What is it about?
People can change in complex ways over time. Their symptoms, behaviors, or relationships may develop gradually, but they can also shift suddenly from one state to another. Intensive longitudinal data, collected repeatedly from the same individuals, can capture these changes, but analyzing them requires models that account for both individual differences and changing dynamics. In this tutorial, we introduce Dynamic Latent Class Structural Equation Modeling (DLCSEM), a flexible framework that combines dynamic structural equation models with hidden-state models. We build the approach step by step, from factor analysis and time-series models to the full DLCSEM, and show how it can be implemented and interpreted using an example from psychotherapy research on anxiety symptoms and the therapeutic alliance.
Featured Image
Photo by Shawn Day on Unsplash
Why is it important?
Many processes studied in psychology and the behavioral sciences are not stable over time. People may move between different psychological states, experience sudden changes, or show changes in the way underlying constructs are expressed. Standard longitudinal models can make these transitions difficult to detect. DLCSEM provides a framework for studying such changes while simultaneously modeling within-person dynamics, between-person differences, and latent psychological constructs. By providing a hands-on, step-by-step introduction and practical implementation, this tutorial lowers the barrier to applying these models and makes a powerful class of methods more accessible to applied researchers working with intensive longitudinal data.
Perspectives
Intensive longitudinal data are becoming increasingly common, but researchers still have relatively few practical resources that guide them through the implementation of models capable of capturing their complexity. This creates a gap between the availability of rich longitudinal data and our ability to analyze them appropriately. At the same time, these data make it possible to investigate dynamic phenomena that were previously difficult, or even impossible, to study with more traditional research designs. Our motivation for this tutorial was therefore not only to introduce DLCSEM, but also to provide researchers with a practical route to implementing it and to make these emerging analytical possibilities more accessible.
Roberto Faleh
Eberhard Karls Universitat Tubingen
Read the Original
This page is a summary of: Dynamic latent class structural equation modeling: A hands-on tutorial for modeling intensive longitudinal data., Psychological Methods, August 2026, American Psychological Association (APA),
DOI: 10.1037/met0000849.
You can read the full text:
Contributors
The following have contributed to this page
