Two nights before the New Hampshire primary, I was sitting in a Manchester bar arguing with a fellow reporter about the latest Al Gore-Bill Bradley polls. "I'm not gonna believe any poll," my colleague told me, "until I see John Zogby's name on it." A little-known independent pollster based in Utica, N.Y., Zogby rocketed to fame by correctly predicting the results of the 1996 presidential election. But like all religions, the cult of Zogby relies on myths about his feats and denial about his failures.
Zogby's dead-on prediction in 1996—he forecast Bill Clinton's eight-point win over Bob Dole, while most pollsters expected a much wider Clinton margin—won him a burst of media attention. What is odd is how Zogby's reputation has mushroomed ever since. Pundits, reporters, and chat show hosts now routinely genuflect to him. "Joining us now from Detroit is the nation's most accurate pollster, John Zogby," said Bill O'Reilly of Fox News Channel's The O'Reilly Factor in November. Robert Novak calls him "the country's hottest pollster." Dick Morris says he's "New York's most accurate pollster." The Washington Times' Wes Pruden bows to the "the hottest (and most accurate) pollster." Chris Matthews, a leading Zogby acolyte, welcomed him to Hardball in 1998 by saying, "John Zogby, you're the best pollster."
The cult grows despite Zogby's embarrassing miscues. He stumbled several times during the 1998 election cycle, the only big election since his 1996 triumph. One of the closest Senate races that year was the Al D'Amato-Charles Schumer contest in Zogby's home state of New York. Most polls found Schumer opening up a discernible lead in the final days of the campaign. Not Zogby, who predicted a D'Amato victory after his final numbers showed D'Amato holding a razor-thin lead. Schumer trounced D'Amato by 11 points.
Zogby went out on another limb in Illinois, where one-term Sen. Carol Moseley-Braun faced a tough fight against conservative challenger Peter Fitzgerald. Though Moseley-Braun managed a small comeback in the campaign's final days, most polls showed her trailing by as much as 10 points. But numbers released by Zogby International Nov. 3 gave Moseley-Braun a three-point lead. Mort Kondracke enthused the night before the election, "If Carol Moseley-Braun wins in Illinois and Al D'Amato wins in New York, John Zogby is going to go … right into the Polling Hall of Fame … because nobody agrees with him." Fitzgerald beat Moseley-Braun by three points.
Zogby is not a bad pollster. He's a very good one. He's just not head and shoulders above the rest. He flubs his share of races, but the media ignore his failures. They seize on evidence that confirms their beliefs and disregard contradictory data. Take Matthews' spin on Zogby's New Hampshire polls. No pollster predicted the scale of John McCain's 19-point victory over George W. Bush. But Zogby did give McCain one of his highest margins, a 12-point lead over Bush in his final tracking poll. So Zogby kind of called it, right? Matthews thought so. "You were clearly the closest pollster getting it," he gushed to Zogby two weeks later. What Matthews didn't mention, or maybe didn't notice, was that Zogby's final numbers also gave Al Gore a 12-point lead over Bill Bradley, even though Gore only won by five points. Most other pollsters noticed Bradley's late surge and got that five-point margin right.
So what explains Zogby's outsized reputation? Pundits have convinced themselves that he discovered a magical technique that elevates him above mere polling mortals: He only polls "likely voters." It's true that news organizations frequently release polls of registered voters or even all adults. But every serious political pollster screens for likely voters. The skill is figuring out who's a "likely" and who's not.
The most important reason for Zogby's popularity is that his polls make Republicans feel good. Conservatives clutched at his accurate prediction of the 1996 race because it seemed to show that Clinton wasn't so popular after all. Since then, Zogby's numbers have usually shown Republicans doing better than they do in other polls. (Zogby is a registered Democrat and, he says, a liberal.) My hunch is that Zogby's method of determining who's a "likely voter" emphasizes low-turnout elections, especially ones in which Republicans are disproportionately able to mobilize their base. That allows him to notice some Republican upsets that other pollsters miss. But it also sometimes leads him astray, as it did in the D'Amato-Schumer race.
It's hard to blame Republican partisans for treating Zogby's calculations as electoral truth revealed from on high. But what's the media's excuse?