Modeling dynamic processes with panel data: An application of continuous time models to prevention research

Abstract: 

This presentation will explore the article by Deboeck, Cole, Preacher, Forehand, & Compas (2021). The article argues that effective interventions may not be limited to changing means but also alter how variables interact over time. Consequently, considering dynamic models of change may be crucial for intervention research. This presentation will start by introducing derivatives, which provide a framework for expressing and understanding change processes and different models of change. Using derivatives as a foundation, several models will be presented, beginning with the latent growth curve model and the latent change/difference score model, and building to a common continuous time model. The application of continuous time models to prevention research will then be discussed, particularly when there are relatively few repeated observations over time (panel data). Finally, a substantive example will be used to compare models that suggest different continuous time processes, fitting the models to panel data and examining the dynamics between control and intervention groups.