Talk:P-rep

From Wikipedia, the free encyclopedia

"... a p-value indicates the probability of obtaining a result by chance alone, ..." incorrect statement - see the article on p-value

also: it would be best to delete the "approximation" part again: if there exists a mapping from p-rep to p-value, then p-rep would have the same shortcomings as p, because it would carry exactly the same information. over-seeing the linked article i believe this "approximation" is obtained for a certain preassumed family of probability distributions on the sample, probably the distribution the "inventors" of p-rep believe to be the most common. But then again for this family of distributions p and p-rep are as similar as the approximation is good.

I'm not saying p-rep is scientifically useless, but very sometimes it seems that proponents for p-rep prefer it over p just because they don't understand the concept p. It appears to me as if there are assumptions necessery for calculation of p-rep, that are inherently arbitrary. The classical p-value does not need any such assumptions, which makes it more meaningful in an abstract sense. Obviously the 0.05 bound for p is ridiculously arbitrary, though. —Preceding unsigned comment added by 91.19.105.203 (talk • contribs) 14:18, 19 September 2007