HICKS: The problems with polling and predictions

Just days before a presidential election, there’s no doubt we will be bombarded with poll results and election models designed to predict winners and losers. It is useful to explain how these work without technical jargon. Let’s begin with polling.

Conducting voter polls has become more difficult in recent years, despite (or maybe because of) a spate of technical developments that should have made it easier.

Simply getting ahold of voters should be easier due to the ubiquity of cell phones. Unfortunately for pollsters, human behavior has adjusted—many of us no longer pick up calls from unknown numbers. And more families—including mine—have disconnected their landline telephone.

This means a number of potential voters are largely unreachable by phone. If these people were randomly distributed across the population spectrum, it wouldn’t bias a poll—but they are not.

Moreover, how much they differ from the population as a whole is not well-known for the same reasons we cannot contact them to ask about elections.

We do generally know that owners of landlines tend to be older. We also know that a higher share of younger people have cell phones than do older folks.

So when we call a few thousand people in order to get ahold of a few hundred, we don’t really know if they are a good sample of the population as a whole. This generates the first real polling problem: getting a representative sample of the population.

The second problem pollsters face is determining which of the folks contacted will actually vote. This is very important because only a little more than half of all eligible Americans are likely to vote on Tuesday. Getting this even slightly wrong can skew a poll.

As with the cell phone problem, if the likelihood of voting were evenly distributed across the population, this wouldn’t bias the poll results.

However, the propensity to vote is affected by age, income, enthusiasm for a particular candidate, and/or the belief your candidate will win—factors that tend to favor one candidate or another, making a pollster’s job even tougher.

And then there are those people who lie about whom they will vote for.

In light of these types of problems, economists (and increasingly political scientists) favor observing what people do, rather than asking them about it. This is what statistical vote models do, albeit mostly using historical data.

A number of researchers have built models of vote predictions that include candidate favorability ratings, national and local economic conditions, consumer sentiment and the like.

The best of these models look at state or sub-state regions, predicting voter turnout and winners based upon historical relationships between these conditions and votes. All in all, we should expect some significant surprises on Tuesday night.•

__________

Hicks is director of the Center for Business and Economic Research at Ball State University. His column appears weekly. He can be reached at cber@bsu.edu.