Analysis of covariance
ANCOVA or analysis of covariance is a variant of the ANOVA model in which the statistical effect of a nuisance variable is removed mathematically from the analysis in order to clarify the relations between the independent and the dependent variables (Belin and Normand 2009). The optimal situation would be if the independent variable levels or groups were not related to the nuisance variable. However, if the nuisance variable is related to the dependent variable and if the nuisance variable is systematically represented among the independent variables, ANCOVA may be used to partial out the statistical effect of the nuisance variable or covariate. This is not a substitute for removing the effect through experimental design. For example, level of education may be statistically related to performance on a memory test. If two groups of depressed and nondepressed individuals differ systematically on the basis of their level of education, any...