Personality predicts internalizing symptoms and quality of life in police cadets: A comparison of artificial intelligence and parametric approaches

Abstract

Background: Police cadets undergo persistent and elevated stress due to continuous training and evaluation. Identifying resilience and risk factors in this population can thus crucially inform management decisions within the police force. Here, in two large cohorts of police cadets (n=1069, 30% women and n=1377, 35% women) we investigated whether broad personality traits could predict internalizing symptoms (somatization, depression, and anxiety) as well as mental health-related quality of life (MHRQoL). Moreover, we compared seven popular artificial intelligence and linear regression models (Elastic Net, General Linear Model, Lasso Regression, Neural Networks, Random Forests, and Support Vector Regression) in predicting MHRQoL as a function of all other variables. Results: A Random Forest accounted for about half of the observed variance in MHRQoL, and outperformed all other models by up to 12% in an out-of-sample cross-validation. In all analyses, emotional stability emerged as the primary personality trait linked to MHRQoL, with anxiety and somatization symptoms partially mediating this relationship. Conclusions: Our findings delineate the personality factors that best predict internalizing symptoms and MHRQoL among cadets, and tentatively suggest that Random Forest models might be a powerful forecasting tool in police management.