Thursday, June 21, 2012

This is deeply misleading

Arnold Kling writes:

"In a podcast with Russ Roberts, [James] Manzi says,

Is the number of stars in our galaxy odd or even? Well, there's a real answer to that question. If you have a bunch of people yelling odd and a bunch of people yelling even, one of those two groups is right. But unless one of them has access to knowledge that I don't think we have a species right now, we don't know. And that doesn't mean it is a theoretically technically unanswerable question, how many stars are in the galaxy, but we don't have the knowledge right now and we don't have the capacity of knowledge right now. And that's the way I feel about that debate.
He is referring to the debate about whether the stimulus created jobs. As you know, I think his book, Uncontrolled is excellent. The podcast is also recommended."
 
Whether the number of stars in our galaxy is odd or even is a question that we can't answer because we can only produce an estimate of this number that we can infer from what we observe. Estimates have errors and as long as we estimate these figures we can't say anything about them that you need zero error to say.
 
Similarly, we can't say whether the stimulus produced an odd or an even number of jobs. We just can't.
 
But we can provide an estimate, with error, of the number of stars in the galaxy and we can provide an estimate, with error, of the number of jobs created by the stimulus. Those are the scientifically accessible and the practically relevant questions.
 
This is really irresponsible blogging.

9 comments:

  1. "This is really irresponsible blogging."

    Which is why it's time to stop paying any attention to it whatsoever!

    ReplyDelete
  2. We can provide estimates, but for both questions we cannot provide anything more than an estimate of the error for our estimate...
    This is not to suggest that we shouldn't attempt to estimate those numbers, but for practical/policy purposes, they should be taken with a grain of salt. The trouble, in m eyes, is that these estimates are often presumed to be far more certain than is scientifically possible.

    ReplyDelete
    Replies
    1. Agree with the first half of your comment, but I don't think you have grounds to say we should take it with a grain of salt.

      Delete
    2. Maybe that language was too extreme, would you accept modest amount of skepticism?

      Delete
    3. Sure! I have a modest amount of skepticism when I decide that the leftovers are still good enough to eat. I'm certainly willing to extend it to multiplier estimates!

      Isn't this the heart and soul of statistical estimation, though?

      Delete
    4. Yes. Maybe that was implicit in your initial post (if so, my apologies), but it wasn't clear when I read it.

      Delete
  3. Is the definition of "jobs created" precise enough to allow this question to be answered, even with perfect information?

    ReplyDelete
  4. I think you completely miss Manzi's point. His point is that the number of jobs created cannot be estimated because of the problem of causal density. Any "point estimates" or "standard errors" are phony.

    I would recommend reading his book. You clearly were misdirected by the star-counting analogy.

    ReplyDelete
    Replies
    1. I completely agree with a lot of the summaries of the book and the obstacles to empirical policy work (I talk about these issues a lot on here, if you don't follow it). It's the leap to the "cannot be estimated" that concerns me.

      Delete

All anonymous comments will be deleted. Consistent pseudonyms are fine.