Confirmatory Factor Analysis of the Jouve Cerebrals Crystallized Educational Scale: Evidence for a Three-Factor Structure With a Dominant General Factor

This study examined the latent factor structure of the Jouve Cerebrals Crystallized Educational Scale (JCCES) using confirmatory factor analysis with 1,553 examinees across 129 items organized into three subtests: Verbal Analogies (VA), Mathematical Problems (MP), and General Knowledge (GK). Four competing models were evaluated: a unidimensional model, a three-factor correlated model, a higher-order g model, and a bifactor model. The three-factor correlated model (CFI = .911, RMSEA = .123) fit significantly better than the one-factor model (delta chi-squared = 1,370, delta df = 3, p < .001). Schmid-Leiman decomposition yielded omega hierarchical = .879, omega total = .937, and ECV = .826, indicating a dominant general factor that accounts for 93.7% of the reliable variance in total scores, with the Mathematical Problems subtest retaining the most unique reliable variance (omega_s = .260). Because conventional fit cut-offs are miscalibrated for parcelled binary indicators, a benchmark simulated from a model correct by construction attains RMSEA = .063 against the observed .123, indicating that the elevated RMSEA is part artefact and part genuine misspecification.