fb-pixel Skip to main content

DURHAM, N.H. — Pinning down voters about who they might support in the New Hampshire presidential primary may be political gold for a candidate, but history has proven it’s not an easy race to predict.

“More often polls are wrong than they are right,” said Andy Smith, director of the University of New Hampshire Survey Center. “I think the 1984 Gary Hart win was the worst. Any pollster who says they know who will win the primary doesn’t know what they’re talking about.”

New Hampshire is well known for unpredictable voters who can launch underdogs into the national spotlight. Hart’s win in the New Hampshire primary, upending predictions favoring Vice President Walter Mondale, is an often-cited example of pollsters who spent weeks surveying voters getting it wrong.


Smith said a number of factors make New Hampshire voters tough to predict — especially during the presidential primary — but one constant has remained true over the years.

“People will make up their minds at the very end,” said Smith, who recently coauthored a book with David Moore titled “The First Primary: New Hampshire’s Outsize Role in Presidential Nominations.”

With 2016 promising a double-digit field of Republican candidates, pollsters may be facing an even harder go of it while their surveys are being called upon in new ways to shape the election. National polling is already being used by CNN and Fox News to narrow the number of candidates on stage during the first two Republican presidential debates.

“That’s a real world implication for this cycle that wasn’t present in previous cycles,” said Tom Jensen, director of Public Policy Polling in Raleigh, N.C. “The other area where polling really matters is for fund-raising purposes, especially if you’re one of those candidates who are skirting the line.”

A handful of Republican hopefuls have already hired private polling firms as part of their New Hampshire operations in advance of a possible presidential run.


“A real survey helps you determine who is responding to the things you want to talk about and how to best communicate your message,” said David Carney, a GOP strategist based in New Hampshire. “That’s what it’s about.”

Campaign polling should not be confused with public opinion polling reported on by the media, Carney said. Campaign polls have no interest in the horse race aspect of elections, instead delving into questions that help a candidate learn the “where, who and how to best deliver your message,” he said.

Carney criticized public opinion polling, saying it falls short with how and what kind of information it collects from voters. At the same time, campaigns will be quick to use a public opinion poll when it works to the advantage of their candidate, and just as quick to trash one that goes against them, Carney said.

“The political spinning of polls is the most egregious part of a campaign,” he said. “It’s the most overused technique.”

Jensen said polling results in New Hampshire and Iowa will likely play a significant role in shaping the 2016 election in the months before voters reach the polls. The prominence in polling comes at a time when fewer people want to participate in surveys, he said. For every 100 people Jensen said his company reaches out to for an automated political poll, he will be lucky if 10 percent will participate. More recently, that number has dropped to 5 percent, he said.


“You now you have a situation where you have 18 or so candidates on the Republican side,” he said. “You worry a little about exhausting the respondents.”

Landline phones are on their way out. People have shorter attention spans, and more pollsters in 2016 are turning to the Internet in an attempt to get a well-rounded view of the electorate, Jensen said.

“The simple reality is there is no one single way to reach everybody in an election,” he said. “We are going to see more of a fusion methodology.”

Despite his criticisms of public polling, Carney said CNN and Fox News should have used a blend of polls from states with early primaries where candidates are spending most their time instead of national polls where voters are not yet paying attention to the election.

“It would have been a more accurate barometer,” he said. “It would have given candidates an opportunity to strut their stuff. You can’t have 18 people on stage.”

Smith said over the years he has honed the types of questions he asks potential voters that have given him more telling results. Likability tends to be a better predictor of who may emerge as a winner in an election, he said.

“It’s always difficult. It’s always a crap shoot, so one of the things I try to do to make my understanding of the race better is to look at favorability ratings,” Smith said.


In 2000, when polls indicated that US Senator John McCain was tanking in race, he still rated well in likability, Smith said. McCain went on to defy polls predicting his loss in the New Hampshire primary and defeat George W. Bush.

Smith now asks voters what candidate they believe has a high likeability factor; what candidate they would never vote for under any circumstance; and finally, regardless of who they vote for, who do they think will win?

“That question is more likely to predict who will win than the horse race question,” he said. “More often, it gets winner right.”