Nate Silver: State Polls vs. National Polls
What State Polls Suggest About the National Popular Vote
By NATE SILVER
Mitt Romney and President Obama remain roughly tied in national polls, while state polls are suggestive of a lead for Mr. Obama in the Electoral College. Most people take this to mean that there is a fairly good chance of a split outcome between the Electoral College and the popular vote, as we had in 2000. But the story may not be so simple
For both the swing state polls and the national polls to be right, something else has to give to make the math work. If Mr. Obama is performing well in swing states, but is only tied in the popular vote nationally, that means he must be underperforming in noncompetitive states.
But polls of noncompetitive states don’t always cooperate with the story. Take the polls that were out on Tuesday.
Mr. Obama trailed by “only” eight points, for instance, in a poll of Georgia that was released on Tuesday. Those are somewhat worse results than Mr. Obama achieved in 2008, when he lost Georgia by five percentage points. But they’re only a little bit worse, whereas the national polls are suggestive of a larger decline for Mr. Obama in the popular vote.
Or take the poll of Texas, also out on Tuesday, which had Mr. Obama behind by 16 points there. He’s obviously no threat to win the state or come close to it, but that still represents only a 4-point decline for Mr. Obama from 2008, when he lost Texas by 12 points instead.
High-population red states like these, Texas and Georgia, are just the sort of places where Mr. Obama would need to lose a lot of ground in order to increase the likelihood of his winning the Electoral College while losing the popular vote.
Perhaps Mr. Obama is underperforming in deeply blue states rather than deeply red ones? Sometimes you’ll get numbers that check out with this assumption: Mr. Obama did get some mediocre polling in Oregon on Tuesday, for instance. But he also got a poll showing him ahead by 23 points in California. Another survey on Tuesday gave him a 31-point lead in Massachusetts.
Yes, I am deliberately cherry-picking a bit. But the discrepancy seems to hold if you look at the data in a more comprehensive way. Nor is it an unusual feature of the FiveThirtyEight model. Rather, pretty much every method for evaluating the election based on state polls seems to hint at a very slight popular vote lead for Mr. Obama, along with an Electoral College one.
snip
You can see that the various projections strongly agree with another, for the most part, in making “calls” about individual states. The only state where different sites show different candidates ahead right now is Florida, where Talking Points Memo gives Mr. Obama a nominal 0.2-percentage point lead while the others (including FiveThirtyEight) have Mr. Romney slightly up instead. There are also four states — New Hampshire, Iowa, Colorado and Virginia — in which some methods show an exactly tied race while others give Mr. Obama the lead.
Although I hope that this chart serves as a useful reference point — and as a reminder that other data-driven sites that look at the polls with the same philosophy that FiveThirtyEight applies are achieving largely the same results — I’m more interested in looking at this data in a macroscopic way.
Suppose, for example, that you take the consensus forecast in each state. (By “consensus” I just mean: the average of the different forecasts.) Then you weigh it based on what each state’s share of the overall turnout was in 2008, in order to produce an estimate of the national popular vote.
Do the math, and you’ll find that this implies that Mr. Obama leads nationally by 1.9 percentage points — by no means a safe advantage, but still a better result for him than what the national polls suggest.
What if turnout doesn’t look like it did in 2008? Instead, what if the share of the votes that each state contributed was the same as in 2004, a better Republican year?
That doesn’t help to break the discord between state and national polls, unfortunately. Mr. Obama would lead by two percentage points in the consensus forecast weighing the states by their 2004 turnout.
Or we can weigh the states by their turnout in 2010, a very good Republican year. But that doesn’t help, either: instead, Mr. Obama leads by 2.1 percentage points based on this method.
(In each of these examples, you’d get almost exactly the same outcome if you used the FiveThirtyEight forecast alone rather than the consensus. We’re on the high end and the low end of the consensus in different states for Mr. Romney and Mr. Obama, but it pretty much balances out over all.)
Whether the state polls or the national polls characterize the election correctly could well determine its outcome.
Mr. Obama’s lead in the Electoral College is modest, but also quite consistent across the different methods. The states in which every site has Mr. Obama leading make up 271 electoral votes — one more than the president needs to clinch victory. The states in which everyone has Mr. Romney ahead represent 206 electoral votes. That leaves five states, and 61 electoral votes, unaccounted for — but Mr. Obama would not need them if he prevails in the states where he is leading in the polls.
But perhaps national polls tell the right story of the race instead — meaning that the state polls systematically overrate Mr. Obama’s standing?
It’s certainly possible. (It keeps me up late at night.) If the polls in states like Ohio and Wisconsin are wrong, then FiveThirtyEight — and all of our competitors that build projections based on state polls — will not have a happy Nov. 6.
With that said, our decision to cast our lot mostly with the state polls is not arbitrary. In recent years, they’ve been a slightly more unbiased indicator of how the election will play out.
Bias, in a statistical sense, means missing consistently in one direction — for example, overrating the Republican’s performance across a number of different examples, or the Democrat’s. It is to be distinguished from the term accuracy, which refers to how close you come to the outcome in either direction. If our forecasts miss high on Mr. Obama’s vote share by 10 percentage points in Nevada, but miss low on it by 10 percentage points in Iowa, our forecasts won’t have been very accurate, but they also won’t have been biased since the misses were in opposite directions (they’ll just have been bad).
snip
We’re approaching the point where Mr. Romney may need the state polls to be systematically biased against him in order to win the Electoral College. And that certainly could turn out to be the case: if Mr. Romney wins the popular vote by more than about two percentage points, for example, he’ll be very likely to cobble together a winning electoral map, somehow and some way. (And he’ll be a virtual lock if the results are in line with Mr. Romney’s best national polls, like the Gallup survey, which put him four or five points ahead.)
But the historical evidence weighs in slightly more heavily on behalf of the state polls, in my view, when they seem to contradict the national ones. If the state polls are right, than Mr. Obama is not just the favorite in the Electoral College but probably also in the popular vote.
snip
http://fivethirtyeight.blogs.nytimes.com/
= new reply since forum marked as read
