Don't believe the numbers.

Photographer: Win McNamee/Getty Images

Hate the Polls? Blame the Media.

Jonathan Bernstein is a Bloomberg View columnist. He taught political science at the University of Texas at San Antonio and DePauw University and wrote A Plain Blog About Politics.
Read More.
a | A

OK, let’s talk early polls in presidential elections. What exactly do they tell us?

They tell us what people are hearing, and sometimes that will matter. While nomination surveys can't successfully predict what will happen in next year’s primaries and caucuses, let alone the general election next November, they can -- interpreted carefully -- say something about what’s happened so far.

Ed Kilgore at Washington Monthly makes that case, in response to a Nate Cohn item at the Upshot (see also Harry Enten at FiveThirtyEight). As Kilgore points out, polling can have real impact even when it's nonsense if the right people take it seriously.

This may have happened on the Democratic side over the last couple of years. Hillary Clinton, temporarily shielded from normal partisanship during her time at the State Department, polled unusually well. It’s possible that party actors -- the politicians, campaign and governing professionals, formal officials and staff, donors and activists, and aligned interest groups and media that matter most in nominations – saw those numbers and believed they would hold up through November 2016.

If so, those polls helped her get a jump on securing the nomination early, even if she will likely perform as well as any Democrat would in the general election.

Even if you believe party actors are only minimally affected by polls and accept that the horse-race question isn't predictive at all, some claim you can find out plenty in the questions about candidate favorability, for example, an argument that Kilgore shares with Cohn and Enten. 

I'm skeptical. 

While some voters will answer a pollster's questions right now, most people aren’t focused on presidential elections this early. Even fairly attentive voters can’t tell a Marco Rubio from a Scott Walker or, for that matter, from a George Pataki. And they may even be a little confused about which Bush they’re being asked about.

So what are we hearing when voters claim to like one candidate and dislike another (say, Chris Christie, who polls poorly now)?

It’s going to be a mix of three things.

One is reaction to opinion leadership. Rush Limbaugh listeners, for example, will adopt Limbaugh’s opinion; liberals may adopt what they’ve heard Nancy Pelosi say.

Second, there's a general echoing of what has been in the news lately and over time. Brand-new candidates are usually introduced with a flattering biography, for example, which can produce a nice bounce in favorable numbers. This can stall or be reversed when new information is reported.

And third, some direct reaction from voters about the candidates is less affected by what they’ve been hearing from others, and might hint at how the larger electorate will react once it has more exposure to the candidate.

Alas, it's impossible to untangle those three factors. And it's likely that the only one with any real value-added -- the third one measuring direct reaction -- has the smallest influence, given that hardly anyone watches a lot of C-SPAN, for example.

If this is true, then all we know from the careful dives into early polling numbers is how the press is portraying the candidates. And we need evidence outside of polling to predict how the candidates will be portrayed in the future.

Take the case of Christie, who is unpopular among Republicans generally, especially among conservatives. This attitude is likely to continue, not because rank-and-file conservatives inherently dislike him, but because Republican opinion leaders, especially conservative ones, will continue to signal they want nothing to do with him. So the place to look isn’t in the polling numbers. It's in careful reporting about what Republican party leaders are thinking. This in turn will eventually affect what the party's opinion leaders say. 

So, when you read polls saying what voters think, take it as a shorthand for what the media has said so far about a candidate. And while some political actors may take it far too seriously and may even make decisions based on that information, it otherwise isn't predictive of anything. For that, we have to wait until voters get engaged – in the last few weeks before they are scheduled to vote. 

This column does not necessarily reflect the opinion of Bloomberg View's editorial board or Bloomberg LP, its owners and investors.

To contact the author on this story:
Jonathan Bernstein at jbernstein62@bloomberg.net

To contact the editor on this story:
Katy Roberts at kroberts29@bloomberg.net