Abstract
As the apartheid system crumbles, test populations become progressively more multi-cultural. Most tests, however, were created and standardized in the days when test populations were made up of a single cultural group. Not surprisingly, questions are increasingly being asked as to the comparability of test scores across cultures. As yet, very little comparability research has been done in the non-cognitive domain. In this article the authors examine the comparability of scores on the South African Personality Questionnaire (SAPQ) in two black and and two white samples. Construct comparability of the five scales was investigated using a number of techniques, including exploratory and confirmatory factor analysis; item bias was investigated through the application of a ‘multimethod-multisample’ strategy. An item analysis was also conducted. The results indicated modest support for the construct comparability of scales in both white and one black sample; but a majority of items failed to satisfy either the no-bias or the item-total correlation criteria. The Rigidity scale performed the worst and the Dominance scale the best. The findings indicate that the SAPQ cannot simply be patched up in order to make it suitable for cross-cultural applications. The authors believe that a clean-sheet approach is needed. They suggest a strategy for the creation of a new personality instrument suitable for cross-cultural use in South Africa.
Get full access to this article
View all access options for this article.
