TOWARDS A RATIONALE FOR TESTING ATTAINMENTS AND ABILITIES IN AFRICA

S ummary . A brief review of recent research in Africa in the field of cognitive measurement is followed by a discussion of the central issue of test score meaning in testing educational skills and aptitudes. Influences on test score meaning are hypothesised, among them sex differences and role relationships. A major influence is the quality of school attended. Two factor analyses are reported in an attempt to clarify the application of traditional constructs in the mapping and prediction of educational skills. They indicate that constructs like ‘ g,’ ‘ v. ed.’ and ‘ n ’ may have legitimate psychological meaning in non‐western cultures. To assess the comparative validity of the same test used on different cultural groups, item difficulties for Raven's Progressive Matrices between British and African groups were correlated. The resulting rank correlation (.695) was taken as prima facie evidence that the test scores were only partly comparable, but encouraged using the Raven as a ‘ g ’ reference axis in factor analysis. The results quoted were gained from a representative sample of over 1,600 African eight‐year educated children in Southern Rhodesia, and from a follow‐up group of 291 from the same sample in Form I (nine‐year educated). In addition, concurrent validities with primary school headmasters' estimates, using combinations of traditional and experimental tests, were in the region of .90 while multiple correlations of .70, uncorrected for range restriction, were gained with a Form I multiple criterion. These correlations were achieved by selecting varieties for regression analysis with reference to the factor analyses carried out, in an attempt to check the psychological meaningfulness of the factor solutions.