Design and analysis of intra-subject variability in cross-over experiments

Area of Science:

• Biostatistics
• Experimental Design
• Statistical Inference

Background:

• Growing interest in inferential techniques for cross-over experiments.
• Emphasis on comparing intra-subject variability over inter-subject variability.

Purpose of the Study:

• To present a general approach for statistical inference in cross-over designs.
• To discuss statistical models for comparing treatment variabilities, particularly intra-subject variability.

Main Methods:

• Incorporation of t-variate random subject effects into three distinct statistical models.
• Development of maximum likelihood (ML) and restricted maximum likelihood (REML) approaches for parameter estimation.
• Consideration of a special case with closed-form variance component estimators.

Main Results:

• Methodologies for statistical inference in cross-over designs are developed.
• Parameter estimators are derived using ML and REML approaches.
• Illustrative analysis of three real-world datasets.

Conclusions:

• The study provides robust statistical methodologies for analyzing treatment variability in cross-over experiments.
• The proposed models and estimation techniques are applicable to various cross-over designs.
• The findings contribute to a better understanding of intra-subject variability in treatment comparisons.