People Can Predict Elections (Even When Polls Can’t)

New research from David Rothschild and Justin Wolfers:

Most pollsters base their election projections off questions of voter intentions, which ask “If the election were held today, who would you vote for?” By contrast, we probe the value of questions probing voters’ expectations, which typically ask: “Regardless of who you plan to vote for, who do you think will win the upcoming election?” We demonstrate that polls of voter expectations consistently yield more accurate forecasts than polls of voter intentions. A small-scale structural model reveals that this is because we are polling from a broader information set, and voters respond as if they had polled twenty of their friends. This model also provides a rational interpretation for why respondents’ forecasts are correlated with their expectations. We also show that we can use expectations polls to extract accurate election forecasts even from extremely skewed samples.

In an Oct. 26-28 YouGov poll, 45% said Obama would probably or definitely win, 18% said Obama and Romney were equally likely to win, and 29% said Romney would win (8% did not know).  In Gallup’s most recent poll, Obama’s advantage is 54%-34% (11% did not know).  Gallup notes that its own question accurately predicted the winner of the popular vote in the 1996-2008 presidential elections.  Wolfers is working on an estimate from 2012, which I’ll report here when it’s ready.

6 Responses to People Can Predict Elections (Even When Polls Can’t)

  1. Andrew Gelman November 1, 2012 at 2:39 pm #

    As I told Justin when I heard about this: it isn’t surprising to me that people can read the newspaper and find out who is predicted to win an election.

  2. Joel W. November 1, 2012 at 2:54 pm #

    But Andrew, that’s clearly not what’s happening. R partisans report thinking Romney is going to win for the most part, and vice-verse for Dems, though more Rs report thinking Obama will win. However, aren’t most newspapers reporting “an extremely close race” right now?

    Moreover, they were definitely reporting that the race was a tossup in the summer, when polls of this nature still gave Obama a large advantage. I tend to think that it is at least somewhat instructive as to the question of how tight likely voter screens should be.

  3. Andrew Gelman November 1, 2012 at 2:56 pm #

    P.S. I think the implied statement in the above headline, “polls can’t predict elections,” is misleading. First, a poll is a snapshot, not a forecast. Second, “people” predict elections given available information, which includes the polls reported in the newspapers as well as easily-available assessments by pundits, assessments that are, in the current TV environment, at least as accessible as the candidates’ actual policy positions. Third, polls can be used to make forecasts, if they are appropriately calibrated (as is shown, for example, in the work of Bafumi, Erikson, and Wlezien).

    You don’t have to survey 1500 people to find out who the experts think is going to win the election.

    I like Rothschild and Wolfers’s idea of asking people how they think their friends and neighbors will vote. I think it would be much more interesting to go for that directly and forget the headline-grabbing but silly “predicting elections” angle.

    • David Pennock November 5, 2012 at 11:11 am #

      I’m not sure Bafumi, Erikson, and Wlezien’s method would generalize easily for primaries or races with true third-party candidates. Rothschild & Wolfers’s method should easily generalize. A prediction market is not much different than Rothschild & Wolfers’s method, though the data suggests they do ever so slightly better still.

  4. scott November 2, 2012 at 8:51 am #

    Vote flipping can be seen here:

    Very small precincts don’t have any votes “flipped”. The percentage of votes that are flipped is small (such as .01%) for small precincts and large (such as 5%) for large precincts with a *gradual* change in percentage “flipped” dependent on the precinct voter population. A recount will check a random number of precincts and a smaller precinct is *more* likely to be audited because there are more of them which is why the perpetrators flip fewer votes in small precincts.

    • Dan Ferguson November 2, 2012 at 4:52 pm #

      Holy Cow! have you shared this with the Obama Campaign or the NYT?
      That is a good but troubling analysis.