Seeing reason: visuospatial ability, sex differences and the Raven’s Progressive Matrices

Abstract

This thesis sought to address the role of visuospatial ability in measures of inductive reasoning, with a particular focus on the Raven’s Progressive Matrices (RPM). Given that males tend to perform better on certain measures of visuospatial ability, sex differences in performance on the RPM tests and in other measures of inductive reasoning were also examined. The issue of the involvement of visuospatial ability in the RPM tests is important at both a practical and a theoretical level. At the practical level, these tests are often used as a sole measure of general intelligence, and conclusions regarding the relationship of general intelligence to other variables are made on the basis of results from this test. If the RPM tests require a substantive amount of visuospatial ability, this is problematic to the interpretation of results on this test as reflective of general intelligence. At a theoretical level, investigation of this question pertains to an understanding of the relationship between visuospatial abilities and fluid ability generally, but inductive reasoning more specifically. Many commonly used measures of inductive reasoning are presented in a visual format (e.g. abstract figures) and these tests are often shown to cross-load on both fluid and visuospatial factors. This thesis addresses the issues of visuospatial ability and sex differences in the RPM by examining (1) the dimensionality of the Advanced RPM tests; (2) the role of Gv in performance on the RPM tests; and (3) sex differences in raw scores on the RPM and other measures of inductive reasoning. Additionally, the psychometric properties of the General Reasoning Test 2 (GRT2) in the Australian population were examined. This included an investigation of the relationship between figural, verbal and numeric reasoning items as well as sex differences. Study 1 used confirmatory factor analysis and Rasch modeling to investigate the dimensionality of the Advanced RPM, measurement invariance and differential item functioning across sex. Study 2 used structural equation modeling to examine, in three separate samples, how well visuospatial abilities could account for the variance in a latent RPM factor not already accounted for by alternative fluid ability measures. This study additionally assessed invariance of the structural relationships between visuospatial ability, fluid ability and RPM across sex. Study 3 used meta-analytic techniques to synthesise research concerning sex differences on measures of inductive reasoning, considering the item stimulus and item type as potential moderators of this difference. Study 4 used exploratory and confirmatory structural equation modeling to examine the psychometric properties of the GRT2. Results indicate that although the RPM tests are largely unidimensional, visuospatial ability is involved in performance. Furthermore, sex differences in raw scores and at the latent level were found, favouring males. Investigation of sex differences in inductive reasoning measures more broadly indicated that the figural format of these tests may contribute to the male advantage often identified; however, examination of the influence of the stimulus and type of question used in reasoning items in the GRT2 indicated that these do not meaningfully impact the latent construct measured.