On box score methodology as illustrated by three reviews of overtraining reversal effects.