Replies to post #153868 on Biotech Values

[iwfal]

I thought p-values took into account small numbers in a trial and there would thus be a need for more robust results to attain stat sig results.

Correct.

And I thought the definition of stat sig means that results, whether they occur in a few patients or in a large number, reflect a pattern and are not just due to chance.

Correct again.

The only counter point is that smaller sample size makes it easier for a 'single' experimental bias to somehow taint the results. E.g. classically, for instance, all patients being from one trial site that adjudges patients differently from everyone else. In something like hcv (with very repeatable and objective criteria) I would take a small trial p=0.05 over a large trial p=0.5 any day (assuming roughly equal size final trials). But in psychiatric disease I'd probably take the large trial because the trial sites are likely to judge very differently. And cancer I'd put somewhere between - e.g. if both were planning a 300 event ph iii, I'd take the trial that was stat sig with 20 event over the one that was stat sig at 100 patient. But not by a large margin.