What the Polls May Really Be Telling Us

There are lots of political polls out there this year and there seems to be something for everyone. Hillary Clinton’s team seizes every opportunity to taut the polls – almost all of which show her leading nationally and in key battleground states – as proof of an inevitable victory, possibly even a large mandate to govern. Donald Trump either selectively chooses the polls that show him doing better or dismisses all polls as part of the rigged system. For supporters of each there is plenty of confusion. How can there not be? Is Clinton leading by 12 points? By 7? By 4? Or is Trump leading by 1 or 2? Or is it tied? To be sure, everyone cites his or her favorite poll and dismisses the others. And, of course, many folks are just dismissing all polls.

Here is what I know and I hope it is helpful. The field of political polling is in crisis. This is not new or all of a sudden. It has been heading for this crash for well over a decade and some of us saw it coming and have written extensively about it. The first problem is technology.When I first started in 1984, 96% of US households had a landline telephone and two out of three reached were willing (even eager) to do a survey. Steadily, answering machines entered the picture, both adults began working, call waiting, Star 69 entered the picture and so on. The “long distance call” was no longer a social event in the household, it was a nuisance. With the advent of cell phones, a whole new problem emerged. Pollsters were now invading people. And a typical poll is not a couple of questions, but perhaps 40, 50 or 60 questions – something hard to do on a cell phone. Response rates plummeted. Today, landline penetration is about where it was in 1957 and response rates border on the negligible. I have been both a pioneer and an advocate of online polling. It is better suited in many ways for most opinion research.

Beyond the technology and social change in the household, we have a breakdown in the stability of our political life. It used to be that the two parties nominated popular characters who could count on 90% support of their respective party’s voters then try to move to the middle to win over enough centrist undecided voters to achieve victory. We have never had two candidates like this year – detested and distrusted by majorities, especially those voters who still have not made up their minds. When voters tell us that “I will vote the lesser of two evils but I just have not made up my mind which one that is”, that makes things unpredictable. It makes turnout unpredictable. In the past decade, as many as 15 – 18% have told us that they did not even make up their minds to vote, let along who to vote for, until Election Day, and the volatility soars.

Thirdly, the expectations for polls just have gotten way out of hand. I heard on the radio the other day, that Gallup – the dean of polling – was “way off in 2012 because they missed the actual results by 2 points”. That is ridiculous. Accuracy is important and the era of being accurate all the time within 1 percentage point just does not exist anymore – without a little luck anyway. So those who track the accuracy of the polls may have some fun but they will put themselves out of business in short order, too.

Finally, a word of caution to those who aggregate polls, track social media, write algorithms, and use artificial intelligence to make projections. We pollsters poll people not numbers. We poll public opinion, values, emotional responses, levels of intensity. It may be elusive at times, but we try to capture what makes people human – how they process information, how they react, what they hold dearest, how they feel about a candidate or issue, even with whom they would like to share a beer. Polling is a very human endeavor. It is not a basket full of data and algorithms. It is a study of people not an abacus counting vegetables and currency.

The polls are telling us a lot today – but more about polling than about the Presidential race.