10/6/2008

This year’s New Hampshire primary will be remembered as the election the pollsters got wrong. For days before the vote on January 8, breathless pundits on cable news shows insisted Barack Obama and John McCain would run away with the election. Instead, Hillary Clinton prevailed by 2 points, and McCain beat Mitt Romney by 5. But polls conducted by the Suffolk University Political Research Center (SUPRC) stood out from the crowd—not for correctly predicting the victors, but for being the only polls that were accurate within the margin of error.

 That night, on MSNBC’s show Hardball, Chris Matthews conceded to former-NBC anchor Tom Brokaw that “maybe Suffolk knew something that everybody else didn’t.”

SUPRC Director David Paleologos would be the first to admit that he didn’t have any secret knowledge—just the attention to detail that has won the SUPRC (and Suffolk) accolades since it was founded in 2002. Paleologos insists that the accuracy of the Suffolk poll was due to clear-headed research and analysis. But if there is a secret weapon Paleologos uses to improve polling accuracy, it’s a technique—unique among pollsters—that he calls “bellwether tracking.” Paleologos picks several towns or counties that have historically reflected the outcomes of the state as a whole. For the New Hampshire primary, he scoured voter lists to identify two towns, Kingston and Sandown, with an uncanny record of picking the exact order of finishers in 1988 and 2000, and getting each winner within 10 percent of the statewide results.

 So on election night, when Paleologos seemed so out of sync with the other pollsters, he wasn’t relying on two polls, but six (one statewide and two bellwethers for each of the two primaries). The bellwethers were more accurate than the statewide polls: both had Clinton leading in the Democratic primary, and for the Republican primary, one had McCain ahead while the other favored Romney. Paleologos sees the bellwethers as a way of adding another layer of confidence to the statewide results—akin to a doctor running three tests instead of just one.
 
The SUPRC was created six years ago, when students in Paleologos’s class, The Art of Polling, conducted a poll on the Massachusetts governor’s race. Their results earned coverage in the Boston Globe and the Worcester Telegram. A few months later, Paleologos began to collaborate with WHDH-TV, Boston’s Channel 7, correctly predicting Mitt Romney’s victory over then-State Treasurer Shannon O’Brien. (The Globe and the Herald predicted an O’Brien win and a tie, respectively.) “And from that moment on,” says Paleologos, “we have been doing polling.”

 Paleologos recognizes that polls are just one part of a complicated political picture; ultimately, it’s the voters who decide. After all, while his polls were the most accurate in New Hampshire, he got caught up in the Obama-mania and called the race for Obama just like everyone else, even though both of his bellwethers showed Clinton ahead. “It’s a good lesson to me, that I need to keep bringing this back to the science,” he says. “If those bellwethers and that poll all show the same thing, I’m prepared to go off the cliff with it.”

The full version of this article ran in the Spring 2008 issue of Suffolk Alumni Magazine.

Back to News »