Comparing treatments under growth curve models: Exact tests using generalized p-values

In this paper we consider a simple growth curves problem for which no exact tests are available. Exact tests using generalized p-values are developed for comparing a number of treatments with balanced and complete repeated observations made under growth curve models with the compound symmetry covariance structure and equal residual error variances. The tests are also established for models with unequal residual error variances. In addition, tests using generalized p-value for the inter-subject variance component (random effects) are also derived for the extended models. The procedures are illustrated with two examples. (C) 1998 Elsevier Science B.V. All rights reserved.