Abstract
The random intercept cross-lagged panel model (RICLPM) decomposes longitudinal associations between two processes X and Y into stable between-person associations and temporal within-person changes. In a recent study, Bailey et al. demonstrated through a simulation study that the between-person variance components in the RICLPM can occur only due to the presence of time-varying covariate processes that are omitted from the analysis model. Therefore, the between-person component was termed illusory. In this article, necessary and sufficient conditions for the occurrence of such illusory between-person components in an RICLPM are analytically derived and systematically investigated in a simulation study.
Notes
1 Semantically, the label „univariate RICLPM“ is a misnomer because the model does not include cross-lagged effects. However, we nevertheless use this label because it eases the generalization to the multivariate RICLPM and is also used in the literature (Andersen, Citation2022).