Abstract

Objectives: The present study aims to bridge this theoretical and methodological gap through an advanced empirical investigation that examines the relationship between the Expected Cross-Validation Index (ECVI) as a predictive indicator of generalizability, and convergent and discriminant validity indices as traditional measures of construct quality. Methods: The Self-Closure Scale was administered to a sample of 788 university students. A total of 500 bootstrap samples (n = 300 each) were generated using random sampling with replacement. For each bootstrap sample. Results: The results revealed no statistically significant relationship between the level of ECVI and convergent validity (r = 0.08, p = 0.12), nor between ECVI and discriminant validity (r = 0.05, p = 0.29). Conclusions: These findings indicate that ECVI and traditional indicators of convergent and discriminant validity represent methodologically independent dimensions of model evaluation. While ECVI primarily reflects the stability of model fit and generalizability across samples, conventional validity evidence focuses on the accuracy of theoretical construct representation.