cross-lagged panel design
Definition
Cross-lagged panel design is a longitudinal method in which two or more constructs are measured at two or more discrete occasions, typically separated by months or years, and the variables at each occasion are regressed on those at the previous occasion to estimate predictive effects over a specific time lag. This procedure yields cross-lagged parameters representing how much a change in one variable predicts a change in another across that interval, while auto-regressive effects control for each variable's prior level. Although such parameters do not support causal inference, a non-zero regression coefficient is a necessary, if not sufficient, condition for establishing causality, so cross-lagged paths are interpreted in predictive rather than causal terms. The design is most appropriate when data span a small number of measurement occasions from a large sample and the research questions concern predictive or potentially causal relations among constructs, as illustrated by longitudinal panel data on students' commitment to school and self-esteem.
Sources: Wysocki et al. (2025)
Related Terms
- network models (1 shared article)
- cross-lagged networks (1 shared article)
Applications
Cross-lagged Panel Design and Network Modeling
The cross-lagged panel network extends the cross-lagged panel design by modeling predictive longitudinal pathways at the level of individual items rather than latent constructs, combining regularized regression estimation with structural equation modeling to recover auto-regressive and cross-lagged paths among observed variables. Applied to data on commitment to school and self-esteem, this approach can identify specific bridge nodes, such as the link between the item 'I don't do well at school' and the self-esteem item 'I'm a failure', that aggregate latent variable models would obscure.
Sources: Wysocki et al. (2025)
Cross-lagged Panel Design and Latent Variable Models
In the traditional application of the cross-lagged panel design, constructs are represented as latent variables whose cross-lagged paths capture predictive effects between unobserved constructs over time, such as one latent construct affecting another across a developmental period. The cross-lagged panel network was proposed precisely because latent variable cross-lagged panel models aggregate item-level information and can miss specific inter-item pathways that generate more detailed and testable hypotheses about how constructs influence one another.
Sources: Wysocki et al. (2025)
