Ordinal measurement invariance of three retained GAEU-1 dimensions in Chilean and Ecuadorian university samples.

Abstract

Meaningful cross-national comparisons of self-regulated learning (SRL), motivation, and related learning-regulation constructs require evidence that the measurement model operates comparably across groups. We tested this requirement for three GAEU-1 dimensions previously retained in Chile and Ecuador using the open Version-5 dataset (N = 1,316; Chile n = 631; Ecuador n = 685). Fifteen ordered indicators representing the GAEU-1 SRL factor, conscious motivational strategies, and deep learning techniques were analyzed with WLSMV, followed by a Wu-Estabrook multigroup ordinal invariance sequence and complementary DIF analyses. Country-specific fit was mixed: CFI and SRMR were relatively favorable, whereas RMSEA indicated residual misspecification, especially in Chile (CFI = 0.947, SRMR = 0.059, RMSEA = 0.094; Ecuador: CFI = 0.952, SRMR = 0.059, RMSEA = 0.085). In the multigroup models, equality constraints on thresholds, loadings, and intercept-related parameters produced no meaningful deterioration in approximate fit, and met the study's prespecified incremental-fit criteria for proceeding to latent comparison within the retained scoring representation. Several items showed FDR-adjusted DIF signals, but none reached the prespecified 0.020 practical-effect criterion. Post hoc diagnostics localized the largest residual association to GA7-GA8 in both samples; a one-factor model fit appreciably worse than the retained three-factor representation (Delta CFI = -0.059 in Chile and -0.069 in Ecuador). Polychoric HTMT values were all below 0.85, although Fornell-Larcker diagnostics remained unfavorable for factor pairs involving the GAEU-1 SRL factor. Observed and latent standardized sample contrasts differed only modestly and are interpreted descriptively because the samples differed markedly in university sector and gender composition. The two planned sensitivity analyses and a post hoc private-university-only sensitivity analysis preserved the main comparability result. Overall, the three retained GAEU-1 dimensions showed stable measurement parameters across these Chilean and Ecuadorian university samples, while broader structural and country-level interpretations remain outside the evidence provided here.