The ability of a study to detect a dose-related effect when one exists, is called its power. Power is defined statistically as the probability of obtaining a significant result of a particular size, given the experimental conditions. It depends not only on study size and variability, but also on the size of effect it is desired to detect, the particular significance test chosen and the validity of the assumptions.
Power calculations are illustrated for the t-test, normal test for proportions, and the analysis-of-variance F-test. Some suggestions for increasing the power are given.
