P-values are the gold medal of research. Finding a p-value less than .05 in your clinical trial is the pinnacle of achievement. Even though there is virtually no difference between p=.047 and p=.053, get the former and your research is a success, the latter and it is consigned to the wastepaper basket of history.
The reason people compete in the Olympics is to win a gold medal. The reason researchers do clinical trials is to prove statistical significance.
We can live with occasional cheats in the Olympics. We can live with occasional rogue outcomes. What we cannot live with is knowing that only 1 in 5 medals was legitimate. Or that the BEST chance the result was fair was 26%.
Yet those are the realities of p values – according to Professor David Colquhoun, FRS.
Anyone thinking about practicing evidence-based (condition-based and population-based) better have a good understanding of the statistics they are using, that’s why I am posting here 36 minutes of the worst (in terms of production) but most interesting video you have ever seen.
[s3video url=”https://www.youtube.com/watch?v=tRZMD1cYX_c” responsive=”on” /]
Why aren’t practitioners rioting about this? Why aren’t those who have taught research apologising to the healthcare community for getting it wrong all these years?
This error of false discovery rates taints nearly all clinical trials done for the last 50 years, yet it is receiving little to no publicity.
If all you have is statistics you better get those statistics right. The fact clinical trials got it wrong for so long is a disaster, and nobody seems to care.