Thursday, June 21, 2012

Read Stuff, You Should

Lot of good choices about a Happy Birthday to Berkeley Breathed, 55. Yes, kids, there was once a time when the comics in your daily newspaper were excellent.

And the good stuff:

1. Fed-watching from Sarah Binder.

2. A very helpful primer on executive privilege, by Andrew Rudalevige. Yes, I'm sending you to the Monkey Cage twice, and really you should just be reading everything at the Monkey Cage, but it seems best to list the individual items that I take particular note of. But yeah, really, just read it.

3.Matt Yglesias makes the case for more low-skill immigration.

4. And a good Nate Silver post about outliers on the occasion of a goofy reading from a normally reliable pollster (see also Mark Blumenthal) opens up a reader opportunity for a catch of the day: did any prominent Democratic or liberal blogs get all excited about yesterday's Bloomberg poll, with its 13 point Obama lead? I didn't notice any, but then again I certainly don't read 'em all. I'd also be impressed if anyone can find a Republican site gushing about the latest Gallup or Rasmussen numbers -- slight Romney lead -- while ignoring Bloomberg entirely. But mostly, I'm interested if any Obama-supporting writers played up the Bloomberg poll.


  1. Nearly everyone has dismissed the Bloomberg poll as an outlier, probably because that's what it is. There may have been a user diary or two about it over at Daily Kos, but I didn't see anything from the big players, either. Nevertheless, I'd rather see an outlier in Obama's favor than one in Romney's. I guess there is something to that.

  2. The coverage on liberal blogs of the poll has fallen into three camps that I've seen:

    1) It's an outlier, so ignore it.

    2) It's an outlier, but I'd rather it be in our favor.

    3) Let's talk about the other findings in the poll, like Romney's unpopularity.

  3. On Daily Kos Elections, the reaction was "huh. Has Selzer done any other national polls? Not too many. Eh, it's an outlier, but it'll balance out Rasmussen". The main "polling digest" post was titled "Is Obama Really Leading Romney by 13?" and started with "Short answer: Almost certainly not." (Paraphrases)

  4. There was this:

    But it's not a major blog.

  5. The Bloomberg "kerfuffle" is a fascinating window into the difference between Republican and Democratic approaches; perhaps suggesting why the Dems tend to do so much worse at spin wars than the Republicans.

    First, as Silver notes, the Bloomberg poll is almost certainly statistically valid (that is, its an appropriate measurement of the information it seeks). The question is whether its reliable (that is, does it get it's measurement right?)

    On the reliability question...who knows? The concept of margin of error is designed to reflect that 19/20 (or 99/100 or whatever) such measurements will produce reliable results; in a zeitgeist bombarded with polls, even a 99/100 confidence interval will produce errant results when there are 100's of polls.

    This fact seems to bother no one, including political scientists, until a poll like Bloomberg shows up, and we (particularly liberals) say ah, this must be the errant 1/100 valid polls. Why? Because it doesn't feel right. Do we know shit, though? Do we know how the Dream Act stuff plays in the populace at large? Last minute buyer's remorse at the impending doom of the ACA? Increased exposure to Mitt Romney's epic, even by politician standards, sliminess?

    We think we know. Why, though? 'Stuff' has happened since the last polling round; stuff that might matter, we think it doesn't, but do we know? There's a general principle in stats that the first observation of a new data set is the most likely mean value of that data set - we (particularly liberals) base our dismissal of Bloomberg on the idea that it doesn't represent a new data set.

    Maybe. Maybe not. The "maybe not" part arguably represented a golden spin opportunity, a gift the recipients of which wanted absolutely no part of. In similar circumstances, I doubt the Republicans would have been so ungrateful. Does explain some things.


Note: Only a member of this blog may post a comment.

Who links to my website?