(613) 818-2848

What the U.S. election polls are telling us

Do you remember the crazy poll swings before the Canadian federal election last year?

Slowly coming back to you now?

Did you spend lots of time reading about the polls, and how things seemed to be dramatically changing day after day, for months? Lots of us did, and we didn’t really learn much along the way.

I was reminded of this the other day when I read an instalment of the blog Wolves and Sheep: Political analysis from outside the Beltway, by Matt Kerbel. This instalment was entitled “About Those Polls”. In the blog, Kerbel argues that, although the U.S. presidential campaign is being written about as if it is just getting into full swing, the reality is that it is beginning to wind down. Even though the pundits are telling everyone who will listen that the race is tightening, Kerbel points out, the fundamental structure of the race has not been changing.

So you could gnaw your fingernails daily reading about the polling numbers, or you could accept that there are fundamental issues at play underneath the numbers, and that the numbers will settle down as election day draws near.

Look again at the first figure and those Canadian polls from mid-August through to election day last year. All over the place. Big excitement as the NDP and the Conservatives, of all people, seemed to be trading places in the hearts of the public. In reality, the polls all the way through to early-mid September bore little relationship to the eventual result, except for the gradual rise in Liberal numbers. At that point, either the public started to solidify their views, or the pollsters started to adjust how they polled in view of the trends, or both. I’ll say more the former, to be charitable.

The second figure plots the difference between the percentage of popular support some of the better-known polling firms were reporting and the final numbers over roughly the last two months of the campaign. I produced this after the election using data from the excellent Wikipedia series “Opinion polling in the Canadian Federal Election, 2015” that ran at the time. The forecasts improved relative to the final result over time, but for most of this period they were still way off. You can see how much better the polling firms did once people had made up their minds, in the last days of the campaign.




So what? What did all of that crazy statistical action and breathless commentary in the months before October add up to? To use a colourful phrase to go with the colourful charts, bupkis. That’s probably what the U.S. polls are worth right now, too.

So, if you are spending your time reading about how the race is tightening, etc., etc. and feel your blood pressure rising, go out and enjoy the lovely fall weather. The U.S. electorate are going to do what they are going to do, and we’ll know soon enough.

* I admit that the fading-in chart at the top was a shameless attempt to catch your attention and to throw in a little custom-coded D3.js chart-making. Check out other examples on this site.

You can subscribe to my blog on the “Signup” page of this site.