Hi,
Interesting question and interesting debate. I have a different opinion! It is not true that the larger the sample size, the better or that sample size for clinical trials is to only to ensure enough power for the experiment. Obviously, from the statistical point of view, larger (not biased) samples, represent more information. And also, it is obvious that is important to have the experiments
properly powered. However, the major problem in clinical trials is exactly the opposite.
The ethics in human experiments (and also animal experiments) is a very sensitive subject. One of the fundamental principals of clinical trials (and this applies mostly to human subjects) is the principal of clinical equipoise. In very simplistic terms, this implies that to be able to randomize patients to two arms of a trial, there should be no evidence of the superiority of any of the arms (again this is much more complex than what I just stated).
For the argument sake, suppose in a trial comparing A and B, treatment A is in fact better than B. If a sample is "too big", there will be a point where patients are being randomized despite existing enough evidence favouring A (even if the researcher does not perform an interim analysis, the information is already collected!). At that point the principle of equipoise is broken and the experiment is not acceptable from an ethical point of view.
Thus, in my opinion, the importance of sample size calculation arises mainly because of the equipoise principle.
This does not mean that a good research plan should disregard sample size justification. In any case, there are many situations in observational studies, where the issue is not what size should my sample have but rather what power do I have with this sample because there are no mode individuals (e.g. rare diseases). Actually, I do believe that in observational studies, power analysis should be the standard requirement and not sample size justification (of course this should be understood as a general recommendation and not a rule).
Best,
Armando Teixeira-Pinto
