Psychometric evaluation of the MoCA-TR in adolescence: Evidence from item-level bifactor analysis and GAMLSS-based...

Objective: Age-sensitive normative data for the Montreal Cognitive Assessment (MoCA) in adolescents are lacking. This study established normative reference values and evaluated the psychometric properties of the Turkish MoCA (MoCA-TR) within Kane's Interpretive Use Argument framework. Method: Participants were 400 community-based adolescents aged 11-18 years (M = 14.5, SD = 2.3; 50% female) recruited from a Family Health Center in Western Turkey. Analyses comprised item-level confirmatory factor analysis (26 binary indicators, WLSMV), bifactor variance decomposition, MIMIC measurement invariance, and GAMLSS normative modeling. Convergent validity used the Stroop Color-Word Test and the BRIEF-2. An exploratory known-groups comparison included 22 adolescents with self-reported neurodevelopmental diagnoses. Results: Six-model comparison supported a three-factor structure (Attention/Executive, Memory, and Language/Abstraction) and a bifactor model with an intermediate unidimensional-multidimensional profile (ECV = 0.512; ωh = 0.584; H = 0.765). Composite reliability was satisfactory (ωtotal = 0.876) and test-retest stability was strong (ICC = 0.83, 95% CI [0.75, 0.89]). Convergent validity was supported by correlations with Stroop interference (r = -0.38) and BRIEF-2 Global Executive Composite (r = -0.36). MIMIC analyses flagged educationally sensitive items (serial subtraction, verbal fluency, and abstraction); Delayed Recall items were sociodemographically invariant. GAMLSS-derived centile curves revealed monotonically increasing scores across adolescence (10th percentile: 24-26). Conclusions: The MoCA-TR demonstrates adequate psychometric properties for cognitive screening in Turkish adolescents. Developmentally calibrated normative values reduce misclassification risk from adult cutoffs, and MIMIC findings support the sociodemographic fairness of episodic memory items for equitable screening.