First, some context – since ZH is the site that most frequently discusses POMO, I have been posting some of my work there in the comment sections pretty regularly. There’s typically quite a bit of “high quality click-through,” especially from some well-known firms in NYC and London. This morning, I posted a comment on my research from this morning on how the ZH’s Submitted/Accepted ratio post yesterday was not totally true on one of the usual ZH POMO posts. In under 30 seconds, it was deleted, and soon after the following email conversation began. Here are my two cents:
- When did ZH get serious about monetizing itself via ad revenue? Is there any talk of going public or is anyone interesting in forming a coalition to LBO them?!
- Since when should the null hypothesis be “conspiracy theory = true” ? Or did Tyler miss STATS100?
Tyler, 9:17am:We appreciate constructive and critical thought. If you would like to buy advertising space, however, to promote your blog, please advise
Me, 9:18am: How do I copy-paste the 1000 words necessary to properly respond into the comment section?
Tyler, 9:21am: I believe blasting every even remotely POMO related post with linkbacks is sufficient. Looking at a history of your comments for the past week shows behavior that some would characterize as self-promoting spam, which by the way is forbidden on zero hedge
Me, 9:25am: I won’t deny that it’s definitely strategic and will stop if it’s really a problem. You let plenty of crazier shit through, which only provides fodder for people who don’t want to believe you’re a credible source of information. I figured that the appearance of some reasonable and rigorous discussion in the comments would make both of us better off.
Tyler, 9:29am: By all means: however for you to post something with a headline that says “zero hedge is wrong” and to have a conclusion “the dollar volume and magnitude of the change are statistically significantly related to the POMO accepted-to-submitted ratio, but the direction is not really guaranteed” seems a little self defeating and itself is somewhat statistically suspect. Did you actually refute the null hypothesis?
As for people who don’t want to believe about ZH being a credible source of information, we really couldn’t give a rat’s ass, and wish them safe travels elsewhere.
Me, 9:33am: You mean I used a headline that was somewhat hyperbolic? Fair point but probably the pot calling the kettle black. As far as the data and numbers, I put all my data and code online, so anybody can download it and run it on their own.
Anyway, regardless of this tiff, thanks for what you do. The world is probably a better place for the site, not just in a feel-good bullshit kind of way.
N.B.: This seems like pretty clear fair use of these emails. If anyone thinks differently (e.g., Tyler(s)), feel free to let me know before calling your lawyer.
The “null hypothesis” is the converse of whatever you wish to test, so it CAN be true, if you are testing the falsity of the proposition.
Fair point, but isn’t the burden on the person trying to prove the claim in the first point? If science ran this way without Occam’s razor, I could get away with arguing that jellyfish are responsible for the creation of the moon. It is possible that I misinterpreted his implication.
Anyway, if he had understood the post, he would have seen that there is very weak support for rejecting H_0: \beta = 0, as I show in the first regression. You have to accept a pretty p-value near 0.1, but in this very simple GLM setup the estimate – s.e. is positive.