Revisiting the nature and strength of the personality-job performance relations: New insights from interpretable machine learning.

Prior research on the relations between the five-factor model (FFM) of personality traits and job performance has suggested mixed findings: Some studies pointed to linear relations, while other studies revealed nonlinear relations. This study addresses these gaps using machine learning (ML) methods that can model complex relations between the FFM traits and job performance in a more generalizable way, particularly interpretable ML techniques that can more effectively reveal the nature (linear, curvilinear, interactive) and strength (feature/relative importance) of the personality-job performance relations. Overall, the results based on a sample of 1,190 employees suggest that nonlinear ML methods perform slightly yet consistently better than linear regression methods in modeling the relation of job performance with FFM facets, but not with factors. On the factor level, conscientiousness exhibits a noticeable curvilinear relation with job performance, and it also interacts with other FFM factors to predict job performance. Conscientiousness displays the strongest feature importance across job types, followed by agreeableness. On the facet level, most FFM facets show limited evidence for curvilinear and interactive (with other facets) relations with job performance. While several conscientiousness facets (order, deliberation, self-discipline) display the strongest feature importance in predicting job performance, some agreeableness (straightforwardness, altruism) and extraversion (positive emotionality) facets also emerge as important features for different sales job types (corporate vs. individual sales). We discuss the implications of these findings for research and practice. (PsycInfo Database Record (c) 2024 APA, all rights reserved).