Standard Article

Analysis of Variance for Longitudinal Data

  1. Graham Dunn

Published Online: 15 JUL 2005

DOI: 10.1002/0470011815.b2a12001

Encyclopedia of Biostatistics

Encyclopedia of Biostatistics

How to Cite

Dunn, G. 2005. Analysis of Variance for Longitudinal Data. Encyclopedia of Biostatistics. 1.

Author Information

  1. University of Manchester, Manchester, UK

Publication History

  1. Published Online: 15 JUL 2005


This entry describes how traditional analysis of variance (ANOVA) techniques might be adapted for use in the analysis of serial measurements. Typically, one might recognize data coming from a longitudinal study as coming from a nested or split-plot design with serial or repeated measurements (subplots) being nested within each of the subjects (plots). The difference between the results of a typical split-plot experiment and that involving repeated measurements arises from the fact that the latter are likely to be serially correlated. Methods of adapting ANOVA methods to allow for serial correlation (i.e. the Greenhouse–Geisser correction) are described, together with suggestions for alternative data analytic approaches.


  • longitudinal data;
  • serial measurements;
  • repeated measures;
  • analysis of variance;
  • multivariate analysis of variance;
  • ANOVA;
  • Hotelling's T2 statistic;
  • split-plot experiments;
  • crossover designs;
  • random effects models;
  • nesting;
  • growth curves;
  • time trends;
  • paired t-test;
  • sphericity assumption;
  • compound symmetry;
  • Mauchly test;
  • orthogonal polynomial contrasts;
  • maximum likelihood (ML);
  • residual maximum likelihood (REML);
  • restricted maximum likelihood (REML);
  • response feature extraction