Numbers in trials are reflected in the null hypothesis and the relative chances of type 1 and type 2 errors.
Type 2 error comes into its own in comparitor trials, eg antidepressant A versus antidepressant B, where the latter is the old treatment. The null hypothesis states that there is no difference. If A seems better, test for type 1 error. If they appear to be equal, test for type 2 error, since there may be a need to increase trial power by increasing the numbers of people in the trial.
Add information to this page that would be handy to have on hand during a consultation, such as a web address or phone number. This information will always be displayed when you visit this page