Why is stress the real culprit behind depression in Asian students and workers?

A new study shines a light on depression among students and workers in India, Malaysia and China. Researchers compared two groups—students and employees—using data on age, workload, perceived pressure, financial stress and personal satisfaction.

The findings reveal pressure as the strongest predictor for both groups. For students, pressure, younger age, study load, money problems and a mental‑health history raise depression risk. For workers, pressure, older age, job load and lower satisfaction play a similar role.

A striking twist is the opposite effect of age: younger students are more vulnerable, while older workers face higher risk.

Statistical models (logistic regression, random forest and causal forest) confirmed that pressure acts directly, not as a mediator of age. Satisfaction also influences depression partly through pressure.

Although the models perform well within each country, their ability to predict across nations is weaker. In practice, tackling academic and occupational stress could be the key to preventing depression in these settings.

The study employed three predictive modeling approaches. Logistic regression offered an interpretable baseline but struggled with non‑linear interactions. Random forest, built on multiple decision trees, improved internal accuracy (AUC ~0.78) by handling correlated variables and uncovering complex patterns. Causal forest was used to estimate heterogeneous effects of pressure across sub‑groups, revealing limited variance: pressure acts almost uniformly on students and workers, with an average effect of about 0.35 log‑odds per unit increase.

Causal mediation analysis (CMA) tested whether intermediate variables, such as perceived pressure, mediate the impact of other covariates (age, satisfaction). Results show pressure does not mediate age’s effect but acts directly. Satisfaction exerts an indirect effect through pressure, with a mediation proportion of 22%.

Regarding generalizability, models lost performance when trained on one country and tested on another (ΔAUC ≈ -0.12), indicating cultural, socioeconomic, and healthcare system factors shape data structure. Enhancing transferability would benefit from adding contextual variables (e.g., welfare policies, educational norms) and employing transfer learning techniques.

In summary, pressure stands out as the main driver of depression, suggesting that interventions aimed at reducing perceived stress—such as time‑management programs, counseling, and balanced workload policies—could significantly improve mental health among Asian students and workers.