What is it about?
This study proposes improved methods for the mixed model for repeated measures (MMRM), a common approach for analyzing longitudinal or clinical trial data. Standard MMRM analyses assume “orthogonality” between fixed effects and covariance parameters, an assumption that rarely holds in practice — especially with small or unbalanced samples. When this assumption fails, standard errors are often underestimated, leading to overconfident results. To solve this, the authors introduce two small-sample adjustment methods that correct the bias and produce more reliable inference. They evaluate these adjustments through simulations and provide an R package for practical use.
Featured Image
Photo by Bozhin Karaivanov on Unsplash
Why is it important?
Small or incomplete data sets are common in medical and social research, where maintaining accuracy is essential. The proposed corrections help prevent false-positive findings and strengthen the credibility of MMRM-based analyses. By combining solid statistical theory with real-world applicability, this research supports more reproducible and trustworthy conclusions in longitudinal studies and clinical trials.
Perspectives
These adjustments are useful in studies with missing data, unequal variances, or subgroup analyses with limited sample size. They enhance the robustness of treatment effect estimation and can be easily implemented in R. Future extensions may include multivariate outcomes or joint modeling, broadening their impact across applied statistics and regulatory science.
Kazushi Maruo
University of Tsukuba
Read the Original
This page is a summary of: Small sample adjustment for inference without assuming orthogonality in a mixed model for repeated measures analysis, Journal of Biopharmaceutical Statistics, October 2024, Taylor & Francis,
DOI: 10.1080/10543406.2024.2420632.
You can read the full text:
Contributors
The following have contributed to this page
