Abstract
This study investigated the psychometric properties of the Dirty Dozen, a measure of the Dark Triad traits (Machiavellianism, Narcissism, and Psychopathy) within the South African context. The measure’s construct validity was investigated by comparing it with the South African Personality Inventory (SAPI), which accounts for culturally specific personality traits in South Africa.
The validation study of the Dirty Dozen measure followed a cross-sectional quantitative research approach. Primary data were collected using a non-probability convenience sampling method from 368 South African working adults, with a subset of 70 participants selected for the test-retest reliability analysis. Data were analysed using classical test theory approaches and software to determine the reliability and validity of the factor structure and determine convergent and discriminant validity. Exploratory Structural Equation Modeling (ESEM) was conducted, confirming a three-factor model. The test-retest reliability correlation coefficients ranged from r = 0.54 and 0.72, confirming stability.
Findings showed that the Dirty Dozen exhibited acceptable reliability and validity within this context and generated clear alignment between the Dirty Dozen’s constructs and certain SAPI dimensions, particularly with the Negative Social-Relational Disposition.
The study acknowledged its limitations, such as potential social desirability bias. While the sample size was sufficient for the analyses conducted, future research could benefit from a larger and more diverse sample size to strengthen generalisability. The results show that SAPI contributes to theoretical and practical applications by validating the Dirty Dozen as a cross-cultural tool for assessing socially aversive traits in South Africa. This validation supports its useability in an organisational setting, where understanding these traits can improve workplace dynamics and enhance the employee selection processes, creating a more
CONSTRUCT VALIDATION OF THE DIRTY DOZEN
6
effective environment. Furthermore, a deeper exploration of alternative statistical models, such as bifactor analysis, would be beneficial for future research.