Analysis of Covariance

Overview: In experimental methods, a central tenet of establishing significant relationships has to do with the notion of random assignment. Random assignment solves a couple of problems. Statistically, it ensures that, in the main, the resulting probability will be independent of the starting conditions of an experiment. Secondly, it is a way of establishing parity, which is to say that it is a method of controlling for what you don’t know. ANCOVA is a method that can be thought of as a cross between ANOVA and Regression. It is, in fact, an ANOVA where the effects of some other variable have been controlled for statistically. There are three reasons why one would choose to use an ANCOVA: 1. Reduction of within group or error variance – to increase the sensitivity of a test of main effects and/or interactions, by reducing the error term. 2. Elimination of systematic bias – to adjust the means on the d.v. to what they would be if all subjects scored equally on the c.v. 3. Stepdown Analysis – to compare scores on a d.v. after they are adjusted for scores on another DV, which is treated as a c.v. (MANOVA model).