Universiteitstoelating in die nuwe Suid-Afrika: ideologiese en statistiese oorwegings

This article discusses the misconceptions which can arise when statistically distinguishable groups with diverse school backgrounds compete for admission to tertiary academic institutions. It is recommended that separate correlations (between predictor and criterion) and separate regression equations (for predicting the criterion in terms of the predictor) be determined for such groups. If this is done, it becomes clear that the lower mean matriculation and/or aptitude test scores of one group are not necessarily prejudicial to its members and that simply no statements can be made about the biasedness or unbiasedness of the predictor on the basis of group differences in the above correlation. Ideological objections to the computation of separate statistical indices and equations for different demographic groups are discussed and regarded as inconsistent.