Raven's Matrices: What do They Really Measure? Implications for Clinical Practice and Diagnosis.

Inconsistent findings across studies on the factorial structure of Raven's Coloured Progressive Matrices (CPM) undermine the validity of this widely used test in psychological, psychopedagogical, and neuropsychological assessments. This study examines these discrepancies by testing competing factorial models. We tested nine CPM factorial models in a sample of 1,088 students (50% male; mean age 9.4 years ± 2.6 SD) from Belo Horizonte, Brazil. Confirmatory factor analysis compared models and evaluated scalar invariance across sex and school year. Structural equation modeling tested the model's predictive capacity for word reading accuracy. As for the Results, a bifactor model, with one general factor (eduction) and two specific factors (intermediate and advanced levels), best represented the CPM's structure. This model demonstrated scalar invariance for sex and school year and predicted 18.9% of variance in word reading accuracy. In conclusion, our findings robustly support the bifactor structure's factorial validity for the CPM. Its factors meaningfully assess individual differences and predict outcomes, reinforcing the test's utility in clinical and educational settings.