Comparing different versions of the Rahim EI questionnaire in a South African context: A confirmatory factor analysis approach

Abstract

Orientation: Given the interest in the importance of emotional intelligence in employees and leaders with regard to performance of their jobs, it is imperative to use reliable and validinstruments to operationalise emotional intelligence.

Research purpose: The purpose of the study was to assess the psychometric properties of three different versions of the Rahim emotional intelligence index (EQI), specifically withregard to its factor structure and reliability, using two different samples.

otivation for the study: No previous study has investigated which version of the Rahim EQI is the most appropriate for conducting research within South African organisations. Inaddition, the question of whether the Rahim EQI measures a strong general factor has notbeen answered.

Research approach, design, and method: A cross-sectional quantitative research design wasused. Two samples were used (n = 470 and n = 308). The first sample completed the 40-itemversion of the Rahim EQI, whilst the second sample completed the 30-item version of the Rahim EQI. The measurement model, representing the 22-item version of the Rahim EQI, was also fitted to both these samples. Confirmatory factor analysis was used to compare thedifferent versions, as well as conceptualisations, of the Rahim EQI.

Main findings: The 22-item version of the Rahim EQI exhibited better model fit than the 40-item and 30-item versions. In addition, the bifactor model suggested that the Rahim EQIseems to measure a strong general factor (emotional intelligence) with very little evidence ofthe presence of unique group factors (self-awareness, self-regulation, motivation, empathy and social skills).

Practical/managerial implications: Both the reliability and factor structure of the 22-item version of the Rahim EQI have been confirmed. The bifactor structure should inform researchers and practitioners that, in order to understand emotional intelligence, it is better to conceptualise it as a unidimensional construct.

Contribution/value-add: In order to identify the most appropriate conceptualisation associated with the Rahim EQI, various goodness-of-fit statistics (e.g. comparative fit indexand root mean square error of approximation) should be consulted. The impact of the removalof items from instruments should be investigated with regard to the accuracy with which the construct is to be measured. The current study has also contributed to the literature byexamining the psychometric properties of the Rahim EQI in a South African sample.