When r and β disagree: suppression-aware interpretation of personality-learning links in Vietnamese EFL classes

Personality effects in English as a foreign language (EFL) studies are often read from zero-order correlations even though Big Five traits and classroom outcomes are both intercorrelated. Using data from 103 first-year Vietnamese EFL students, this study examines whether such bivariate readings can obscure suppression-compatible patterns. Students completed the BFI-44 and a classroom-adapted inventory of group, individual and kinesthetic preferences, together with a collaboration-oriented self-regulated learning scale. Simultaneous five-trait regressions yielded two notable r → β divergences. Agreeableness was essentially unrelated to individual preference at the zero-order level (r ≈ .02) but became negative after the other four traits were controlled (β ≈ − .15). Neuroticism showed the complementary pattern for group preference, moving from a near-zero bivariate association (r ≈ − .02) to a positive unique coefficient (β ≈ .17). Given the modest sample and self-report design, these results are interpreted as pattern-validating rather than parameter-final. The study’s contribution is therefore methodological as much as substantive: it shows why EFL personality research should report zero-order correlations, multivariate coefficients and sensitivity checks together rather than relying on bivariate associations alone.