Multi-Level Multi-Growth Models: New opportunities for addressing developmental theory using advanced longitudinal...

Area of Science:

• * Developmental Psychology
• * Quantitative Psychology
• * Longitudinal Data Analysis

Background:

• * Longitudinal models are powerful for studying within-person processes using repeated measures.
• * Growth models offer a flexible framework for complex, non-linear trajectories and time-varying covariates.
• * Existing growth models can be extended to incorporate multiple simultaneous growth processes on a single outcome.

Purpose of the Study:

• * To outline an extension of growth models for simultaneously modeling multiple growth processes.
• * To demonstrate how this extension can be achieved by treating multiple processes as time-varying covariates.
• * To highlight the critical role of study design, beyond statistical models, for parameter recovery.

Main Methods:

• * Extending general growth models to include multiple time-varying covariates representing distinct growth processes.
• * Utilizing simulation studies to evaluate model performance under various theoretical conditions.
• * Comparing model behavior in cohort and accelerated longitudinal designs, emphasizing planned missingness.

Main Results:

• * The proposed extension allows for disaggregating effects of factors like age and practice/treatment in repeated assessments.
• * Simulations demonstrate the utility of modeling age- and puberty-related effects during adolescence.
• * Planned missingness in observations is identified as key for successful parameter recovery in these extended models.

Conclusions:

• * Statistical models alone are insufficient; study design is paramount for accurate parameter recovery in multi-process growth models.
• * The outlined method provides a framework for addressing complex developmental theories.
• * Future research can leverage this approach for substantive investigations into developmental trajectories.