Background: Trait emotional intelligence (EI) is often assessed using the 30-item Trait Emotional Intelligence Questionnaire-Short Form (TEIQue-SF). However, previous research using item response theory (IRT) modelling has identified several underperforming items. This study aimed to psychometrically evaluate, refine, and optimize the TEIQue-SF using IRT, with the goals of identifying and eliminating underperforming items, and examining whether items in the refined version function differently across sexes. Furthermore, the study sought to further validate the Swedish version of the TEIQue-SF.

Methods: This cross-sectional study surveyed 845 first-year students aged 19–59 (87% women) from seven healthcare and social work programs across six universities in southern Sweden. Participants completed the TEIQue-SF and health-related measures for convergent validity. IRT modeling employed the Graded Response Model (GRM) using the 2-Parameter Logistic Model in IRT for Patient-Reported Outcomes (IRTPRO). Marginal reliability and differential item functioning (DIF) were assessed with IRT, internal consistency with Cronbach’s alpha and mean inter-item correlations, and validity through evaluating Direct Discrepancy Dynamic fit index (DDDFI) and bivariate correlations.

Results: The IRT modeling identified underperforming items, leading to a refined 12-item TEIQue-SF that effectively captures trait EI with high-quality items. The item selection process is detailed and supplemented. The shortened measure showed a strong correlation with the original (r = .94), demonstrated good reliability, and exhibited uniform DIF for only one item (Item 15). A comparison of confirmatory factor analysis (CFA) model fit statistics using the DDDFI indicated a fair fit for the TEIQue-SF. Consistent with previous research on the TEIQue-SF, both 30-item and 12 item versions demonstrated strong convergent validity with health-related measures within the Swedish context.

Conclusions: The 12-item TEIQue-SF is a brief, precise, and valid measure for assessing trait EI while preserving its global conceptual structure. IRT modeling and validity testing against health-related measures confirm that 12-item TEIQue-SF effectively captures trait EI.