Functional analysis of variance involves testing for differences in functional means across k groups in n functional responses. If a significant overall difference in the mean curves is detected, one may want to identify the location of these differences. Two different follow-up testing methods are discussed and contrasted. A point-wise test proposed by Cox and Lee (2008) is compared to a test based on regions of the functional domain. The methods are contrasted in terms of weak control of the family-wise error rate, strong control of the family-wise error rate, and power.