Reciprocal Associations between Stress, Loneliness, and Social Networks in Middle-Aged and Older Koreans: A Random Intercept Cross-Lagged Panel Model

Background: Stress, loneliness, and social networks play pivotal roles in the mental health of middle-aged and older adults. However, their longitudinal interplay, distinguishing stable between-person differences from within-person fluctuations, remains underexplored. This study examines the temporal directionality of these variables in a Korean sample.Methods: Data were collected over three waves at six-month intervals from 297 middle-aged and older Korean adults (Mage = 58.53, range: 40–83). A Random Intercept Cross-Lagged Panel Model (RI-CLPM) was employed to analyze stress (BEPSI-K), loneliness (ULS-3), and social networks (LSNS), with demographic covariates entered as predictors of the random intercepts. Parameters were estimated through maximum likelihood with robust standard errors (MLR).Results: At the between-person level, stress was positively associated with loneliness, whereas social networks were negatively associated with both. At the within-person level, stress and loneliness reciprocally predicted each other over time: higher-than-average stress predicted subsequent increases in loneliness, and higher-than-average loneliness predicted subsequent increases in stress. Social networks demonstrated no significant autoregressive or cross-lagged effects.Conclusions: These findings suggest that the structural aspects of social networks (i.e., contact frequency, network size, and availability of support) and the subjective experience of loneliness may follow distinct pathways at the within-person level. Considering the bidirectional within-person interplay of stress and loneliness, stress management may be considered a potential intervention point for preventing the exacerbation of loneliness among middle-aged and older adults. However, causal interpretation is constrained by the observational design and unmeasured time-varying confounders.