An examination of the psychometric properties and dimensionality of the Aggression-Problem Behavior Frequency Scale in a sample of Black South African adolescents

Abstract

Advancing research into aggressive behaviour in South Africa necessitates a range of suitable measuring instruments. The Aggression-Problem Behavior Frequency Scale is one such instrument that has been extensively used in diverse settings. The authors of the Aggression-Problem Behavior Frequency Scale present it as a multidimensional scale that purports to measure physical, non-physical, and relational aggression. Despite the widespread use of the scale as a three-factor scale, the psychometric properties, specifically the factor structure of the instrument, have not been examined either globally or in South Africa. This article reports on the reliability, factor structure, and dimensionality of the Aggression-Problem Behavior Frequency Scale among a sample of Black South African adolescents. The scale demonstrated a high level of reliability both in terms of coefficient alpha and coefficient omega. Confirmatory factor analysis indicated that a second-order factor model and a bifactor model demonstrated a better fit than a one-factor model. Despite the superiority of fit of the two models, bifactor indices demonstrated that the Aggression-Problem Behavior Frequency Scale is essentially unidimensional. First, Omega Hierarchical Subscale indicated a significant reduction in the reliability of the subscales after partitioning out the variance attributable to the general factor (total aggression). Second, the percentage of common variance explained by the total scale was 75% with only 25% attributable to the three subscales. While the findings of the study tentatively confirmed the suitability of the Aggression-Problem Behavior Frequency Scale for use in South Africa, the results indicate that there is no empirical basis for the use of subscale scores and that the Aggression-Problem Behavior Frequency Scale is essentially unidimensional.