Before Iowa, Hillary was beating Obama in NH by like 20 points, or at least double digits. After Iowa, Obama got this huge surge in the polls. You can see the time series here. It’s a mystery why the polls were so wrong. I think it comes down to:
First – Erikson, Panagopoulos, and Wlezien wrote a paper showing that the Gallup poll overestimates fluctuation in the electorate when using the likely voter screen early in the election. In a nutshell, what happens is this: because the Gallup poll (and most other polls) are interested in interviewing “likely voters” only, they ask a series of screening questions at the beginning of the poll to gauge the respondents’ interest in the election. They then have some formula to determine who is a “likely voter”, and they throw out the remainder of the results. This paper examined the results that were thrown out along with the poll and found that, when something is going wrong for a candidate, their supporters are less enthusiastic and therefore less likely to be considered “likely voters” during this screening process. As a result, many of the supporters of the “losing” candidate just aren’t counted in the poll, because pollsters think they’re not going to vote. This makes fluctuations in polling seem more dramatic than they actually are. ( ) This means Obama was never actually leading in NH, and all this talk about “something happened in the last 24 hours” is all a load of BS.
Second – survey weighting. Whenever a pollster does a survey, they need to make the poll representative of the voting electorate. (They) essentially guess what the demographic makeup of the electorate is going to be. Usually this is done on historical data and census data, but it’s always really hard in primaries because they’re not very consistent. So the pollster will first try to get this breakdown in who they actually talk to, and if they can’t, they’ll then “weight” the survey to simulate the expected breakdown. ( )The pollsters probably saw how weird the electorate was in Iowa (i.e. so many people turned out, and so many young people), that they probably tried to compensate by weighting young people a ton in the following polls to NH. Now, we know that young people support Obama disproportionately. If the pollsters overcompensated for young people, then Obama’s support was artificially strengthened in the polls.
MOMENTUM AND SOCIAL LEARNING IN PRESIDENTIAL PRIMARIES
Brian Knight and Nathan Schiff (NBER Working Paper 13637)
Do outcomes of primaries depend upon the sequencing of states? Do sequential, relative to simultaneous, systems lead to different outcomes in terms of the selection of candidates? In our view, the key distinction is that sequential, relative to simultaneous, elections provide late voters with an opportunity to learn about the desirability of the various candidates from the behavior of early voters. This opportunity for late voters to learn from early voting returns can in turn lead to momentum effects.
We develop and estimate a simple model of voter behavior under sequential elections. In the model, voters are uncertain about candidate quality, and voters in late states attempt to infer private information held by early voters from voting returns in early states. Candidates experience momentum effects when their performance in early states exceeds voter expectations. The magnitude of momentum effects depends upon prior beliefs about the quality of candidates held by voters, expectations about candidate performance, and the degree of variation in state-level preferences. Our empirical application focuses on the 2004 Democratic primary. We find that Kerry benefited substantially from surprising wins in early states and took votes away from Dean, who stumbled in early states after holding strong leads in polling data prior to the primary season.
The estimated model demonstrates that social learning is strongest in early states and that, by the end of the campaign, returns in other states are largely ignored by voters in the latest states. The voting weights implied by the estimated model demonstrate that early voters have up to 20 times the influence of late voters in the selection of candidates, demonstrating a significant departure from the ideal of “one person, one vote.”
Finally, we simulate the election under a number of counterfactual primary systems and show that the race would have been much tighter under a simultaneous system and that electoral outcomes are sensitive to the order of voting. While these results are specific to the 2004 primary, we feel that they are informative more generally in the debate over the design of electoral systems in the United States and elsewhere.