Bioequivalence and Bioavailability Forum

Dear shri!

❝ Please tell me that is it good...

And what if I don't think that it's good?

❝ ...to take pilot or pivotal data in consideration whose power < 80 and the

❝ study was failed, for sample size estimation.

A pilot study can't fail (unless you sc¤ed something up in the design: sampling schedule, bioanalytics) - it's always informative.

A posteriori power doesn't exist; see these posts: #419, #1801, #2338, #4596; and yet another interesting (but lengthy) thread.

If you want to step into the muddle of a posteriori power calculations - which I don't hope - any 'failed' study will show low power. So what? We know already that our assumptions in study planing did not come out the way we expected ('bad' T/R-ratio, higher CV and/or drop-out rate).
For sure you can use the values from a failed pivotal or a pilot in sample size estimation. Keep in mind to adjust for the usual uncertainties (e.g., CV_intra from a small pilot).