The Corner

Politics & Policy

When It Comes to Polls, We Need Something Between Absolute Faith and Absolute Denial

From the Election Day edition of the Morning Jolt

When It Comes to Polls, We Need Something Between Absolute Faith and Absolute Denial

Republicans, conservatives, and all assorted political junkies need to have a discussion about polling after this election.

Yes, the polls in 2012 were off, with the average pollster underestimating Obama’s lead and largely unprepared for how effectively the Obama campaign’s get-out-the-vote effort would change the electorate. But most of the polls were off by one, two, or three percentage points.

Yes, the polls in 2014 were off, with the average pollster overestimating Senate Democratic candidates’ performance by 4 percentage points and overestimating gubernatorial Democratic candidates’ performance by 3 percentage points. But quite a few elections that year were outside that margin.

Yes, every once in a while, you get a race where the election day winner didn’t lead any poll – like Republican Larry Hogan’s victory in Maryland in 2014 and Matt Bevin’s victory in Kentucky in 2015.

Yes, sometimes you’ll get very contradictory results from different pollsters surveying the same state at the same time. In New Hampshire yesterday, Emerson released its final poll of the presidential race, finding Hillary Clinton ahead by a point. WMUR/UNH released their final poll, finding Clinton ahead by 11 points.

But by and large, polls give you a ballpark sense of how a candidate is doing and how a race is shaping up. If a candidate is down ten points going into Election Day, they’re not likely to pull off a miraculous comeback. If a candidate has a small lead in the final polls, the final result is probably going to be close.

They’re not all “rigged.” (Every once in a great while, you will find a pollster whose results are so implausible – and so impossible to duplicate through random sampling – that analysis suggests they didn’t actually do the work and fabricated or manipulated the numbers.)

We can question whether a poll’s sample is precisely what the electorate is going to look like on Election Day – how the demographics stack up in terms of age, gender, race, and yes, party affiliation. Every pollster has to make their best educated guess of what the total electorate is going to look like. But history tells us that most of the polls will be in the ballpark. And when multiple pollsters surveying at the same time find a similar lopsided result, the outcome is pretty likely. In Ohio, Rob Portman’s lead has been in double-digits in every poll since late September. That one’s pretty clear.

We cannot just hope that the pollster who’s telling us what we want to hear the most is the one who is right. When one tracking poll is consistently giving us good news and the rest are consistently giving us bad news, maybe that one tracking poll is right and everyone else is wrong… or maybe it’s just an outlier. It is irresponsible for a campaign and a party to take a leap of faith that the best-case scenario is about to occur.

Absorbing bad news and negative feedback and learning from it is how campaigns get better.  Blind denial of evidence suggesting that what they’re offering isn’t appealing to the largest slice of the electorate is how they go careening off the cliff.


The Latest