Suppose there's a study where an interim analysis was performed to evaluate efficacy with the intention of stopping the study. However, since the predefined success criterion wasn't achieved, the study continued until completion. In this scenario, should the final analysis account for multiplicity, considering that no type-1 error occurred during the interim analysis?


1 Answer 1


Yes you would still need to make some correction to control the probability of a Type 1 error from your experimental procedure as a whole.

If you reset alpha to (say) 0.05 whenever the interim analysis doesn't suggest an effect, then overall your experiment has a probability greater than 0.05 of a type 1 error (so long as the interim analysis had a non-zero type 1 error probability).

That your particular instance of the experiment didn't stop early is not relevant, it's the long run probability of a type 1 error from the procedure as a whole that is important.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.