Recent methodological work has shown that the classical Cross-Lagged Panel Model (CLPM) conflates stable between-person differences with within-person temporal processes, which can produce bias through inflated autoregressive and cross-lagged estimates (Hamaker, 2023; Hamaker et al., 2015; Lucas, 2023; Usami et al., 2019). As an alternative, Curran et al. (2014) presented the Growth Curve Model with Structured Residuals (GCM-SR) which simultaneously estimates latent trajectories and cross-lagged parameters between the residuals of the constructs, allowing for a separation of between- and within-person effects. The present study applies the GCM-SR to the longitudinal link between emotional neglect and juvenile delinquency across five panel waves from the CrimoC study. On the within-person level, emotional neglect showed strong stability while juvenile delinquency did not show significant autoregressive effects. Cross-lagged parameters indicated no significant effect of emotional neglect on subsequent delinquent behavior, and no significant effect of juvenile delinquency on later emotional neglect. On the between-person level, intercept and slope factors of both constructs were negatively associated, indicating that higher initial levels of both emotional neglect and juvenile delinquency predict weaker subsequent increases. The construct’s intercepts were positively associated, but their slopes were not. A negative association emerged between the intercept of emotional neglect and the slope of juvenile delinquency. Compared to an identically constrained CLPM, the GCM-SR yields smaller and non-significant cross-lagged estimates, likely reflecting that the effects are due to stable between-person variances. This supports the use of the GCM-SR instead of the CLPM when aiming for conclusions on within-person processes.
