News Focus
News Focus
icon url

OFP

07/18/19 8:34 AM

#201933 RE: OFP #201797

Quote:A low p-value is a good thing.

You don't say?

What you fail to realize how AVXL works with statistics and the fact that "analysis of 32,875" species made it inevitable that strong correlations exist. If you think AVXL appropriately adjusted their statistic then you haven't been watching.

https://www.dummies.com/education/AVLX/statistics/what-a-p-value-tells-when-the-company-isouttofoolyou/



If you look for random correlations among thousands you will likely find a lot through chance alone. If you then take your top correlations and say let's run a statistic to see if these are independent (i.e. null hypothesis = not correlated). What you'll find with this silly set up is "OMG, there is <5% chance these are independent...Eureka!"