Monday, May 20, 2013

Details

I've got an RA this week, and I've got him working on a lit review on empirical topics like replication, p-hacking, sub-group analysis, pre-specified analysis plans, publication bias, etc.

Anyway, that reminded me of this New Yorker article on how scientific results tend to fail replication: "The Decline Effect." Unfortunately, this article is by Jonah Lehrer, who lost his job because he made up quotes. In a mildly interesting diversion, I came across these articles about Lehrer and the TED conference.

Felix Salmon wrote about Lehrer and the TED conference, and though Lehrer never spoke at TED, he lumps them (and Malcolm Gladwell and David Brooks) together. Appropriately, I think. Basically, they don't really understand science very well.

Yes, that's right: I strongly dislike TED. I think it's style over substance. Despite saying TED talks have "a somewhat vaporous tone," this (gated, unfortunately) New Yorker article by Nathan Heller about the history of TED is mostly laudatory:
Why speak rigorously to an audience of hundreds when you can ham it up a bit and spread the fruits of your research to millions?

Wow, that is aggravating. I much prefer the argument of Evgeny Morozov in The New Republic:
But surely, "modern attention spans" must be resisted, not celebrated.
Agreed. I don't think the world is simple. I don't think many useful answers are that simple. Conservatives like to mock the number of pages in Dodd-Frank, or they like to mock the fact that salmon are regulated by a different agency when in freshwater than when in saltwater, and we all know what crap John Kerry got when he said he voted for it before he voted against it, but guess what? The world is complicated. Policy, in order to be good, is most likely going to be complicated. More complicated than a TED talk.

(The salmon and Dodd-Frank bits are from this good Yglesias blog post.)


3 comments:

  1. for what it's worth: http://priceonomics.com/is-this-why-ted-talks-seem-so-convincing/

    ReplyDelete
  2. Interesting, but N=42.

    ReplyDelete
  3. yeah, i can't view the original paper but i guessed the sample size was small. still sort of interesting. i'm going running now. ugh, i sound like you.

    ReplyDelete