discuss the properties and limitations of the two coefficients and illustrate their use by reanalyzing some published findings from studies of gender differences.