My Polling Pledge

In the last few days, a number of polls (see here and here) have shown a dip in support for President Obama, and the reasons are not entirely clear. Is it the rise in gas prices? Maybe. But what about the positive signs on the economy? All well and good, but perhaps the administration is undermining itself by making too much of them. But there are still almost eight months until Election Day, so we'd all be well advised not to make too much of any one poll or any momentary fluctuation.

Because that's what these kinds of tracking polls do. They fluctuate. Between now and Election Day, I promise you there will be polls that show Obama comfortably leading, polls that show Romney leading, and polls that show a tie. That was what happened four years ago, and what happens in nearly every election. Take a look at this chart of the 2008 election, from The trend lines show averages of all the polls—with Obama leading until March, then McCain leading for a couple of months, then Obama leading until September, then McCain leading briefly again, then Obama leading until the end. But you'll also notice the points, each of which represents an individual poll. They show a huge variation. At the beginning of March, a Washington Post/ABC News poll had Obama leading by 11 points. A few weeks later, a Rasmussen poll had McCain leading by ten points. Which was right? Neither, both, some combination of the two, and it depends on how you look at it.

In the end, Obama beat McCain by seven points, making all the hand-wringing along the way from Democrats seem kind of silly. Looking at these latest polls, I'm sure that conservatives are given great hope. Liberals, I know, are doing one of two things—dismissing the results as a momentary and inconsequential blip, or beginning to get very nervous. It's possible that my own inclination toward the former is just a product of my ideological bias and the desire it produces to convince myself that the facts actually are what I would like them to be. So I hereby make a promise to you, gentle reader.

Over the course of this campaign, there will be many, many more polls. Some of them will be interesting enough to comment on. But I pledge to you now that I will try not to make too much of any one poll, or even a couple of polls. I will remember that there are always outliers—by the accepted standards of statistical accuracy, if you ask a question a hundred times, five of the answers you get will be completely wrong. I will endeavor to maintain this long view even when a poll tells me what I want to hear. It won't be easy, and I can't promise that I won't ever slip up. But I'll do my best.

You may also like