Item Response Theory Analysis of a Work Ethic Scale: Evidence From Multi-Industry Samples in West Africa

Abstract

In light of growing interest in cross-cultural workforce behaviors and the need for psychometrically sound tools to assess employee values, this study investigated the validity and fairness of a Work Ethic Scale across key industries in West Africa. This study examined the psychometric properties of an adapted Work Ethic Scale across major industries in West Africa using item response theory (IRT). A sample of 800 full-time employees from education, finance, manufacturing, and healthcare sectors in Ghana and Nigeria completed the scale. A two-parameter logistic model was applied to assess item discrimination, difficulty, differential item functioning (DIF), and measurement precision. Dimensionality assessment via exploratory factor analysis and Horn's parallel analysis confirmed essential unidimensionality. The first factor had an eigenvalue of 6.87 (34.35% variance explained), exceeding the parallel analysis criterion of 1.42. Subsequent factors fell below their respective thresholds, and model fit was confirmed [KMO = 0.91; Bartlett's χ²(190) = 2892.31, p < 0.001; TLI = 0.91; RMSR = 0.04]. IRT analysis showed strong item-level performance. Discrimination values (a-parameters) ranged from 0.98 to 2.25, with 17 of 20 items exceeding 1.20. The highest discrimination was observed for “I believe in earning rewards through effort” (a = 2.25, SE = 0.15), and the lowest for “I see leisure as less important than work” (a = 0.98, SE = 0.09). Difficulty parameters (b-values) ranged from −0.70 to 0.80, indicating that most items effectively targeted mid-levels of the work ethic trait. Model fit for all items was adequate (S-X² p values > 0.05). DIF analysis revealed five items with significant DIF across industries. Uniform DIF was found in “Work is central to my life” [χ²(3) = 19.88, p = 0.0002, η² = 0.08], while nonuniform DIF was observed for “I feel guilty when I'm not being productive” [χ²(3) = 21.10, p = 0.0001, η² = 0.07]. Test information function analysis showed peak information at θ = 0.0 (I = 1.93, SE = 0.79, R = 0.92), with decreasing precision at θ = ±2.0. A key recommendation is to revise or add items to enhance measurement precision at the lower and higher extremes of work ethic, ensuring a more balanced assessment across all trait levels.