What's Behind the State-National Poll Divergence?

By Sean Trende - October 31, 2012

‹‹Previous Page |1 | 2 |

After adding the totals up, the results were plain: If the state polls are right, even assuming Romney performs as well as Bush did in the states without polling, Obama should lead by 1.18 points in the national vote. Given the high collective samples in both the state and national polling, this is almost certainly a statistically significant difference. It’s also a larger margin than all but one of the polls in the national RCP Average presently show.

But what if my assumptions about the states without polling are incorrect? To double-check this, I turned to Drew Linzer’s “Votamatic” model. It provides estimates for all 50 states. While some of these seem a bit off (I would bet $10,000 of Mitt Romney’s money that he will win Tennessee by more than 10 points), it still gives us a nice uniform data set. The result: When weighted by 2008 voting patterns, these data suggest that Romney should lose the popular vote by 2.5 points -- more than any national poll is presently showing.

Again, this makes sense given what we see in the state polling. If Romney loses big in California, New York and Illinois, and loses or is effectively tied in Florida, Pennsylvania, Ohio, and Michigan, we’ve accounted for seven of the 10 largest states, comprising about 40 percent of the U.S. population. The gains he would have to roll up in places like Texas, North Carolina, and Georgia to offset this simply aren’t possible, given the realities of minority voting patterns in those states.

As a further check, I used Bush ’04 vote totals in the states, rather than 2008, and the numbers didn’t change appreciably.

But what if there is a huge surge in Republican voting? Well, this would be felt across all states, at least somewhat, and should be turning the purple states red. But regardless, I took Linzer’s model, assumed a 20 percent increase in turnout in the red states (those projected to go for Romney by more than 8 percent), and assumed a 6 percent net increase in Romney’s vote share over the baseline projection. The result? Romney wins the popular vote by 0.7 points, still less than the current RCP Average.

You can poke holes in this model, to be sure, but I think the simplest explanation is that the state and national polls really are saying different things, at least for now. In other words, if you are calling for the state polls to be right, you are pretty much necessarily calling for the national polls to be wrong, and vice versa.

How do we resolve this? Which will be correct? My best answer is “I don’t know; it is a source of uncertainty in projecting the election.” I suspect one group of polls will converge upon the other in the next week, and we’ll get a better idea.

After all, there are several good arguments for favoring the state polling: (1) you have more polls -- a much larger collective “n”; (2) you compartmentalize sampling issues -- pollsters focused exclusively on Colorado, for example, seem less likely to overlook downscale Latinos than pollsters with a national focus; and (3) the state pollsters were better in 1996 and 2000, two years that the national pollsters missed (although the truly final national pollsters in 2000 got it right, suggesting that perhaps there was a late shift in the race).

But this is by no means a cut-and-dried case. Among national pollsters, you have a battle-tested group with a long track record performing national polls. Of the 14 pollsters producing national surveys in October, all but three were doing the same in 2004 (although AP used Ipsos as its pollster that year rather than GfK, and I believe a few others may have changed their data-collection companies). Of the 14 pollsters surveying Ohio in October, only four did so in 2004 (five if you count CNN/USAToday/Gallup and CNN/Opinion Research as the same poll).

Pollsters such as ABC/Washington Post, Gallup, Pew, Battleground, and NBC/WSJ are well-funded, well-staffed organizations. It’s not immediately obvious why the Gravises, Purple Strategies and Marists of the world should be trusted as much as them, let alone more. And since virtually none of the present state pollsters were around in 1996 or 2000 (except Rasmussen Reports, which had a terrible year in 2000 and has since overhauled its methodology), it’s even less clear why we should now defer to state poll performance based upon those years.

Finally, remember that in 2008, the national polls were pretty much spot-on; the state polls were off by a couple of points. Although they actually tended to favor McCain that year, it’s just further proof that the larger “n” isn’t a guarantee of greater accuracy.

For now, I think the best thing to do is wait, and to remember that there is probably more uncertainty in this election than partisans on either side would care to admit.

UPDATE: A few people have emailed to ask if we shouldn’t just average the national and state polls. Obviously, we at RCP love averages. But you have to be careful with them. Example: If you say the capital of the United States is Washington, D.C., and I say it is New York, the best course of action is not to compromise and call it Philadelphia. Similarly, we have enough national pollsters consistently saying one thing, and enough state pollsters saying another, that I’m fairly sure the “right” answer won’t be in the middle. It really is a “two men enter, one man leaves” situation. 

‹‹Previous Page |1 | 2 |

Sean Trende is senior elections analyst for RealClearPolitics. He is a co-author of the 2014 Almanac of American Politics and author of The Lost Majority. He can be reached at Follow him on Twitter @SeanTrende.

Mitt Romney for Mayor
Carl M. Cannon · November 16, 2014
A President Who Is Hearing Things
Richard Benedetto · November 12, 2014
Bret Stephens' Call for Robust U.S. Foreign Policy
Peter Berkowitz · November 16, 2014

Latest On Twitter