Residual plots for repeated measures.

Multivariate data are difficult to analyse partly because of difficulty in looking at the data. Repeated measures data have a special structure that makes the parallel plot particularly effective for viewing the data. In this paper we use parallel plots to display not just raw data but also residuals from standard models fit to repeated measures data. The plots are useful for determining how well a particular model fits the data, identifying outlying observations and suggesting terms missing from the linear predictor.