A unified framework of longitudinal models to examine reciprocal relations.

Abstract

Inferring reciprocal effects or causality between variables is a central aim of behavioral and psychological research. To address reciprocal effects, a variety of longitudinal models that include cross-lagged relations have been proposed in different contexts and disciplines. However, the relations between these cross-lagged models have not been systematically discussed in the literature. This lack of insight makes it difficult for researchers to select an appropriate model when analyzing longitudinal data, and some researchers do not even think about alternative cross-lagged models. The present research provides a unified framework that clarifies the conceptual and mathematical similarities and differences between these models. The unified framework shows that existing longitudinal models can be effectively classified based on whether the model posits unique factors and/or dynamic residuals, and what types of common factors are used to model changes. The latter is essential to understand how cross-lagged parameters are interpreted. We also present an example using empirical data to demonstrate that there is great risk of drawing different conclusions depending on the cross-lagged models used.

Comorbidity between depression and anxiety: assessing the role of bridge mental states in dynamic psychological networks

“…Although multilevel lagged regression models with person mean-centering have been discussed as a method that can control for the effect of unmeasured time-invariant confounding variables (cf. [95]), there are likely numerous unobserved time-varying confounders present in the current setting (such as additional unmeasured symptoms, mental states, and contextual factors). Additionally, misspecification of the causal timescale, as previously discussed, and the functional form of the temporal relationships would also undermine veracity of causal conclusions made on the basis of estimated cross-lagged effects [72].…”

Editorial: Some roads less travelled—different routes to understanding the causes of child psychopathology

“…Models like the ALT‐SR model (and similar models like the random intercept‐cross lagged model (RI‐CLPM); (Hamaker, Kuiper, & Grasman, 2015) and the stable trait autoregressive trait and state (STARTS) model (Kenny & Zautra, 1995)) can therefore control for unobserved trait‐like confounders that are stable across time (i.e. the between‐person differences, see Usami, Murayame, & Hammaker, 2019). For this reason, these models are to be preferred over the traditional CLPM because they allow us to model the predictors of change within an individual (does person A with high externalising problems at time 1 show larger increases in internalising problems over time than person B who has lower externalising problems at time 1?…”

Meta-analysis of Lagged Regression Models: A Continuous-time Approach

“…Furthermore, while the interest of researchers applying these methods is typically in making some inference about (Granger-)causal relationships (cf. Usami, Murayama, & Hamaker, 2019), the order of magnitude of cross-lagged effects does not necessarily ensure the same ordering of 'causal dominance' relations: further assumptions are necessary for these lagged parameters to reflect causal relationships. Nonetheless, the availability of a method to meta-analyze cross-lagged parameters from studies with different measurement designs is a necessary first-step on the road to establishing any reliable conclusions regarding the nature of these relationships in principle.…”