(Total Views: 296)
Posted On: 09/14/2017 1:50:52 PM
Post# of 72448
If you read Leo's email response regarding the interim results, he specifically cites "alpha" risk.
Alpha or Type I error, in contrast to Beta or Type II error http://www.nejm.org/doi/full/10.1056/NEJM1990...#t=article
Type 1 error would mean that there is no correlation, but we think there is one, meaning P does not work, but we think it does.
Was this a clue? Leo did not cite beta/type II risk (it looks like P works, but in fact it does not).
This of course is based on the premise that Leo may get some aggregate/blinded data.
If you were to negotiate a deal with BP, I'd think you'd want your statistical significance as strong as possible and minimize your type I (and type II) errors. This would remedy the number one reason BPs have small companies do trials before partnering- to eliminate risk...
So is Leo stating "we think P works, so we don't want to cast doubt on that"?
Alpha or Type I error, in contrast to Beta or Type II error http://www.nejm.org/doi/full/10.1056/NEJM1990...#t=article
Type 1 error would mean that there is no correlation, but we think there is one, meaning P does not work, but we think it does.
Was this a clue? Leo did not cite beta/type II risk (it looks like P works, but in fact it does not).
This of course is based on the premise that Leo may get some aggregate/blinded data.
If you were to negotiate a deal with BP, I'd think you'd want your statistical significance as strong as possible and minimize your type I (and type II) errors. This would remedy the number one reason BPs have small companies do trials before partnering- to eliminate risk...
So is Leo stating "we think P works, so we don't want to cast doubt on that"?
(1)
(0)
