Comment of the Day: JEC: HOW WOULD HEALTH-CARE REFORM AFFECT PATIENT HEALTH?: "I'd also point out that Marty is performing the 'null hypothesis bait-and-switch', which relies on nobody noticing that you've swapped the maintained and null hypotheses... http://www.bradford-delong.com/2017/07/must-read-naughty-naughty-marty-you-know-better-you-say-patients-in-the-oregon-medicaid-study-show-no-signif.html?cid=6a00e551f08003883401b8d299d84a970c#comment-6a00e551f08003883401b8d299d84a970c
And he's pretty blatant about it:
...enrollees in Medicaid show no significant improvement in clinical physical health outcomes. This was the main finding of a large 'natural experiment' supported by the federal government...
This is flatly false.
There is no way to read that statement other than "The study rejected a null hypothesis of non-zero improvement in clinical physical outcomes in favor of the maintained hypothesis of zero improvement." That is what "finding" means. And it's a specific, provable, knowing lie.
The really interesting question is why Feldstein thinks he can get away with this transparently false claim. Is is the post-truth chaos of Trump-era politics? I don't think so. As it happens, this specific move, treating "failure to reject the null" as a "finding" in support of of the hypothesis one actually maintains, has long been characteristic of our favorite school(s) of macroeconomics. ("See: this test fails to reject the EMH! EMH must be true! Yay team EMH!")
So...consider it yet another example of why the deteriorating intellectual standards in certain corners of the discipline matters A LOT.
