Disentangling the Mediation Pathways of Depression in Asian Students and Workers

Depression is a major global mental health issue shaped by cultural, demographic, and occupational factors. This study compares predictors of depression across student and worker populations using datasets from India, Malaysia, and China. The India dataset was split into student and worker groups, while the Malaysia dataset includes only students and the China (CHARLS) dataset includes only workers. After harmonizing variables, we applied logistic regression, random forest, and causal forest models to identify key predictors and subgroup-specific effects, and conducted causal mediation analysis (CMA) to assess whether variables operate through intermediaries such as perceived pressure. Among students, pressure, age, workload, financial stress, mental health history, and satisfaction were significant predictors; similar factors emerged for workers. Notably, age showed opposite effects across groups: younger students were more likely to experience depression, whereas older workers showed higher risk. Model performance showed moderate internal accuracy but weaker external generalizability across countries, with random forest outperforming logistic regression. Causal forest results indicated limited heterogeneity in the effect of pressure, while CMA showed that pressure does not mediate the effect of age but operates more directly, and satisfaction influences depression partly through pressure. Overall, pressure consistently emerged as the strongest predictor, suggesting that interventions targeting academic and occupational stress may help reduce depressive symptoms.