John Dickerson on the perils of polls

John Dickerson on the perils of polls
John Dickerson on the perils of polls 02:42

We are in the high season of political polls. It feels like we are pelted with a new one by the hour. Public interest and the looming Election Day charge the atmosphere. 

Are the polls solid? Can they be trusted? What's the sample size? I've never been called. 

Partisans bicker over interpreting the polls as if the election were on the line in that moment. In jittery times, polling keeps everyone hopped up. Maybe we should ignore them. We'll know soon enough.

But we should not ignore the polls, if for no other reason than political polling encourages humility. That's useful in politics, or any other public issue, in an age where everyone thinks they're so right about everything.

plethora-of-polls-620.jpg
Despite what we think we know, there is uncertainty in political polling - and danger in relying too heavily on what polls "predict." CBS News

Over the whole stew of political polling looms the belief that the "polls were wrong" in the last election. This is the popular view. It is also the wrong view.

In 2016, the average of national polls showed that Hillary Clinton was leading by around 3%. When the votes came in, she won the popular by a hair over 2%, very close. 

What was wrong was the way a lot of us thought about polls, and thought about the forecasts being made about who might win the election. Hillary Clinton was given anywhere from a 70 to 99% percent chance of winning. Many people, even some who follow elections for a living, decided to round that number up to 100 percent.

The polls aren't to blame for that, any more than the weather forecaster deserves blame for your lack of an umbrella when a 30-percent chance of rain is predicted.

In this way, the political class repeated a familiar mistake of leaning too hard on the numbers. In 1936, they over-read a poll taken in Literary Digest which showed Alf Landon beating Franklin Delano Roosevelt. What they missed is that all they'd predicted is how the readers of Literary Digest felt, not how the voting public felt. 

They mis-read the electorate, which is a version of what was behind pollster mistakes in states in the Midwest where they did get it wrong last time. 

It's the reason pollsters will be the first to warn about the uncertainty of polls. Voters and pundits may need certainty from them, but that's on us. Don't blame the polls for that.

         
Story produced by Ed Forgotson and Robert Marston. Editor: Joe Frandino.