yeah well An anova test is used with three or more sets of data and is used to determine if there is a significant difference between the groups. Generally one might run just multiple t-tests but this isn't appropriate and can skew your data.
Let me stop you at your first query (in the link you provided). If the groups are equivalent at the pre-test stage, then there is no need to use ANCOVA (MACOVA). That said, just use ANOVA (MANOVA) with the usual parametric post-hoc pairwise analyses to compare groups.