Analysis of Repeated Measurements from Animal Experiments using Polynomial Regression

Abstract

Methods are presented for analyzing repeated measurements data (also called profile or growth curve data) from animal experiments when the "split-plot-in-time" analysis is not valid. The procedure is to first fit a polynomial to each animal's profile of measurements. Appropriate analyses of variance are then performed on the regression coefficients so obtained to test hypotheses posed. This approach ordinarily produces more powerful tests than if the multivariate analysis of variance is applied to the original profile of measurements. It also allows a multivariate analysis to be performed in cases where it would not be possible otherwise, (for example, data with missing observations or experiments in which there are more measurements per animal than there are animals). Finally, one can obtain a fitted curve for each treatment. These procedures are applicable to most experimental designs and all methodology presented can be efficiently computed with existing statistical computer programs. Copyright © 1983. American Society of Animal Science. Copyright 1983 by American Society of Animal Science.