The polling debacle of 2008 (so far!) was the industry-wide projection of a win for Barack Obama in the New Hampshire Democratic presidential primary. The all-time champ remains the blown call in the 1948 presidential election, when the pollsters were certain Dewey would defeat Truman. Yet these high-profile goofs distract from a larger problem with the surveying of public opinion - a problem that journalists and pollsters alike need to confront.

While horse-race polls get the most attention, in one key sense they're harmless. Eventually they give way to actual voting results: the only poll that matters, as the candidates trailing in the polls always remind you.

Not so issue polls, the surveys that purport to tell us that 72 percent of the public supports increased federal funding for children's health insurance. Such definitive-sounding results produce definitive-sounding headlines, and interest groups or politicians who like the result file them away for future citation. But have we really measured the pulse of the American people?

Don't bet on it, says David Moore, and he should know. He's the founder of the University of New Hampshire Survey Center and for 13 years helped run the Gallup polls. In his forthcoming book, The Opinion Makers, Moore makes a persuasive case that you don't really know what you think you've learned from public opinion surveys.

Take that 72 percent support for kids' health insurance. As you may know (Moore suggests that's pollster-speak for "fat chance"), a year ago President Bush and Congress battled over renewing the program known as SCHIP, the State Children's Health Insurance Program. It was an ABC News/Washington Post survey that announced the 72 percent figure. A CNN survey put the number at 61 percent. Close enough, right?

CBS pegged SCHIP support at 81 percent.

Gallup said it was only 40 percent.

This riddle is not solved by deeming one or more of the polls bogus.

The surveys came from some of the biggest polling and media companies, and Moore accuses none of rigging the outcome toward one policy preference or another.

But all these polls, Moore argues, are rigged to produce something that can be called Public Opinion. The media, he says, have no interest in reporting that most of the country isn't hanging on the outcome of a policy debate getting lots of news coverage. The pollsters, therefore, do their best to produce a definitive result.

Alone among the four SCHIP pollsters, Gallup asked people how closely they had been following the news. Only 17 percent said "very closely," 34 percent chose the socially acceptable "somewhat closely," 30 percent copped to "not too closely," and 19 percent confessed "not at all."

In other words, some 80 percent of those surveyed were not engaged enough to have an informed opinion on whether SCHIP rules should allow a family of four to qualify based on an income of, say, $41,000 or $62,000.

No matter. Gallup went ahead and told interview subjects that the Democrats wanted the higher figure and Bush wanted the lower figure and then asked, "Whose side do you favor?"

The kicker is, only 7 percent said "neither" or expressed no opinion.

From abortion to war funding, most Americans are not deeply engaged in the issues. That's okay: It's why we elect representatives. What's not okay is asking a random sample of us 15 or 20 minutes' worth of questions on all these issues and then announcing that almost all of us a) have an opinion and b) know what we're talking about. It isn't so, and there's no need to pretend that it is.