Detecting Transition Points in the Slope-Intercept Relation in Linear Latent Growth Models: Bayesian Semiparametric Approach

• In Event: AERA Roundtable Session Sunday 8:00 am Four Seasons Ballroom 2-3
  In Roundtable Session: Applications and Advances in Structural Equation Modeling (Table 6)

Abstract

In a linear latent growth model parameterized by intercept and slope factors, their relation is often of interest. The model typically captures this through the covariance parameter, assuming linearity. However, what if it’s not? For instance, below a certain intercept threshold there might be no relation, with the intercept and slope only related above it. Thus, while the growth model is linear, the relation between the growth factors is not, indicating a transition point. To address this, we propose a transition point detection method after modeling nonlinear relations using Bayesian P-splines using MCMC techniques, combining spline coefficient differences with Bayesian model comparison. Simulation results demonstrate our approach’s effectiveness within single transition point scenarios, offering deeper insights into growth dynamics.

Authors

• Dayeon Lee, University of Maryland

• Gregory R. Hancock, University of Maryland