Analysis of variance for repeated measures. Data: a generalized estimating equations approach

Area of Science:

• Biostatistics
• Statistical Modeling
• Longitudinal Data Analysis

Background:

• Analysis of repeated measures data requires accounting for the dependence between observations from the same subject.
• Existing methods often depend on specific distributional assumptions and study designs.
• A unified approach is needed for robust and flexible analysis of such data.

Purpose of the Study:

• To examine the application of generalized estimating equations (GEE) for analyzing repeated measures data.
• To evaluate the performance of GEE with an identity link for estimating mean responses.
• To demonstrate the simplicity and consistency of GEE compared to specialized methods.

Main Methods:

• Utilized generalized estimating equations (GEE) as proposed by Liang and Zeger.
• Applied GEE with an identity link function to estimate the mean response for each covariate combination.
• Analyzed various types of repeated measures data using this unified approach.

Main Results:

• The computations for fitting the GEE models were found to be exceptionally simple.
• Numerical examples demonstrated that the GEE approach yields estimation results consistent with specialized methods.
• Hypothesis testing results using GEE were also found to be consistent with established techniques.

Conclusions:

• Generalized estimating equations provide a unified and computationally simple method for analyzing repeated measures data.
• The GEE approach effectively accounts for the dependence structure in longitudinal data.
• GEE offers a reliable alternative to specialized methods, providing consistent estimation and hypothesis testing.