Low-burden data-driven work addiction screening: Comparing demographic and psychosocial predictors using machine...

Area of Science:

• Organizational Psychology
• Occupational Health
• Data Science in HR

Background:

• Identifying employees at risk for work addiction is vital for early intervention.
• Systematic screening for work addiction presents practical challenges for organizations.
• This study compares demographic-only versus psychosocial screening models for work addiction.

Purpose of the Study:

• To compare the predictive performance of a demographic-only screening model against a psychosocial model for identifying employees at risk of work addiction.
• To evaluate the feasibility and trade-offs of different screening strategies in organizational settings.

Main Methods:

• Analyzed data from 22,136 (psychosocial model) and 23,219 (demographic model) employees across 85 countries.
• Assessed work addiction risk using the International Work Addiction Scale.
• Evaluated five machine learning algorithms (Logistic Regression, Random Forest, XGBoost, Neural Networks, Ensemble Voting) on selected feature sets using cross-validation.

Main Results:

• The psychosocial model demonstrated superior performance, with the best model achieving 72.7% accuracy in identifying at-risk individuals.
• The demographic-only model showed moderate performance but retained practical utility, detecting 57.0% of at-risk cases.
• High-sensitivity configurations in the demographic model achieved 71% detection but with increased false positives.

Conclusions:

• Psychosocial variables offer superior prediction for work addiction risk.
• Demographic-only models serve as a practical screening alternative, identifying a substantial portion of at-risk employees without additional data burden.
• The choice between models involves a trade-off between predictive accuracy and implementation feasibility for organizational screening programs.