Connected or Isolated? Examining the Associations between Smartphone Use, Mental Health, and Phubbing: Moderated Mediation via Bayesian Networks with Machine Learning

Abstract This study investigated whether smartphone addiction mediates the relationships between symptoms of stress, anxiety, and depression and phubbing behavior in Brazilian university students. A total of 348 students (67% women; M age = 22.7) participated in the study, responding to the Smartphone Addiction Scale–Short Version (SAS-SV), the Phubbing Scale, and the Depression, Anxiety, and Stress Scale (DASS-21). The analyses included Pearson correlations, multiple regression, mediation models with bootstrapping, and Bayesian networks. The results indicated that smartphone addiction was strongly associated with phubbing and moderately related to mental health symptoms. In the multiple regression model, stress (β = .42, p = .001), anxiety (β = .28, p = .004), and depression (β = .19, p = .032) together explained 10.5% of the variance in phubbing (R² = .105). The mediation models suggested that smartphone addiction acted as a partial mediator of the relationships between stress/anxiety and phubbing and as a full mediator between depression and phubbing. Bayesian networks confirmed that the probabilistic dependencies were consistent with these findings. Additional analyses revealed variations in effects as a function of sex and race, although estimates in smaller subgroups should be interpreted with caution. The results reinforce smartphone addiction as a plausible mechanism connecting emotional symptoms to problematic social behaviors in digital contexts.