Comment Re:Whoa there... (Score 1) 397
I think you're giving too much credit to the TFA. While it's true that you can essentially interchange between R and Chi-squared with some easy statistics, that doesn't mean that this guy was thinking that.
Everything in TFA indicates he just saw R as being low, thus declared it insignificant. Later on in the TFA, he declares R > 0.10 significant (complete with a citation, which I cannot imagine to be even remotely correct). This assertion makes no sense, as R does not take sample-size into account, where every other significance test on the planet does for obvious reasons.
Honestly, you give the author way too much credit. You're right, he got the conclusion right. But not because of any actual sound statistical logic.
p.s. I really think your guy's argument down below is just a matter of semantics. Statisticians will argue that correlation will never imply causation -- other disciplines (especially the ones in which just about everything can be controlled, ie physics) will go a bit farther, essentially saying "well everything else is controlled for, so we must be proving causation", and then they just accept being wrong as much as their alpha allows.
Everything in TFA indicates he just saw R as being low, thus declared it insignificant. Later on in the TFA, he declares R > 0.10 significant (complete with a citation, which I cannot imagine to be even remotely correct). This assertion makes no sense, as R does not take sample-size into account, where every other significance test on the planet does for obvious reasons.
Honestly, you give the author way too much credit. You're right, he got the conclusion right. But not because of any actual sound statistical logic.
p.s. I really think your guy's argument down below is just a matter of semantics. Statisticians will argue that correlation will never imply causation -- other disciplines (especially the ones in which just about everything can be controlled, ie physics) will go a bit farther, essentially saying "well everything else is controlled for, so we must be proving causation", and then they just accept being wrong as much as their alpha allows.
