Public opinion polling has gotten out of hand. You do not have to take a survey to verify it. Nor do you have to be a statistician to understand the numbers:
In the last month-long period, there were 11 polls conducted in Iowa, 10 in New Hampshire and nine nationally. There were polls focused on 10 different states. And that was despite the holiday breaks.
The proliferation of polls might not in itself be troubling if they were a reliable sampling of public opinion. But they are not. Despite the claims of scientific method and total impartiality, the poll results are wildly contradictory.
Take, for example, the polls in New Hampshire last week. Bernie Sanders was ahead of Hillary Clinton by three points — according to Gravis Marketing and One America News Network. However, the research fellows at CNN and WMUR got a slightly different result — Sanders up by 27.
In actual elections, they have been a poor guide to what the people want. They wrongly predicted the outcomes in the recent parliamentary elections in the UK and Israel, as well as the 2014 midterms in the United States.
Last year, in the Kentucky governor’s race, the polling consensus had the Democrat, Jack Conway, beating the Republican, Matt Bevin, by five points. But this was only in the pollsters’ minds. On Election Day, Bevin won by nearly nine.
Nobody, including the industry professionals, dispute these numbers. Furthermore, they acknowledge that the problem is getting worse.
In large part, that’s because the available pool of respondents has been shrinking. The accuracy of a survey rests on the principle — or hope — that those who answer the phone will give answers representative of the larger population that is not reached. But that is only true if the sample is large enough.
Response rates used to be much higher. Back in the 1980s, a 60-percent rate was not unheard of. Pollsters feared a drop in that rate, and that if it ever got down to, say, 20, they would be out of business. Response rates these days are often less than 10 percent!
The change is due to a law preventing polling firms from accessing mobile phones by automatic dialing, at a time when millions of people are switching to them and abandoning their landlines. It is also because many people who are reached by phone will not talk to pollsters, because they distrust them. A 2013 study found that three out of four Americans suspect polls of bias.
Here, we may see a vicious cycle at work. Polling accuracy is down because of the smaller sample; because of declining accuracy, people are less inclined to trust them, thus shrinking the available pool of respondents; the subsequently even smaller sample further skews the results, and so on.
None of this bodes well for an industry that makes claims to scientific accuracy. Yet, as we noted at the beginning of this discussion, polls are playing an ever bigger role in the election campaign. It’s an obsession.
Politicians have for decades used them to make their campaign strategy, journalists to fill news space, ordinary citizens to follow the action. Now, the media debate organizers are using them to determine who shall be allowed to participate on a crowded debate stage. Carly Fiorina was excluded from a recent debate because her poll numbers were too low. Even the candidates’ positions on stage have this season been decided according to their rankings in the polls.
Thus, the influence of the polls increases, even as general awareness of their inaccuracy mounts. As one prominent pollster exclaimed after the Kentucky race mentioned above, “It was 14 points off. But everybody shrugs and moves on down the highway.”
There are, however, signs of a change in direction. The sinking response rate is one important sign. Another is that politicians themselves are beginning to shake their addiction to polls. Mitt Romney’s pollsters believed, even as of Election Day, that he would win the 2012 presidential election, never mind what the polls said.
The decline of polls may be in sight. There may come a time when newspapers and web sites won’t bother publishing poll results because nobody will be interested in reading them. But it won’t be anytime soon. Too many people have a vested interest in keeping the industry going.
Is it a bad thing? Some think so. As Irving Kristol once observed, the purpose of a political campaign is to let the people know what the candidates stand for, not to let the candidates know what the people want them to stand for, so they can tell them what they want to hear.
Then again, scientific polling was just an enhancement of what had been going on in American politics from its inception. Indeed, there was a school of thought in the early days of the nation that held that the task of an elected official was to represent his constituents’ opinions and nothing more. Even those who believed that in a republic their job is to provide wise leadership for the uninformed, uneducated masses admitted that they should not stray too far from popular sentiment.
Periodic elections were the constitutional guarantor that they would not. Thus, the most successful politicians were those who could tell which way the wind was blowing by sniffing it at a party meeting or street rally.
Then, too, the polls have become part of the hoopla of American politics, like marching bands — hardly a rational way to choose a candidate, but tradition and reason need not be synonymous.
Most of all, though, in a world so ruled by technology and data, the failure of polling to rule us is a happy thing. The polls, in their inaccuracy, provide a foil to the splendor of human unpredictability, the triumph of the voter’s last-minute decision-making over the legions of statisticians and their computer projections. It is the citizen’s way of thumbing his nose at the know-it-alls, of asserting his sovereign right of free choice.