ABSTRACT
In essence, ANCOVA works by simply including the additional variable (the covariate) in the regression, but, by doing so, it allows the effects of that variable (such as age, or aptitude scores) to be separated out from the response variable. In this way, ANCOVA is like partial correlation, which we saw in Sections 6.5.1 and 6.5.2, because it includes the variable whose effects we want to “partial out” in the analysis in order to separate them from the other effects. ANCOVA works like the repeated-measures or mixed-effect designs that we saw previously in Chapter 12 as well to reduce the amount of variability in the model that is unexplained. If we think that scores on an aptitude test help account for the variability on the response variable, then by including the aptitude test in the design we help reduce the amount of variability that is unexplained.
