Media polls expose their limitations

The widespread unease with the performance of the political press in this week’s Wisconsin Democratic primary raises questions worth keeping in mind as the presidential race picks up velocity.

The most consequential of these questions center -- once again -- around the accuracy and impact of the various news organizations polls:

Why did they fail to pick up the surge in support for John Edwards?

Did postelection reports and analysis seeking to compensate for that failure overstate the surprising quality of the North Carolina senator’s close second-place finish to front-runner John F. Kerry?

The answers to both questions tell us a number of significant things about the limitations and shortcomings of the media’s political polls. Some have to do with the simple mechanics of polling; others have to do with the way in which the news organizations’ techniques for surveying public opinion may be falling out of sync with the contemporary electorate.

First of all, no media organization polls right up to election day. Everybody stops at some point, and from that moment on, political reportage is quantitatively blind until surveys resume with the exit polls. In Wisconsin, the last independent measures of voter sentiment of any sort occurred 48 hours before the polls opened. As we now know, an extraordinary number of primary voters made up their minds during the period after all polling had stopped.

In other words, you can’t measure a sentiment people don’t yet have.

CNN political analyst Bill Schneider, who also is a highly regarded scholar of public opinion, said that “last Monday, the day before the primary, my producer kept asking me, ‘Doesn’t anybody have some fresh numbers?’ I had to keep reminding him that the last news organizations in the field had wrapped up their polls Sunday morning. That was before the candidates’ debate on Sunday night and before many Wisconsin Democrats learned that the state’s leading newspaper had endorsed Edwards. Those two events turned out to be immensely influential in helping undecided voters make up their minds, and there were no polls to measure their impact.”

This isn’t a new problem. The biggest electoral surprise ever -- Harry S. Truman’s upset win over Thomas Dewey in the presidential race of 1948 -- occurred in a year in which the last national poll was conducted a full month before election day. In the political forest, a tree can fall with nobody around to hear, and nevertheless end up hitting somebody between the eyes.

In Schneider’s view, the point at which polling stops has become a more consequential issue as the national electorate’s behavior has changed. Wisconsin may not be so unique a case. “These days, people are making up their minds later and later than ever in elections at all levels,” the CNN analyst said. “This has become a very real problem for political polling. Cutbacks and cost-consciousness among media organizations have reduced the number of polls conducted and, nowadays, most are done on the weekend because it’s cheaper to survey people then.

“The problem is that you can’t force people to pay attention to politics and most of them don’t until they have no choices, which means they’re making up their minds closer and closer to election day. In the presidential election of 2000, for example,” Schneider said, “public disinterest in the contest between George W. Bush and Al Gore was as high as we’d ever measured in a national campaign. Nobody cared. Then Florida threw the whole thing into overtime, and public interest became intense -- as if it were a sporting event.”

Critical readers of the news media’s political polls need to keep some other inherent limitations in mind. When they’re well-conducted, preelection polls are extremely good at measuring the movement in sentiments and issues pollsters already know exist. Preelection polls are like a series of snapshots and if you look at them in sequence, you get something like one of those old-fashioned flip books.

That’s also true of the exit polls, which are designed to elicit the reasons people voted as they did and to provide material for postelection analysis. In essence, the problem is that even the best pollsters can’t write questions to tell them about things they don’t already know exist. Like ordinary mortals, pollsters don’t know what they don’t know.

The real question about Edwards’ surprisingly strong finish in Wisconsin is whether it can be attributed to new voter sentiments and electoral alignments that none of the media polls are picking up because no one -- up to now -- has known they were there to be measured. Some of the post primary polling seems to indicate that may be true.

Edwards’ campaign is built around a couple of key issues he argues are interrelated: the economy’s continuing failure to arrest the loss of jobs and the domestic impact of free trade agreements, like NAFTA, which both Kerry and Bush supported. In Wisconsin, those themes struck a strong chord in unusual places. Historically, the more highly educated and affluent a voter, the more likely they are to support free trade. In last Tuesday’s primary, Edwards beat Kerry among college-educated voters and among voters who earn more than $50,000 a year. Three-quarters of all the voters who went to the polls Tuesday told the exit pollsters they believe free trade agreements have cost the United States more jobs than they have created.

In American politics, anti-free trade sentiments traditionally have been populist staples. The Edwards voters may be something new -- what might be termed an anti-Wal-Mart bloc. (Without passing judgment on the discount giant’s business model, it increasingly is seen as the symbolic apex of a globalized trade arrangement indifferent to issues of equity.)

“For upper-middle-class Americans,” Schneider said, “trade is no longer an economic issue, but a moral issue. An increasing number of people seem to be saying, ‘I won’t throw other Americans out of work so I can pay less for whatever at the local big-box discounter.’ That’s significant because whatever their incomes or education, American voters don’t really think about the economy. They think about jobs. Without jobs there is no America and no American dream.”

These used to be the sentiments of organized labor and the working class. But in an era when even the new economy’s knowledge workers are seeing their jobs move offshore -- to India and Singapore, for example -- such social distinctions are breaking down. As Schneider put it, “people of all sorts seem to be telling us that a jobless economic recovery is no recovery at all.”

Whatever difficulty the polls had in hearing people say that in Wisconsin, it seems likely they’ll be listening closely between now and Super Tuesday.