The hypothesis is tested that the Raven's Advanced Progressive Matrices (APM) has the same construct validity in African university students as it does in non-African university students. Analyses were made of scores from 294 highly select 17-23-year-olds in the Faculties of Engineering and the Built Environment at the University of the Witwatersrand (187 Africans, 40 East Indians, 67 Whites; 70 women, 224 men). Out of a total of 36 problems, the African students solved an average of 22, the East Indian students, 25, and the White students, 29 (P<.001), placing them at the 57th, 64th, and 86th percentiles, respectively, and yielding IQ equivalents of 103, 106, and 117 on the 1993 US norms. Four months earlier, they had completed the Standard Progressive Matrices. The two tests correlated .60 or higher for both the Africans and the non-Africans, and both tests predicted final end-of-year grades with mean r's = .30 (P's<.05). Items found difficult by one group were difficult for the others; items found easy by one group were easy for the others (mean r's = .90, P<.001). The African-East Indian-White differences were "Jensen Effects," being most pronounced on the general factor of intelligence (measured in this instance by items with the highest item-total correlations). Indeed, the g loadings showed cross-cultural generality: For example, item-total correlations calculated on the East Indian students predicted the magnitude of the African-White differences. When each of the 36 g loadings and race effects were aggregated into nine four-item "subtests," the magnitude of the Jensen Effect was Spearman's ρ = .52. There were no sex differences. © 2003 Elsevier Science Inc. All rights reserved.
