Whenever there is a substantial correlation between the performance prior to treatment and that during (or following) treatment, as in the case being discussed, some measure of the change will be more sensitive to treatment differences than an absolute measure.