Comparing machine learning methods for predicting dark triad personality traits using social media text data

• Dictionary features from Facebook updates enabled Dark Triad trait prediction. • Random Forest achieved the lowest RMSE for all Dark Triad traits. • Prediction bias was small and similar across all evaluated models. • Bootstrap comparisons showed Random Forest beat most alternatives. The Dark Triad (DT) personality traits, characterized by manipulativeness, callousness, and egocentrism, are linked to both negative outcomes such as aggression and delinquency, as well as positive outcomes like career success. This study aims to compare different machine learning models for predicting DT traits − Narcissism, Machiavellianism, and Psychopathy − using social media text data from Facebook status updates and personality questionnaires. Various machine learning models were evaluated. Across traits, Random Forest achieved the lowest RMSE, outperforming most other models, followed by Support Vector Machines and Gaussian Processes. Bias was similar across all models. These findings highlight the potential of social media data to offer insights into users’ personalities and carry methodological implications for future research on personality assessments.