Revisiting the multidimensionality and measurement of hostile masculinity: testing bifactor s-1 cfa and esem models

Abstract

The Confluence Model of sexual aggression identifies hostile masculinity (HM) as a central risk factor characterized by hostility toward women (HTW), adversarial sexual beliefs (ASB), and sexual dominance (SOD). Despite its theoretical importance, the dimensionality and measurement of HM remain insufficiently evaluated, with prior research relying on inconsistent operationalizations and composite scoring practices. This dissertation refined the measurement of HM in a three-study psychometric investigation. Study 1 applied integrative data analysis (IDA) to two existing datasets (total N = 720), using the graded response item response theory model (GRM-IRT) to evaluate item performance and differential item functioning (DIF) analyses to assess measurement equivalence across two datasets. Items showing weaker psychometric performance were removed to develop shortened measures of the three HM components: HTW, ASB, and SOD. Study 2 used the same pooled dataset to compare higher-order, bifactor, and bifactor S-1 specifications estimated using confirmatory factor analysis (CFA) and exploratory structural equation modeling (ESEM). Results indicated that the ASB- anchored bifactor S-1 CFA model demonstrated the best balance of model fit, parsimony, and interpretability, χ2(174) = 603.34, CFI = .943, TLI = .931, RMSEA = .067, SRMR = .033, supporting a general HM factor anchored in ASB and three residual-specific factors representing SOD and two HTW dimensions. In Study 3 (N = 329), bivariate correlations and cross-validated regression analyses supported the convergent and nomological validity of the ASB-anchored model relative to an HTW resentment- anchored alternative and a conventional unidimensional HM total score. Together, the findings support the ASB-anchored bifactor S-1 CFA model as a parsimonious and theoretically interpretable representation of HM.