Two Thoughts, Interconnected


Aaron David

A fourth generation Californian, befuddled.

Related Post Roulette

7 Responses

  1. Avatar Tod Kelly says:

    I agree that the Dems are not making it easy on themselves.

    Still, using polling data to call a race before we know who’s going to be running seems a bit click-baity.Report

  2. Avatar Burt Likko says:


    If whatever Matt Y wrote today is still true then, I’ll deal with it then.

    Until then, Vox away, my friends. Vox yourselves out.Report

  3. Avatar Michael Drew says:

    If Hillary actually manages to lose, there’s going to be a whole other kind of fallout the party will work through before even getting to what Yglesias is talking about, though that will come too since that would constitute a bottoming out. (Of course, this eventuality would actually bring about confronting the Yglesias Problems sooner than if she wins.)

    But I’ll say this. I think it’s a little bit out of perspective to say that all a Hillary win would do would unproductively delay a necessary reckoning with the Yglesias Problems. Rather, a Hillary win would confirm that the situation is more in balance than the way Yglesias presents his case. Yes, the Dems would still need to work on their strength in all down-ballot races. But it’s actually a very major thing if the situation is that they keep on winning presidential elections, but for the moment are not doing well for other offices. To the extent there’s a trade-off there it may be one that they would actually take. Regardless, it’s a very different situation than if they’re losing everything. Winning a bunch of presidential elections in a row (all but one by popular vote in the century thus far) is very different from bottoming out. It doesn’t mean you don’t work on what the problems are down ballot. But it’s a far cry from bottoming out, and it’s pretty far from being in deep trouble (without the “at the state and local level” qualifier attached, where it would indeed be true), too.Report

  4. Avatar Chip Daniels says:

    My perspective as a California resident my be clouding my judgment a bit since the Dens hold every statewide office, and have since 2010.

    Having said that, I do think MY has a point that Dens fixate on the Presidency more than the state and c local races.

    As for the second article, it seems a lot of woo. It references patterns to achieve a certainty that seems implausible. How does a layperson know that they haven’t missed or misapplied a variable? We don’t so we really have to accept it on faith. Further it doesn’t offer any convincing theory why the public would prefer the GOP.
    It reminds me of that trick in statistics about the odds of a coin flip. A flipped coin will always have a 50% chance. Even if I just flipped heads 20 times in a row? Yes because the coin doesn’t remember.
    The election will not be between Historical GOP and Historical Dem. It will be between real candidates with all the complexities that come with it.Report

  5. Avatar Doctor Jay says:

    Kevin Drum trots out a model every presidential election that uses a similar principle – voters tend to get bored with the party that has the presidency. But that’s just one factor – the economy is another. The polls of the incumbent are not part of this model. So, I’m feeling very cautious about predicting two terms for Hillary, but I have no problem predicting that she will win the presidency.

    (We haven’t had one party hold the presidency for four terms since Roosevelt/Truman did it for five. It could happen, but I’m cautious.)

    And let me remind you that at this point in 2011, everyone was thinking that Obama was going to lose. Now some of that may be people trying to motivate their volunteers and donors, but not all of it was.

    I don’t think the polls at this point mean all that much.Report

  6. Avatar trizzlor says:

    It’s really cool that Reuters is running this kind of model and explaining the details to it’s readership. Contra Tod above, I think the idea that you can model election outcomes quite accurately *without* knowing the candidates is really important and something that gets lost in crossing from political science to political media.

    One nit I have to pick is that that use of international results (which make up the bulk of the training set) is sorta swept under the rug. I would love to see how well a model trained on non-US data predicted US elections.Report