Tuesday, December 6, 2011

Ignore Those Polls!

Gallup today started running a daily national GOP nomination tracking poll. Ignore that poll!

Yes, I know, you're going to peek at it, and I'm sure I will too. But we know how this works. People in California, and Texas, and New York, and everywhere else are going to be heavily influenced by the results of the early primaries and caucuses, and by the way the press interprets those events (which in turn is partially influenced by what elite GOP party actors say, but is also influenced by various press biases). Meanwhile, the national polls have little effect on what happens right now.

So if you ignore them, you don't miss their predictive power (because it's so small) and you don't miss their current importance (because it's also tiny).

What it's certainly time to do, however, at this point is to start watching the Iowa and New Hampshire polls. That Newt is up in Iowa at this point doesn't mean he'll necessarily win -- his lead isn't that large, and his support is soft -- but it certainly shapes the race for now, and the outcome in Iowa matters for New Hampshire, and the two of them matter for the next steps.

Everyone should just remind themselves that candidates who don't do well in the early states just completely collapse nationally -- see, for example, Rudy Giuliani 2008. And candidates who are at zilch nationally and then score in Iowa and/or New Hampshire can wind up with great national numbers almost overnight. Especially when, as is the case this year, everyone's support (except perhaps Ron Paul's) is incredibly soft.

17 comments:

  1. You're right but the reverse phenomenon is true and no one reports on it. Yes, Iowa and New Hampshire affect subsequent national polls. But national polls before Iowa and New Hampshire affect them. This is truer for presidential politics in the age of the internet. Gingrich leading national polls is the equivalent of what winning Iowa used to be (so I would disagree on ignoring them but I do agree on ranking them less important than Iowa, NH, SC, FL polls). Whether or not Newt can survive the next four weeks without blowing his lead in Iowa is a very open question.

    ReplyDelete
  2. Just to be snarky for a moment, does Jonathan's advice to "dismiss the polls" also extends to ignoring his oh so frequent posts on said poll? ;-)

    ReplyDelete
  3. Greg,

    Yes! OK, I have blogged on the national polls for finding the anti-Romney vote. I don't remember doing to much, if anything, on the head-to-head national polling, other than to say that it doesn't mean much.

    Anon,

    Fair point.

    ReplyDelete
  4. To run with Anon's point:

    I don't think one should totally ignore national polls, precisely because the media don't. Because they're like winning IA used to be, they also annoint frontrunners....which then drives a "how the mighty have fallen" story hook.

    When the results in Iowa read X then Y then a threeway tie between Z, A and Q, those results are going to be interpreted in light of national polls, because that's the water reporters swim in. If X is up in the national polls, the results will "reinforce the conventional wisdom that the election is X's to lose." If Y was ahead, the stories will say "Y falters in IA." If Z was ahead (even if not currently, but was recently), the news will show 600 clips of Z yelling "Yargh!"

    So, I would say, don't ignore these polls. Just don't look at them as evidence of who's going to win or lose. Rather, treat them as the backdrop of the stories that you're going to read.

    In more practical terms, if Gingrich wins IA, you'll see some stories saying "it's ABR" and some saying "two person race: G vs R." If Romney wins, my guess is that the only stories will be "R vs #2." Rather than use prose, I'll just summarize below (first letter is who wins IA/2nd letter is who wins NH).

    R/R: it's all over/R vs #2
    G/R: G vs R
    GPR (menaing Paul gets 2nd in IA over Romney)/___: (look at the polls in IA...it's possible) Gingrich, ABR, and look at Paul!
    R/G: (very unlikely) G vs R
    G/G: people just don't like Romney

    ReplyDelete
  5. Of course, R primary voters will be glued to those national polls, as they give indication of a candidate's status around the country, which individual state primary voters make use of.

    For example, Gingich in that poll has Romney doubled up with conservatives, and nearly tripled up with the Tea Party supporters, while being tied with moderate/liberals and non Tea Party supporters. Thus, those national results can leverage votes for Gingrich in individual states, when people see those breakouts.

    Similarly, Gingrich is ahead of Romney in all regions, and is only close in the East (like that matters?). Gingrich also dominates all age groups. That too might give folks pause about casting a vote for Romney, if he's well behind in multiple demographics like that

    So again, the polling is useful, it just requires careful filtering and usage. Primary voters are more likely to apply filters, and poke and squeeze candidates.

    ReplyDelete
  6. I've never bought into the whole idea of "tracking" polls anyway. I know campaign consultants like them, but they like them for the same reason horse race handicappers like workout information on first time starters-- because that's all the information they have.

    Just about every tracking poll is based on a tiny sample and the daily ups and the downs are always within the margin of error anyway. All you are doing is following statistical noise.

    ReplyDelete
  7. Aren't these polls just a self-fulfilling prophecy? It seems to me that if the polls say that one candidate is doing better, then they do better in the next poll.

    ReplyDelete
  8. If I understand correctly, the reason we should ignore polls at this juncture is because they elevate candidates who may not have the chops to win the nomination. Gingrich is the poll-friendly flavor-of-the-month, which is - I guess - as meaningless as the previous boomlets of Cain or Perry or...but

    Was that really a flaw in those polls? Saying that earlier pro-Cain or pro-Perry polls were inherently flawed is tantamount to saying that those candidates had no chance to win, which ignores the fact that each of them made historic laughingstocks of themselves around the time their boomlets burst.

    In other words, polls are being blamed here for a more fundamental flaw: the candidate himself sucked, and when folks discovered the epic candidate weaknesses, the polls followed suit.

    To me, polls showing Gingrich soaring in early states matter a lot, since they tell us that, if Gingrich's baggage is not a show-stopper, he's got the inside track to the nomination.

    Gingrich's baggage may well be a deal-breaker. But saying the polls are irrelevant is a question-begging way of asserting that Gingrich's baggage is in fact a deal-breaker without bothering to clarify specifically why.

    ReplyDelete
  9. The other possible explanation is that the invisible primary is such inside baseball that, even at this late hour, we the common people have no idea who the kingmakers prefer.

    Perhaps the insiders are even now madly scheming to find a road for Buddy Roemer to the WH? If so they best tell us ambulatory hamburgers soon, cause shortly we'll be headed to the voting booth, and none of us are currently for Buddy.

    In other words, whatever the mechanism of transferring the winner of the invisible primary to the visible one, that must be taking place either now or very shortly, with the success of the transfer reflected by polling.

    Gingrich may well die on the vine. But its hard to believe the insiders haven't thrown him the vine, cause, if not, to whom - and when - were they planning to do so?

    ReplyDelete
  10. One final thought: according to the Gallup poll, we're about 2 weeks into the Gingrich boomlet (while 4 weeks shy of the Iowa caucus). It would seem to be getting awfully late for the insiders to squelch the riff-raff's Gingrich fascination, assuming the insiders are so motivated.

    At the bottom is a link from a certain famous conservative radio personality who did a flip-flop on Gingrich and endorsed him today. A couple weeks into the boomlet. 4 weeks short of the first caucus.

    Of course, maybe the radio personality became a billionaire by some other means than regularly singing from the insider Republican choirbook. Maybe he got all that money cause of his good looks and charm.

    http://www.cbsnews.com/8301-503544_162-57337610-503544/rush-limbaugh-signals-support-for-newt-gingrich/

    ReplyDelete
  11. CSH,

    Well, yes, it's been two weeks, and perhaps only a week in which he was really opening a gap. And the point isn't to influence who wins Iowa; it's to influence who wins the nomination. Lots of candidates have won in Iowa or NH and then faded.

    Rush counts, although it's not always clear exactly how or how much he counts, and he's not exactly out-and-out endorsing Newt, at least not yet.

    Anyway, the point is I don't think it's really late for Newt to fade. Four weeks is a long time, and there's a much longer time until Supertuesday.

    Anyway, to your previous point...the reason we should ignore national polls is that they do a poor job of predicting how people will vote.

    But to everyone else: yes, again, it is true that national polls can affect people's behavior even if they are just smoke and mirrors.

    ReplyDelete
  12. Jonathan - thanks for the reply. This may be nitpicking, but it seems to me there's a difference between "fundamentally irrelevant" and "superceded by later information" as reasons a poll might not matter. We must be talking about the latter in the case of Gingrich's recent success, because if his boomlet is fundamentally irrelevant at this late hour, its hard to figure how the insiders communicate their preference to us outsiders.

    To put it in a more familiar context, if the Giants open a five game lead in May, that is unimportant in the sense that many things will happen to the team in the remaining four months that may override that lead. But we shouldn't ignore that result, since, if all else plays out equally, the division is now the Giants to lose.

    It's hard to imagine how the nomination is not, as of now, Gingrich's to lose. Its hard to see how this is not the equivalent of the 5-game lead in May. One counterargument might be that Gingrich will surely lose that lead because of his epic baggage.

    OTOH, all of that baggage is known, especially to the inside movers. As such, perhaps the more interesting argument is just how much the insiders must hate Romney, Perry and the rest of the circus. If Gingrich got here in spite of his epic baggage, he must be on fairly solid ground indeed.

    ReplyDelete
  13. CSH,

    A couple of things. One is that Iowa polls are not irrelevant at this point...they aren't perfect predictors, but they're useful. So to the extent the analogy holds, it holds with early state polls. National polls don't add (much) to that.

    But second -- the analogy doesn't really hold. Winning games in April may or may not predict what will happen in the rest of the season, but whether they do or not they really count as wins. Good polls right now just don't count in the same way.

    The other thing? Everyone is massively overestimating how well known Newt's baggage is to voters, and even to party actors. If you're a 35 year old Tea Party activist who wasn't really involved in politics until 2009, there's no reason you would know about the ethics violations or even the issue deviations.

    ReplyDelete
  14. If you're a 35 year old Tea Party activist who wasn't really involved in politics until 2009, there's no reason you would know about the ethics violations or even the issue deviations.

    Really? I'm almost 35, and I have a pretty vivid recollection of Newt's career in the '90s, when I wasn't exactly "involved" in politics.

    Granted, there are 20-somethings with no direct memory of Newt when he was in power, but I don't think they constitute the bulk of the Tea Party.

    ReplyDelete
  15. Jonathan,

    Thanks again for the comment. I take your point about the difference between an early poll and a spring baseball game; we could ask Rudy Giuliani how valuable those early poll wins turned out to be (someday politicians will hopefully name this the "Giuliani principle" as a consolation prize for that guy).

    So wrt the primary system where insiders feed information to the media which feeds information to the voters who feed their votes back to the eventual nominee, its actually a complex feedback loop, with some information flowing the other way up the chain, particularly out of the voting in Iowa, NH and SC.

    So Gingrich is doing much much better in NH today than a month ago, but he's still trailing Romney by quite a bit. A double-digit loss there would make him look a lot different than his double-digit gains the last month.

    Actually, one way politics differ from sports is in violating the sporting axiom that "each loss only counts for one". You know, the proverbial WS interview with Ron Washington after Pujols hits three long balls and blows out the Rangers by 13 runs, and Washington says "It still only counts for one game".

    In politics, the same doesn't exactly hold. Rudy Giuliani throws a shoe at the tv every time a sports figure makes such a comment.

    ReplyDelete
  16. Oooh.. an irresistible baseball analogy.

    Let me try and rough this out..

    Invisible Primary = Spring Training
    Iowa & New Hampshire = April, early season games.
    S. Carolina & Nevada = May, the pennant race takes form.
    Florida = 4th of July, the contenders are few.
    Super Tuesday = Labor Day, almost impossible to catch the leader now.

    And Clinton/Obama aside, I think you have to go back to Mondale/Hart in '80 or Ford/Reagan in '76 where the eventual nominee isn't clear by Super Tuesday/the first 6 contests or so.

    ReplyDelete

Who links to my website?