Test Bias: The Validity of the Junior Aptitude Tests (JAT) for Various Population Groups in South Africa regarding Constructs Measured

Abstract

The Junior Aptitude Tests (JAT) were administered to black, white, coloured and Indian pupils in Std 7 to determine whether the test battery measures the same psychological constructs in the various population groups. Factor analyses of the intercorrelations of the JAT test scores showed a three-factor solution for each of the groups. In the case of the white pupils (the reference group) these three factors were tentatively identified as a spatial, a memory and a verbal factor. These constructs were subsequently specified in a LISREL-type model and fitted to the data for all four population groups by means of the computer program RAMONA. Although a poor fit was obtained the same measure of fit was obtained for all four population groups. This finding points firstly to the absence of bias in the construct validity of the JAT and secondly to a greater resemblance between the cognitive structures of the different cultural groups in the RSA than is generally believed. The use of language in tests appears to be a potential source of bias in the case of the black pupils.