Meta-analytic reviews must exclude regression-only studies because effect sizes require zero-order correlations

Evidence

We decided upon correlation coefficients for our effect size measure, so any potential study had to include a correlation coefficient describing the relationship between a type ofthreat and outgroup attitudes or provide enough information to calculate one (e.g., t-values, F-values, means and SD, etc.). We chose to use correlation coefficients because they are commonly used as a measure of effect size in meta-analyses and were reported in the vast majority of studies. This required the exclusion of any study that only reported results from multiple regression or structural equation modeling analyses, which are not useable in a meta-analysis (Lipsey & Wilson, 2001). — Blake M. Riek, Eric W. Mania, and Samuel L. Gaertner, Intergroup Threat and Outgroup Attitudes, 77

Categorization

• XREF: Methodological constraint relevant to the MOC’s empirical backbone — shapes how the threat-attitude effect sizes were aggregateable at all. — connects to: Conceptual heterogeneity, not just design variance, makes meta-analysis impossible. — Lea Echelmeyer, Anne-Marie Slotboom, and Frank Weerman, The Putative Effect of Identity on…, 30; A dispositional spillover thesis needs psychological data, not correlational evidence. — Russell Hardin, Trust and Trustworthiness, 739
• QUESTION: Worth checking how many threat-attitude studies were dropped for reporting only SEM/multiple regression — a silent source of selective exclusion in this literature.

Broader Connections

• Conceptual heterogeneity, not just design variance, makes meta-analysis impossible. #qualifies (0.80) — Lea Echelmeyer, Anne-Marie Slotboom, and Frank Weerman, The Putative Effect of Identity on…, 30
• A dispositional spillover thesis needs psychological data, not correlational evidence. #qualifies (0.72) — Russell Hardin, Trust and Trustworthiness, 739