Monday, November 7, 2011

Cranky Monday Blogging 1


The AP has an awful "one year out" scene-setter for the presidential election. "It's virtually certain that the campaign will be a close, grinding affair." Really, AP? Guess what: it's not in fact virtually certain that it will be close (or, for that matter, grinding, whatever that's suppose to mean. We still don't know what the economy will do over the next nine months; we still don't know the GOP nominee. Seth Masket graphed some results from Nate Silver's fun election predictor, and gets plausible results ranging from close to 100% chance of an Obama re-election to well under 40%.  And that's without factoring in anything else that could affect matter how true it is that economic and other fundamentals matter a lot, there's always the chance that a significant scandal or particularly poor out-party candidate or world events could push things significantly.

Much of the rest of the AP story is cliche and meaningless junk. This election will be "more partisan" than 2008? Huh? And what exactly does this mean from Chris Lehane:
It's going to be extremely different, with much more hand-to-hand combat, from one foxhole to another, targeted to key states.
Obama and John McCain didn't target key states in 2008? And what exactly does "hand-to-hand combat" mean (and, by the way, you don't actually have hand-to-hand in foxholes, do you? I suppose you can).

Also: the AP predicts that Obama will run a negative campaign after not doing so in 2008, but as John Sides keeps pointing out Obama was quite negative in 2008.

Hey, they got a good quote from Dan Schnur, so that's something. And I do think that the AP was correct that the general election will be one year from now. Totally nailed that part.


  1. It's going to be extremely different, with much more hand-to-hand combat, from one foxhole to another, targeted to key states.


    Obama and John McCain didn't target key states in 2008? And what exactly does "hand-to-hand combat" mean (and, by the way, you don't actually have hand-to-hand in foxholes, do you? I suppose you can).

    2012 will be a completely different election than 2008, so the guy's correct.

    Obama didn't fight hand to hand in 2008. He made appearance, and even deigned to come among us, permitting worship.

    And once he waved off public financing (and I was amused at how he breezed that aside, after previous ironclad declarations to the contrary, and at how easily the worshipful accepted that decision), he was able to outspend McCain in states dramatically. He owned the airwaves. Add in the crash, and McCain's foolish wanderings and statements about it... and the election was mostly an afterthought.

    2012 will be quite different, as the electorate knows Obama now. There was no vetting process in 2008, but there's been one since. Bailouts, Porkulus, Cap & Tax and ObamaCare will be at issue, as they were in last November's mid-term. You can throw in recent scandals, as these continue to wax.

    And then there's the 43 cents of every federal dollar spent that he's borrowing, and that he wants to continue on that same corrosive path, as his submitted budgets demonstrate.

    Incumbents face different environments than empty vessels, which folks have filled with their favorite filling. Grinning Georgia peanut farmers and cold blooded community organizers can fool much of the people some of the time, but after that...

    The electorate knows this guy now. From foxhole to foxhole and state to state, those issues will be fought hand to hand, as applicable and relevant from foxhole to foxhole and state to state. The greek columns won't work this time.

    Yes, Obama ran plenty of negative advertising in 2008, and 2012 will be no different. The problem for him is, the hopey changey nonsense won't be there to counterbalance that negative campaiging. It can't be, as the electorate knows better now.

    And yet, without the hopey changey nonsense, he can't attract fringe voters and drive the massive turnout he desperately needs to win reelection. Pure negative campaigning drives them off, in fact. His model is conflicted, then.

    He's facing a terrible dilemma here. His campaign shysters will want to spend out all that loose cash he'll have, to line their own pockets, but I'd recommend he keep it in his own pocket, or find other things to do with it, because assaulting candidate Perry/Romney/Cain may well drive down his own turnout model, and reinforce their attacks on him.

    Yes, the lefties will be ecstatic over every one of those attacks, but the people who matter won't have that same reaction...

  2. Shorter AP: We really, really hope this next election is more interesting than the last one.

  3. "Pure negative campaigning drives them off, in fact."

    Anon #1:

    What are the data that negative advertising depresses turnout? Here data are summarized that it doesn't:

  4. What are the data that negative advertising depresses turnout? Here data are summarized that it doesn't:


    No, that meta analysis data linked in Silver's article doesn't support your proposition. Silver's article makes that claim as well, but Silver's often quite sloppy, as we know.

    Rather than reading Silver's tale of what the data says, go into the actual meta data he linked to: APPENDIX TABLE 1A Description of Negative Campaign Studies Used in the Meta-Analysis, Including Old (from Lau et al. 1999), Updated, and New Findings. There, you'll find studies reporting such as this:

    Negative ads depressed intended
    turnout, d = -.10.

    Negative ads decreased intended
    vote for the attacker during primary
    elections, d = -.14

    States with more negative Senate
    election campaigns had lower
    turnout, d = -1.27.

    Recall of negative ad associated
    with lower probability of voting
    compared to recall of positive ad,
    d = -07.

    Negative message associated with
    trivially higher turnout in both
    Minnesota (d = .01) and Los
    Angeles (d = .004).

    And those are just a few cuts from the first few of the 111 studies listed in the table.

    So in general, I'd say the story isn't as clear cut as Mr. Silver fantasizes, not if you're actually reviewing data (As with most of the sabermetric geeks, Silver is generally more interested in twisting data around than properly reviewing and presenting it.).

    If Obama blowtorches the country with negs, as his campaign shysters will inevitably be motivated to have him do, you can bet that it's gonna depress the massive turnout he needs. Think the OWS crowd will respond well to that? Think that's what will motivate the 18-29 age demographic? I don't. They're the core of his hopey changey strategy, and he can't win absent that, and them.

  5. "No, that meta analysis data linked in Silver's article doesn't support your proposition."

    1. Yes, it does:

    "It follows that the research literature provides no general support for the hypothesis that negative political campaigning depresses voter turnout. If anything, negative campaigning more frequently appears to have a slight mobilizing effect."

    What you did, as you note, was pick out the results of a few studies included in the meta-analysis. The meta-analysis itself considers the effect sizes of all the studies. And, as the authors note, the statistical conclusion is quite clear.

    2. The article I linked to was by political scientist John Sides, not Nate Silver.

  6. Anon 8:58,

    But what then are we going to make of Anon's (2:46) claim that "Silver's often quite sloppy, as we know."???

  7. Yes, anon, but you're making the same mistake that the sloppy Silver made and makes... that the same conclusion is true in ALL cases. Not true, and the meta data proves that, as the examples I cited showed. Negative campaigning DOES indeed drive down turnout, in many cases, as I cited from the study. Not all, as the data tells us, but not not all, as Silver claimed.

    We are mostly inured to the negative campaigning these days, if it's stock schlock. That's not gonna move the needle on turnout, one way or another, as the meta data tells us.

    However, if Herman Cain reaches under the skirts of his employees on a regular basis, proven so, and winds up on a ballot, I guarantee that that will be campaigned negatively by his opposition, and it will drive up the vote count in that election.

    On the other hand, if Obama blowtorches the country with negative advertising in 2012, I can with just as much certainty guarantee that his turnout will drop markedly from 2008 levels. He's got soft support, dependent on a flighty voter demographic. He'll wind up as one of those cites I listed for you, that the meta study demonstrates exists in the electoral pool... with depressed turnout. We know those cases exist, right? How do we know? The data tells us, and unlike Silver, we have to pay attention to the data.

    Try thinking outside blacks and whites. That's Silver's problem as well... twisting data to confirm his bias, rather than merely presenting it.

  8. 1. Again, it's political scientist John Sides who wrote that blog post, not Nate Silver. Nate Silver has nothing to do with this. (And, for what it's worth, that means you have Sides, Bernstein, and the political scientist who authored the study all drawing the same conclusion.)

    2. Significant correlations and effect sizes from single studies don't mean very much. They could be false positives, reflect poor research methodology, reflect biases (conscious or not) on the parts of the investigators, etc. The meta-analysis helps us to get beyond that statistically. And again, the conclusion the authors of the meta-analysis reach is the opposite of what you hypothesize.

    3. Is it possible that those studies were not any of the above and that you, by plucking data out, are calling our attention to "existence proofs" of negative campaigning driving down turnout? Yes, it's possible. However, we don't know if that's the case. Furthermore, even assuming the former, do they support your contention that Obama's negative campaigning would drive down turnout for the reasons you describe? You would have to systematically analyze the data, because we can't prima facie assume that.

    To argue that the meta-analysis itself statistically demonstrates that in certain cases negative campaigning does drive down turnout and that Obama using lots of negative campaigning would depress turnout are both arguments from the vacuum: interesting hypotheses with no empirical support.

    In the meta-analysis, each of those studies is treated as a separate data point. You can't pull out single data points and derive conclusions from them.

    4. To put it another way: the meta-analysis statistical findings tell us that in Given Case X, if we hear negative campaigning is being used, we should assume that it will have no effect on turn out or perhaps a slight positive effect UNLESS there is compelling empirical evidence to tell us otherwise. Hypothetical suggestions about how negative campaigning would specifically impact Obama's campaign are not compelling empirical data.

  9. anon, doesn't matter who it is that's ignoring the data, you, Silver or whoever. You're ignoring the data, and I'm just pointing out that fact.

    The conclusion that the authors of that meta analysis give is that negative campaigning depresses turnout in some cases, and in others it doesn't, and overall it might not have an effect, in the mean. But your statement, and Silver's, is that it never depresses turnout. You're ignoring the data, in other words.

    I'm not pulling out comprehensive conclusions from those data sets. You are. I'm only pointing out to you that the comprehensive conclusion that you're drawing, that negative campaigning does not haves effect on turnout, is not valid, and case by case, can be proven invalid, as the meta data tells us.

    You're correct, I'm unable to document that Obama's blowtorching the country will depress his turnout, although it should be intuitively obvious that it would, given his flighty voter demographics, and need for massive turnout in that demographic, as per 2008. That needed demographic isn't gonna be inspired to vote by a blowtorching... and only the hopey changey stuff worked last time. How do we know this? We go back and check turnout demographics, and spot out the 2008 anomaly.

    I think you're confusing total turnout for all candidates, as per that study, with specific turnout for one particular campaign. It'd also be intuitively obvious that total turnout in a group of studied elections might not change much due to negative campaigning, regardless of negative campaigning. However, the composition and spread of that turnout might drastically change... voter intensity and all that. That's what Obama's got to fear. He'll have more money than God to spend, but he'd best be careful about how he spends it.

    I think we're going to get to test your theory, as Obama's handlers will want to spend that money, and he doesn't seem the type to resist. If he wins it still, in this environment, then I'd be willing to concede the point. If he loses, let's compare turnout demographics and see what we find.

  10. Anon, you're refusal to acknowledge that Silver never made any statement on this is disconcerting because it suggests that you are trolling. Anyway, I continue to think you're interpreting the meta-analysis incorrectly for the reasons I described above.

  11. Also, one last point:

    "I think we're going to get to test your theory, as Obama's handlers will want to spend that money, and he doesn't seem the type to resist. If he wins it still, in this environment, then I'd be willing to concede the point."

    I wouldn't expect you to concede the point for that reason. As I hope I've made clear, I almost never consider an n=1 example to be sufficient scientific evidence from which to draw conclusions.

  12. Yes, Silver said that negs didn't work. Here's the quote from Silver's link:

    "The most comprehensive meta-analysis of research into negative advertising found no conclusive evidence that they work"

    And then we go to the link he provided to the meta analysis, and we find that in fact, negs DO work... in many, many cases. That is, Silver's hacking around with the data, as he and the other sabermetric geeks are prone to do.

    Sorry, but the data is the data, and the man's words are his words. You seem to be ignoring both, and that's the root of this. That's what I'm pointing out. You're ignoring plain data and reality, laid out right in front of you. I can't put this any plainer than that.

    I'm not "interpreting" that meta data, not at all. I'm pointing it out to you. You're the one who's ignoring it.

    And you're correct, one election wouldn't conclusively prove anything, either way. But I'd hope that at least, you'd intuitively understand that Obama's playing with fire, when it comes to the voter demographic he desperately needs to win reelection. We know for a fact that negative turnout has been proven to depress turnout in many elections. How do we know? The data tells us. Now, the question is to isolate the manner in which it does so, when it does so, and identify the voter demographics involved. Mr. Obama would do well to pay careful attention to isolating those demographics, as they are near definite to include portions of the massive turnout he needs for reelection.

    And I think we're going to get a chance to test your theory. Obama just isn't temperamentally capable of avoiding that outcome. ;-)


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

Who links to my website?