The Factorial Structure of the Raven Coloured Progressive Matrices Test: a Reanalysis

Abstract

Based on data from 783 primary-grade children, results of princi pal component factor analyses of the Raven Coloured Progressive Matrices are reported. Analyses were done using phi coefficients, phi/phi max ratios, and tetrachoric correlation coefficients. The re sults indicated that phi/phi max coefficients and tetrachoric correla tions yielded identical factor patterns that were not affected by item difficulty. Factors derived from phi coefficients tend to be, at least partially, an artifact of item difficulty. Regardless of the coefficient used, three factors were identified. For phi/phi max and tetrachoric correlation coefficients the three dimensions were identified as: Clo sure and Abstract Reasoning by Analogy (Factor I), Pattern Com pletion through Identity and Closure (Factor II), and Simple Pattern Completion (Factor III).

Do Raven's Colored Progressive Matrices function in the same way in typical and clinical populations? Insights from the intellectual disability field

“…First, it does not require the calibration of a unidimensional item response model. This is particularly important in this context because, like the Raven's Standard Progressive Matrices (van der Ven & Ellis 2000), the RCPM is not strictly unidimensional (e.g., Carlson & Jensen 1980;Green & Kluever 1991). Second, both uniform DIF (usually characterized by a difference in item difficulty levels across the groups) and nonuniform DIF (caused by differences in item discriminative power) can be modeled and tested at the same time, whereas other classical methods (such as Mantel-Haenszel method) are designed to test only one of those effects.…”

The Flynn effect and its relevance to neuropsychology

“…The Colored Matrices test is even less likely to be comparable to the Advanced Matrices test that was examined by Carpenter et al (1990). The degree to which mathematical reasoning is required by the different forms of Raven's test is difficult to specify, but it does not appear to be a salient demand of either the Colored or Standard test (Carlson & Jensen, 1980;Costa, 1976;Villardita, 1985;Waltz et al, 1999). Also, Lynn's account fails to explain why the magnitude of the Flynn effect is so large for the Wechsler Similarities subtest.…”

Structural Properties of Raven's Coloured Progressive Matrices for a Sample of Gifted Children

“…The split-half and Cronbach alpha coefficients have been estimated at values ranging from .56 to .99 (1,4,7). Validation information has been obtained by correlating scores on the matrices with measures of Piagetian conservation concepts and by factor analyses.…”