An application of diagnostic modeling to a situational judgment test assessing emotional intelligence

Abstract

This study directly addresses important psychometric issues concerning emotional intelligence situational judgment tests (EI SJTs), including nonsensical dimensionality results, ambiguous facet constructs, and low Cronbach’s alpha, and then introduces an alternative methodology, cognitive diagnostic modeling (CDM), which seems to provide a better framework for the item level multidimensionality of these measures. This dissertation is the first study investigating the multidimensional nature of SJTs assessing EI using the CDM approach. Two ultimate purposes of this study include better understanding of the EI construct and advancing the psychometric analysis of EI measures assessed with the SJT format. The results of this study found that there are five dimensions underlying an SJT measuring emotion understanding (STEU) and they tend to have noncompensatory relationships. An SJTs measuring emotion management (STEM) showed four strategies required to perform well on the test, which interact in a compensatory manner. As hypothesized, the G-DINA (generalized deterministic inputs, noisy “and” gate) model best reproduced the SJT data among the other reduced models due to its statistical generality. However, a higher order structure of EI was not found in the CDM analysis. Among other commonly used methodologies, the CDM approach fully reflected the theoretical framework of EI and provided finer-grained information on the test and examines, which can be reflected in better feedback to assessees.