A Comparative Study of Power and Sample Size Calculations for Multivariate General Linear Models.

Area of Science:

• Social and Behavioral Sciences
• Biostatistics
• Psychometrics

Background:

• Repeated measures and longitudinal studies are crucial in social and behavioral research.
• Accurate sample size calculation is essential during study planning for robust results.
• Existing methods for power and sample size calculations may not fully address complex longitudinal data structures.

Purpose of the Study:

• To present methods for power and sample size calculations for normal outcomes in repeated measures and longitudinal studies.
• To extend existing generalized estimating equation and likelihood-based approaches.
• To accommodate both fixed and random effects models within a unified framework.

Main Methods:

• Utilizes multivariate general linear models as the foundational analytical framework.
• Extends generalized estimating equation (GEE) and likelihood-based statistical approaches.
• Incorporates both fixed and random effects models for comprehensive analysis.

Main Results:

• Provides direct extensions of established statistical methods for power and sample size calculations.
• Demonstrates the accommodation of both fixed and random models, enhancing flexibility.
• Illustrates the practical application of the proposed methods with a child development study example.

Conclusions:

• The proposed methods offer a robust framework for sample size determination in longitudinal and repeated measures studies.
• The inclusion of both fixed and random effects models provides greater utility for researchers.
• Monte Carlo simulations confirm the adequacy of the developed sample size formulas.