Psychological and Social Correlates of Mental Health among College Students: A Correlational and Predictive Analysis

Description

DOI: https://doi.org/10.5281/zenodo.21868706

Aviral Goswami and Roshan Lal Zinta (Department of Psychology, Himachal Pradesh University, Shimla, Himachal Pradesh)

Depression, anxiety, stress, self-efficacy, and perceived social support are all implicated in student mental health, but their joint contribution is usually estimated in a single pooled model that assumes an invariant association structure. Methods. Four hundred college students of District Shimla, Himachal Pradesh, completed the Depression Anxiety Stress Scales, the General Self-Efficacy Scale, a three-source perceived social support measure, a drug abuse screening test, and a six-dimension mental health inventory. Pearson correlation matrices and multiple linear regressions predicting mental health were computed separately within each of eight demographically defined subgroups, and multicollinearity was assessed by variance inflation factor and condition index. Association structures were markedly heterogeneous: the depression with anxiety correlation was positive in all three rural subgroups and negative in all three urban subgroups; the distress with support correlation ranged from -.38 to .72 across subgroups; and self-efficacy was independent of all other variables in five of six subgroups but formed a near-collinear cluster in one. Four of eight regression models could not be estimated, and only one satisfied the multicollinearity assumption. In that model, no substantive predictor reached significance once the criterion’s own constituent dimensions were set aside. The predictive question posed by the study cannot be answered by the models as specified. A hierarchical specification is set out, together with the assumption tests required, so that the analysis can be re-estimated on the existing data.