Repeated measurements, design and analysis for

Abstract

This article discusses important aspects of the design and statistical analysis of repeated measurements studies. The scope of study designs includes split‐plot experiments, longitudinal studies, crossover study designs, and sources of variability studies. Analysis methodologies are organized by univariate and multivariate methods, along with comparable nonparametric strategies. ANOVA and ANCOVA methods are discussed for univariate summary functions of the repeated measurements for each subject, while mixed models for repeated measurements (MMRM) are presented for continuous outcome data and generalized estimating equations (GEE) are presented for categorical outcome data. Nonparametric methods include versions of the extended Mantel–Haenszel statistic, extensions of the Wilcoxon signed ranks statistic, the multivariate Kruskal–Wallis statistic, and multivariate Mann–Whitney estimators. Nine examples are provided to illustrate applications of the statistical methods to specific study designs.