Statistical "Significance" is one the most abused concepts in the pharma/medical industry. The Sola nonsense is a great case study.
https://eighteenthelephant.wordpress.com/2016/04/29/how-do-i-hate-p-values-let-me-count-the-ways/
You completely mis-characterize the argument. My intent is not to trash statistics, but to point out that it is abused, in particular the use of p-values. Perhaps, if you would actually studied the subject you might understand. Bottom line, the Lilly Sola team allowed itself to be deceived by a p-value that was randomly "significant". A classic case of p-value hacking.Yes, statistics are awful. Let's just go with our gut on things. We don't actually need to try and prove that our results aren't a product of chance - as long as I think our drug works, we're good.
You just revolutionized the entire clinical trial process. We'll call it the Trump Method of science - "If I think it, it must be true."
You completely mis-characterize the argument. My intent is not to trash statistics, but to point out that it is abused, in particular the use of p-values. Perhaps, if you would actually studied the subject you might understand. Bottom line, the Lilly Sola team allowed itself to be deceived by a p-value that was randomly "significant". A classic case of p-value hacking.
How many reps reading this have gone around spouting about some subset analysis with "statistical significance". In Oncology that was how we operated...well that was years ago anyway.
As someone who makes a living in statistics, your argument has gone from the sublime to the ridiculous.Your argument is that a significant p-value isn't significant at all. Like it or not, you're trying to trash the entire discipline of statistics. Good luck with that argument.
And, thanks for pointing out something you don't like without any sort of alternative method. I was wrong about you being like Trump. You're much more like Ross Perot.
You completely mis-characterize the argument. My intent is not to trash statistics, but to point out that it is abused, in particular the use of p-values. Perhaps, if you would actually studied the subject you might understand. Bottom line, the Lilly Sola team allowed itself to be deceived by a p-value that was randomly "significant". A classic case of p-value hacking.
How many reps reading this have gone around spouting about some subset analysis with "statistical significance". In Oncology that was how we operated...well that was years ago anyway.
The ASA Board was also stimulated by highly visible discussions over the last few years. For example, ScienceNews (Siegfried 2010) wrote: “It’s science’s dirtiest secret: The ‘scientific method’ of testing hypotheses by statistical analysis stands on a flimsy foundation.” A November 2013, article in Phys.org Science News Wire (2013) cited “numerous deep flaws” in null hypothesis significance testing. A ScienceNews article (Siegfried 2014) on February 7, 2014, said “statistical techniques for testing hypotheses …have more flaws than Facebook’s privacy policies.”As someone who makes a living in statistics, your argument has gone from the sublime to the ridiculous.
Geez, I wonder what the American Statistical Association has to say about it?
Oh, well here it is...
http://amstat.tandfonline.com/doi/abs/10.1080/00031305.2016.1154108
Read it, if you can.
Alternatives? Don't hack p-values and don't quote them as some gospel. Sir Ronald Fisher, who designed the concept, never intended it to be used as a binary selection criterion.
