ABSTRACT
Using ANCOVA allows levels of an IV to be compared after an adjustment has been made to allow for the differences between the levels of the IV which exist in a continuous variable (a covariate). The covariate could be a different variable from the DV or it could be pre-treatment values of the DV.
The results of an ANCOVA are interpreted most straightforwardly when the allocation to different levels of the IV has been made randomly. When allocation is not random, including in pre-existing groups, then greater caution has to be taken over use of ANCOVA. When there are more than two levels of the IV, tests of contrasts can be conducted after an ANCOVA but, in addition to the factors which have to be considered when following up an ANOVA, the method of calculating the contrast has to take into account whether the allocation to groups was random and whether the covariate was fixed or random.
