Predicting Work–Family Conflict and Marital Satisfaction Using Machine Learning: The Role of Job Stress, Emotional Exhaustion, Dyadic Coping, and Partner Support

Abstract

The present study aimed to predict work–family conflict and marital satisfaction among married employees using machine learning algorithms based on job stress, emotional exhaustion, dyadic coping, and partner support. This cross-sectional predictive study was conducted among 584 married employees recruited from governmental and private organizations in Tehran, Iran, through multistage cluster sampling. Participants completed the Work–Family Conflict Scale, Revised Dyadic Adjustment Scale, Job Stress Scale, Emotional Exhaustion subscale of the Maslach Burnout Inventory, Dyadic Coping Inventory, and Spousal Support Scale. Following data preprocessing, normalization, and missing-value treatment, the dataset was divided into training (80%) and testing (20%) subsets. Several machine learning algorithms, including Multiple Linear Regression, Support Vector Regression, Random Forest Regression, Gradient Boosting Regression, Artificial Neural Networks, and Extreme Gradient Boosting (XGBoost), were implemented. Model performance was evaluated using the coefficient of determination (R²), Root Mean Square Error (RMSE), Mean Absolute Error (MAE), and Mean Absolute Percentage Error (MAPE). Feature importance analyses were conducted using permutation importance and SHAP techniques. Correlation analyses revealed that work–family conflict was positively associated with job stress (r = 0.68, p < .001) and emotional exhaustion (r = 0.64, p < .001), while negatively associated with dyadic coping (r = -0.49, p < .001), partner support (r = -0.56, p < .001), and marital satisfaction (r = -0.62, p < .001). Marital satisfaction demonstrated significant positive relationships with dyadic coping (r = 0.73, p < .001) and partner support (r = 0.76, p < .001). Among all predictive models, XGBoost demonstrated the highest accuracy. The optimal model explained 81% of the variance in work–family conflict (R² = 0.81, RMSE = 5.14) and 84% of the variance in marital satisfaction (R² = 0.84, RMSE = 4.37). Feature importance analyses indicated that job stress and emotional exhaustion were the strongest predictors of work–family conflict, whereas partner support and dyadic coping emerged as the most influential predictors of marital satisfaction. The findings demonstrate that machine learning approaches can accurately predict work–family conflict and marital satisfaction using a combination of occupational and relational variables. Job stress and emotional exhaustion substantially increase vulnerability to work–family conflict, whereas dyadic coping and partner support function as powerful protective factors that enhance marital satisfaction. These results underscore the importance of strengthening both workplace well-being and couple-based coping resources to promote healthier marital relationships and better adaptation to work–family demands.