A guide to the analysis of growth curve data with special reference to SAS.

The analysis of data variously termed growth curve data, repeated measurements data, or profile data is discussed. Typically, one has measurements taken at several points in time for each experimental unit. Usually there are also treatments randomized to these experimental units according to a suitably chosen design. The conditions under which the "split-plot-in-time" analysis is valid are given and the Geisser-Greenhouse approximation is discussed when it is not valid. Often, however, a low degree polynomial can be fitted to the profile of measurements for each experimental unit, thereby reducing the dimension of the profile considerably. An exact multivariate analysis can then be performed on the regression coefficients. This removes many of the objections usually raised against multivariate techniques. All of this is easily implemented on the statistical analysis system (SAS) and sample programs are given.