Political polls can have bad consequences for election outcomes

Oct 24, 2014

On Tuesday, The Detroit Free Press came out with a poll showing Gov. Rick Snyder eight points ahead of his challenger, Mark Schauer. That was the widest margin we’ve seen in a while. Most polls have had it much closer.

But within a day after that poll, news stories started matter-of-factly referring to the “fact” that Snyder was eight points ahead, as if these were actual, counted votes, or bushels of grain.

The sheer silliness of that wouldn’t matter much, except that polls drive pretty much everything in a campaign these days: Nobody wants to give money to a loser. Nobody wants to stand in line in the rain to vote for one, either. Polls can be self-fulfilling prophecies.

Except … those of us who have been around a while remember the governor’s race in 1990, which seemed like a one-sided affair. Going into the final weekend of the campaign, the last poll showed Jim Blanchard, the incumbent, ahead of challenger John Engler, 54% to 40%.

But in the upset of all upsets, Engler won. Stunned, one pollster went back and took another post-election poll.

Voila – it once again indicated Blanchard had won, when he had, in fact, lost. Usually the polls are more or less right, but when they are wrong, they are sometimes catastrophically wrong.

Usually the polls are more or less right, but when they are wrong, they are sometimes catastrophically wrong.

If predicting the wrong winner was the only bad thing about them, it wouldn’t matter much. But again, the polls badly distort our election campaigns and how we cover those campaigns as well.

This isn’t just the case in Michigan, it’s everywhere. Earlier this week I interviewed Ed FitzGerald, the Democratic nominee for governor of Ohio, on a half-hour TV program I host in Toledo.

FitzGerald has little money and is given next to no chance to win. But I mostly ignored that. Instead, I focused on what he would do and what kind of governor he would be if he did win.

I asked tough and hard questions. Afterwards, he told one of the producers that I was not only fair, but almost uniquely relevant. That is, I asked him questions about how he’d do the job.

Most of the time, FitzGerald said, all he was asked about was polls, campaign fundraising, and two minor scandals.

Few journalists had asked much about what he’d do as governor – which I think is sort of dereliction of our right-to-inform duty. Thirty-five years ago, I worked as a reporter for a newspaper publisher who was also a scientist. He would not publish any head-to-head candidate poll results in his newspaper.

He told me that was because he believed the famous Heisenberg uncertainty principle in physics applied to politics.

Roughly, this means that to observe a phenomenon is to affect it. That’s bad enough; what he thought was even worse was taking a momentary snapshot of a tiny fraction of voters that might or might not be a representative sample, and indicating this was reality.

At the time I thought not publishing polls was horribly backward. Today, I tend to think he may have been absolutely right.

Of course, I’ll have to wait for a poll to show what percentage of the voters feel the same way.   

Jack Lessenberry is Michigan Radio’s political analyst. Views expressed in the essays by Lessenberry are his own and do not necessarily reflect those of Michigan Radio, its management or the station licensee, The University of Michigan.