Detecting Transition Points in the Slope-Intercept Relation in Linear Latent Growth Models

Area of Science:

• Statistics
• Psychometrics
• Developmental Psychology

Background:

• Latent growth models (LGMs) typically assume linear relationships between intercept (α) and slope (β) factors.
• The covariance parameter in standard LGMs inherently assumes linearity, which may not always reflect complex developmental processes.
• Nonlinear relationships between α and β, potentially with distinct segments and transition points, are common but often unaddressed.

Purpose of the Study:

• To develop and validate a semiparametric method for modeling nonlinear α-β relations in LGMs.
• To introduce a flexible approach capable of detecting transition points in the α-β relationship.
• To provide a more nuanced understanding of individual growth trajectories by accounting for segmented α-β associations.

Main Methods:

• A two-stage semiparametric approach combining Bayesian P-splines and segmented regression.
• Bayesian P-splines are utilized for flexible nonlinear modeling of the α-β relationship.
• Segmented regression is employed for detecting transition points in the α-β association.

Main Results:

• The proposed method effectively models nonlinear α-β relationships, including those with a single transition point.
• Simulation studies demonstrate the approach's accuracy in estimating parameters and identifying transition points.
• An empirical data illustration confirms the practical utility and interpretability of the method.

Conclusions:

• The semiparametric approach offers a powerful tool for analyzing complex, nonlinear α-β relationships in latent growth modeling.
• This method enhances the understanding of developmental processes by identifying critical transition points in growth.
• It provides a more nuanced and accurate representation of individual differences in growth trajectories.