A Rasch analysis of the high potential trait indicator : a South African sample

Abstract

The reliability and validity of the six traits comprising the High Potential Trait Indicator (HPTi) were evaluated using Rasch analysis. Focus was designated to the unidimensionality and local independence of each subscale; fit to the Rasch model; person reliability and separation; and differential item functioning (DIF). Secondary data, obtained from intellectual property rights holder Thomas International, were used for analysis with a sample of 1257 South African respondents. One of the six traits, Curiosity (0.73), was found to be reliable. Traits Adjustment (0.69) and Competitiveness (0.69) border on the accepted cut-off of 0.70. Risk Approach (0.64) obtained the lowest reliability, closely followed by Conscientiousness (0.65) and Ambiguity Acceptance (0.65). Six of the 78 HPTi items did not fit the Rasch model, all of which underfit the model. Trait Curiosity was found not to be unidimensional, while the Ambiguity Acceptance scale approached the value at which a scale is considered multidimensional. One item was identified to be threatening the unidimensionality of the Curiosity scale based on both the factor loadings of the principal components analysis of the residuals and underfitting the Rasch model. The differential item functioning (DIF) analysis found no item bias between genders, female and male. Eleven items displayed DIF across ethnicities and home language groups. The most severe instance of DIF occurred in trait Competitiveness, yet it had only one item experiencing DIF. Trait Conscientiousness, however, contained four items experiencing various severities of DIF.

CONTRIBUTION : This study highlighted the shortcomings of the current HPTi in the South African context through Rasch analysis. The findings illustrate the difficult nature of creating ideal personality instruments in the South African context, thus contributing to the body of knowledge of personality assessments in South Africa.