The longitudinal interplay of depressive symptoms and loneliness: causal effects work in both directions and decay to zero before six years.

Prior research on the longitudinal associations between depressive symptoms and loneliness has conflated between-person and within-person effects, resulting in confusion over causal influences. The present study used the random-intercept cross-lagged panel model (RI-CLPM) to disentangle between-person from within-person sources of variance. The study used a representative sample of adults older than 40 years from the German Ageing Survey (DEAS). To examine the influence of time lag on the strength of the predictive paths, analyses were conducted with lags of 3 and 6 years. The results showed that, at the within-person level, auto-regressive and cross-lagged paths were significant with lag length of 3 years but not 6 years. It can be concluded that there is a reciprocal relationship between the variables that operates at intervals shorter than 6 years. However, the stable trait-like components of the variables account for much of the covariance between them. Recommendations include paying more attention to the factors that contribute to the stability of the 2 variables across time and situation in addition to short-term and symptom-focused interventions.