Multiple Comparisons of Parametric Models in Longitudinal Studies

Abstract

In longitudinal analysis, interest lies often in comparing parametric models over time. For example, if the trend over time is linear, commonly a test is performed to evaluate whether the rate of change in the mean response is the same for all treatment groups. For parametric models, treatment profiles can easily be compared using confidence bands. The authors propose simultaneous confidence bands and discuss two methods of calculating the associated critical point for comparing several parametric models over time. This issue has never been addressed before in the context of longitudinal studies. An example illustrates the proposed method.