Analysis of Covariance

Abstract

Traditionally, analysis of covariance (ACOVA) has been used as a tool in the analysis of designed experiments. Suppose one or more measurements are made on a group of experimental units. In an agricultural experiment, such a measurement might be the amount of nitrogen in each plot of ground prior to the application of any treatments. In animal husbandry, the measurements might be the height and weight of animals before treatments are applied. One way to use such information is to create blocks of experimental units that have similar values of the measurements. Analysis of covariance uses a different approach. In analysis of covariance, an experimental design is chosen that does not depend on these supplemental observations. The concomitant observations come into play as regression variables that are added to the basic experimental design model.