Nonlinear and Quasi-Simplex Patterns in Latent Growth Models

Abstract

In the SEM literature, simplex and latent growth models have always been considered competing approaches for the analysis of longitudinal data, even if they are strongly connected and both of specific importance. General dynamic models, which simultaneously estimate autoregressive structures and latent curves, have been recently proposed in the literature. We discuss the properties of Autoregressive Latent Trajectories (ALT) with the aim of deriving their relationship with nonlinear growth models. We show how the quasi-simplex part of the ALT admits an equivalent nonlinear growth representation. A simulation study is performed to examine how the relationship holds in the presence of polynomial and bounded growths over time, whereas an empirical application on student achievement highlights the usefulness of the equivalence. The evaluation of the formative process in the European University system has been assuming an ever increasing importance since the beginning of the Bologna process. In this context, the analysis of student performances and capabilities using different approaches plays a fundamental role.

Notes

*Significant at 5% level.