Investigating the construct validity of a development assessment centre

Abstract

ORIENTATION : The assessment centre (AC) is a prominent measurement tool for selection and development. RESEARCH PURPOSE : The aim of this study was to determine the construct validity of a one-day development assessment centre (DAC) using a convenience sample of 202 managers in a large South African banking institution. MOTIVATION FOR THE STUDY : Although the AC method is popular, it has been widely criticised as to whether it predominantly measures the dimensions it is designed to measure. RESEARCH DESIGN, APPROACH AND METHOD : The fit of the measurement models implied by the dimensions measured was analysed in a quantitative study using an ex post facto correlation design and structural equation modelling. MAIN FINDINGS: Bi-factor confirmatory factor analysis was used to assess the relative contribution of higher-order exercise and dimension effects. Empirical under-identification stemming from the small number of exercises designed to reflect designated latent dimensions restricted the number of DAC dimensions that could be evaluated. Ultimately, only one global dimension had enough measurement points and was analysed. The results suggested that dimension effects explained the majority of variance in the post-exercise dimension ratings. PRACTICAL/MANAGERIAL IMPLICATIONS : Candidates’ proficiency on each dimension was used as the basis for development reports. The validity of inferences holds important implications for candidates’ career development and growth. CONTRIBUTION/VALUE ADD : The authors found only one study on construct validity of AC dimensions in the South African context. The present study is the first use the bi-factor approach. This study will consequently contribute to the scarce AC literature in South Africa.