IntroductionCross-cultural psychometric validation of personality and self-regulated learning (SRL) instruments remains limited in Central Asian institutional contexts. This study examined the structural validity, reliability, measurement invariance and criterion validity of Mongolian adaptations of the Big Five Inventory-2 Short Form (BFI-2-S) and the Motivated Strategies for Learning Questionnaire strategy subscales (MSLQ-M).MethodsIn 307 students at a Mongolian security and defense training institution, Exploratory Structural Equation Modeling (ESEM) with GEOMIN rotation was conducted in Mplus 8.3, alongside an independent-clusters confirmatory model, gender invariance testing, and a full ESEM-SEM linking personality, SRL and self-reported GPA.ResultsThe BFI-2-S fitted well (CFI = .933, RMSEA = .052) but yielded a dominant general evaluative factor spanning Agreeableness, Conscientiousness and Open-Mindedness, with Negative Emotionality the only cleanly recovered domain, a consequence of the all-positive item adaptation adopted for cultural appropriateness. An independent-clusters confirmatory model fitted substantially worse (CFI = .851) and returned latent correlations of .83 to .97 among the four positively-valenced domains, indicating that the conventional model relocates rather than detects this artefact. The MSLQ-M supported a two-factor higher-order structure, Cognitive-Metacognitive and Behavioral-Social SRL (r = .764), although its nominal nine-subscale structure was not recovered. Reliability was strong across all 14 scales (α = .760–.947) and full scalar invariance across gender was supported. Four of five personality domains correlated significantly with all nine MSLQ-M subscales (r = .315–.497), whereas Negative Emotionality showed near-zero associations. In a full ESEM-SEM the general evaluative factor predicted both SRL factors (β = .366 and .440), whereas no cleanly measured domain did, and neither SRL nor personality predicted self-reported GPA, a null robust across three criterion specifications.DiscussionFindings highlight the psychometric costs of all-positive item adaptations and indicate that apparent personality–SRL associations under such adaptations may be carried substantially by shared response valence.