Validating Indigenous Versions of the South African Personality Inventory

Personality assessments are frequently used to make decisions and predictions, creating a demand for assessments that are non-discriminatory. South African legislation requires psychological tests to be scientifically proven to be valid, reliable, fair and non-biased. In response to the necessity for a measure sensitive to indigenous differences, South African and Dutch researchers developed the South African Personality Inventory (SAPI). The SAPI represents a theoretical model of personality that uses an indigenous (emic) and universal (etic) approach to capture South Africa’s rich multicultural and multilingual view of personhood. The development of SAPI items and its simultaneous translation from English into all official languages necessitated the investigation of all the translated language versions’ psychometric properties. This study used Exploratory Structural Equation Modelling to examine the factor structure and model fit of two indigenous language versions of the SAPI, targeting the Tshivenda and the Southern Sotho languages. To accomplish this objective, Study 1 (N = 290) was done in South Africa among the Tshivenda ethnic group, while Study 2 (N = 293) was conducted in South Africa among the Sesotho ethnic group. An acquiescence response pattern was noticed in both studies, possibly to adhere to group consensus and emphasizing harmony within relationships. The ESEM solutions generated an excellent fit for both language versions, and most facets loaded acceptably on their expected factors. The Neuroticism factor proved to be problematic in both language versions. Within the Tshivenda version, the Emotional Stability facet did not generate adequate loadings on any SAPI factors. In contrast, neither Emotional Stability nor Negative Emotionality loaded sufficiently on the Neuroticism factor for the Southern Sotho version. While the overall fit of the six-factor model was excellent, the language in which a person completes a personality questionnaire seems to influence such an assessment’s factor structure. The Tshivenda and Sesotho versions of the SAPI cannot yet be positioned as equitable alternatives when using an indigenous version of the SAPI is needed. The implications of the results and proposals for future studies are discussed.