Political Polls In The Final Days Are No More Accurate Than Any Other Polls

Nov 3, 2014

FiveThirtyEight's polling model predicts a good night for Jeanne Shaheen.
Credit fivethirtyeight.com

New polls out over the past few days show all four of New Hampshire's major races in the state to be too close to call. 

That might prompt us to believe that anything could happen tomorrow, but as poll watchers will tell you, any single poll is just that: a single poll.

NHPR's Brady Carlson spoke with Harry Enten, a senior political writer with FiveThirtyEight -- the politics blog that introduced many politcal watchers to predictive elections models -- about just that. 

Carlson: First Harry, looking at the marquee national race in New Hampshire, Jeanne Shaheen versus Scott Brown, how much did the polls put out since Friday sway the candidates’ odds?

Enten: So Jeanne Shaheen has been a favorite throughout this entire campaign from the day it started until today. Her odds have shifted a little bit downward, so perhaps she had closer to a ninety percent chance of winning when the weekend started, and today it's closer to eighty percent but either way she is the odds on favorite to win.

Carlson: So win, not necessarily by a large margin, but to come out on top.

Enten: Right, look, the name of the game is to win. At this point we're saying she's going to win by about 2 points. That's not a large margin, it's certainly smaller than it was at the beginning of the campaign, but the fact of the matter is it doesn't really matter how much you win by you don't get an extra pretzel or cookie or anything for winning by larger margins.

Carlson: More generally, then, since every poll is a snapshot of a small chunk of the populace, how much should one read into these last minute polls compared to other polls? Don’t they mean more than the polling done before simply by virtue of coming out at essentially the moment that voters are either decided or not?

Enten: Well I wouldn't take too much into a last second poll as compared to a poll taken a week before, that's a huge difference, especially in a general election where most of the minds have been made up a long time ago. However, if you're taking a last second poll now as compared to a month ago or two months ago, then yeah those last second polls are certainly going to be more accurate.

Carlson: So to what degree should one be skeptical when a campaign declares that any given poll shows the candidate has momentum?

Enten: I'm always skeptical of the word momentum, whether in baseball or in politics. 

Carlson: Your model uses something called the “state fundamentals,” to help predict elections. What are those and what do they look like in New Hampshire?

Enten: The state fundamentals rely on a number of factors, number one is the partisan lean in presidential elections,. Two which is a big thing which takes into account the individual candidates, you're looking at the how much fundraising they've done. You know candidates who raise a lot more money are generally thought to be stronger candidates. 

And in New Hampshire what those numbers are suggesting are that number one New Hampshire, although it's a purple state has leaned a little bit more in presidential elections than it has red: obviously John Kerry won in '04 and Obama won both in '08 and '12. Also what you're seeing with the fundraising, there hasn't really been a large difference necessarily between the two, or not large enough to have much of an impact. So the fundamentals are slightly favoring Jeanne Shaheen. 

Which is much different than say North Carolina where the polling average looks very similar to New Hampshire, but the fundamentals there favor the Republican candidate, which is why fivethirtyeight thinks there's a slightly better chance of an upset in that state.

Carlson: The secretary of state in New Hampshire, Bill Gardner, who uses previous elections to predict voter turnout, said he believes this year could break records for turnout in a mid-term, sitting at just shy of 54 percent. Who does a number like that help and who does it hurt?

Enten: You know it really does depend on the state and the voters who are turning out. I remember in 2004, for instance in Ohio, Democrats were all excited about higher turnout. And there was, and it turned out Bush and his team were actually turning out more Evangelicals than ever and he was getting them to vote for him so higher turnout actually benefited the Republicans. Versus say in like a Southern state like North Carolina where African Americans might fall off during a mid-term and not vote as much as during a presidential year, higher turnout might benefit the Democrats.

But in New Hampshire where there isn't a lot of racial disparity between the voting blocks because New Hampshire obviously has a high percentage of whites compared to say the nation as a whole, I'm not really sure that higher vote turnout will necessarily help either candidate... or at least we don't know how it would at least not yet.

Carlson: And as candidates love to say, the poll that matters most is the one on Election Day itself. Harry thanks so much

Enten: Thank you.