Interpretive Difficulties in Using Difference Score Measures: A Commentary on Gligorić et al.’s (2025) Country-Level Findings and a Reanalysis of the Data

Abstract

Gligorić et al. reported based on country-level data that poorer and more exploited countries showed more outgroup favoritism on competence ratings and more ingroup favoritism on warmth. However, their analyses used a difference score measure of outgroup favoritism. I discuss problems with this approach, and reanalyze the authors’ data using polyvariate regression, which is specifically meant to test congruence hypotheses as outcomes. This revealed that greater poverty was associated with warmer ratings of the ingroup, but was unrelated to warmth ratings of outgroups. Thus, the relationship between wealth and warmth stereotyping may be driven more by ingroup perceptions than perceptions of outgroups. No other multivariate associations of either exploitation or wealth with trait ratings were statistically significant, although there were strong, but nonsignificant, trends in the data. Finally, I show how difference scores rarely provide statistical information beyond the two individual measures used to create them.