Assessing mediational processes using piecewise linear growth curve models with individual measurement occasions

Area of Science:

• Statistics
• Biostatistics
• Longitudinal Data Analysis

Background:

• Longitudinal studies often involve concurrent processes with potentially non-constant change over time.
• Existing multivariate growth models primarily focus on correlated, rather than unidirectional, relationships between outcomes.
• Unidirectional pathways in longitudinal data require specialized statistical approaches for accurate estimation.

Purpose of the Study:

• To develop and present novel longitudinal mediation models for analyzing unidirectional relationships among multiple growth processes.
• To utilize a linear-linear piecewise functional form to effectively capture complex change patterns over time.
• To provide a statistical framework for estimating direct and indirect effects in longitudinal mediation.

Main Methods:

• Development of two distinct longitudinal mediation models.
• Application of a linear-linear piecewise functional form to model change patterns.
• Validation through simulation studies and real-world data analysis.

Main Results:

• Simulation studies confirmed the proposed models yield unbiased point estimates and accurate coverage probabilities.
• Empirical analyses successfully demonstrated the models' capability to estimate direct and indirect effects of covariates on outcome changes.
• The models provide a robust method for understanding mediational processes in longitudinal research.

Conclusions:

• The developed longitudinal mediation models offer a reliable method for analyzing unidirectional associations in longitudinal data.
• These models enhance the understanding of complex developmental trajectories by quantifying indirect effects.
• The study provides practical tools and code for researchers working with longitudinal mediation.