Residual analysis in random regressions using SAS and S-PLUS

Area of Science:

• Statistics
• Biostatistics
• Quantitative Genetics

Background:

• Random regression models are crucial for analyzing longitudinal data and repeated measures.
• Residual analysis is essential for validating model assumptions and identifying deviations.
• Existing methods may lack comprehensive tools for advanced residual diagnostics.

Purpose of the Study:

• To introduce the Random Regression Residual Analysis Program (RRRAP).
• To provide a software package for performing detailed residual analyses in random regression models.
• To facilitate the assessment of model fit and detection of outliers.

Main Methods:

• Utilizes SAS PROCEDURE MIXED for statistical inference.
• Employs both elementary-level and individual-level residuals.
• Incorporates S-PLUS programs for residual transformation and diagnostic statistics.

Main Results:

• S-PLUS programs offer orthogonalization of correlated residuals.
• Provides statistics and plots for checking model assumptions.
• Enables detection of outlying individuals and assessment of model fitting.

Conclusions:

• RRRAP is a valuable tool for researchers conducting random regression residual analysis.
• The package enhances the ability to validate statistical models and interpret results.
• Facilitates robust analysis through advanced residual diagnostics and outlier detection.