In other words, the smaller the difference between two conditions (say, experiment and control group, or Treatment A and Treatment B), the more experimental subjects you need in order to show that the difference is real, and not simply due to chance.
