Analysis of repeated measures designs with changing covariates

A model proposed by Patel (1986) is re-examined. It is shown that the model can be written as seemingly unrelated regressions and estimation can be carried out simply. Estimation under constraints given by the general linear hypothesis is also discussed and a two-stage procedure proposed. Connexions with seemingly unrelated regressions with constraints between the regression parameters and the sum of profiles model of Verbyla & Venables (1988) are also given.