Nov. 11–DOVER — Pollsters at the University of New Hampshire have taken criticism from both sides of the aisle this year for their measurements of the presidential race, but in the end, it appears the UNH Survey Center was on target.
UNH conducted its final “WMUR Granite State Poll” on the presidential race between Nov. 1-4. The poll found Obama holding a 4 point advantage over Republican Mitt Romney, leading by a tally of 50 percent to 46 percent among likely New Hampshire voters.
The final poll conducted by New England College also had Obama winning New Hampshire by the same margin.
While none of the major polls conducted the week before the election hit the nail on the head, UNH and New England College appear to have come closest to accurately gauging voter sentiment.
On Tuesday, Obama carried New Hampshire by close to 6 points, beating Romney by a tally of 52.2 percent to 46.4 percent, according to results compiled by WMUR.
Voter turnout was high for both parties in New Hampshire this year, and with Democrats now appearing to hold an edge in the state’s electorate, the results of the election didn’t come as a surprise, according to UNH pollster Andy Smith.
“The biggest thing is there are just more Democrats than Republicans in New Hampshire,” said Smith, an associate professor who directs the UNH Survey Center. “We’ve been seeing for the last several years that Democrats have about a 5 percentage point advantage over Republicans in numbers in the state.”
Nearly all the of the polls released on the eve of the election showed a tight race in New Hampshire. Public Policy Polling and Rasmussen Reports both gave the advantage to Obama in their final polls, estimating that the president carried a 2-point advantage over Romney.
Gravis Marketing, a public opinion research firm that also develops software for campaigns, released a poll on Nov. 1 that showed Obama holding a 1-point lead, and carrying 50 percent of the vote.
“I think in the end, pollsters did a fairly good job,” said Dick Bennett, president of American Research Group, a polling firm based in Manchester.
ARG’s final poll predicted that Obama and Romney were tied among likely New Hampshire voters at 49 percent each. Bennett said ARG’s poll accurately predicted trends that emerged on election day, such as the schism between Democratic voters in urban areas and Republicans in more rural areas.
“We saw that coming,” Bennett said. “I think where our numbers diverged, the Republicans who were telling us they were likely to vote, the proportion, we were slightly high on that, and that’s basically the difference.” Bennett said one reason that UNH’s polling data has come under fire this year is because it tended to show large swings between polls.
One poll conducted by UNH showed Obama jumping from a 5-point lead in early September to a 15-point lead over Romney at the end of the month. A week later, Obama was back down to a 6-point lead.
Former Republican Gov. John Sununu was among those who criticized UNH’s projection of a 15-point margin in Obama’s favor. Sununu called UNH’s polling ”extremely volatile,” and dismissed the poll as a “piece of garbage” during an appearance on MSNBC, according to a report from the Concord Monitor.
“We’ve gotten a lot of abuse this year from Republicans who didn’t like the results they were seeing in our polls, but that’s part of the game,” Smith said after last week’s election.
Although some Republicans were dissatisfied with the UNH Survey Center’s polling methodology this year, Democrats have also voiced objections in the past when polls weren’t going in favor of their candidates.
In 2009, an editor at the progressive website Daily Kos called Smith’s polling “unreliable,” and said his analysis is “consistently Republican spin.”
“In short, Andy Smith acts, in his polling and his analysis, as a partisan Republican — but because he says he’s independent, reporters swallow every ridiculous claim he makes as truth from on high,” wrote freelance editor Laura Clawson.
In October 2012, Daily Kos Political Director David Nir encouraged readers to “not pay a lot of attention to UNH’s whacky results.” The message came after a poll showed Republican Congressman Frank Guinta shooting up in the polls and taking a 7-point lead over Democrat Carol Shea-Porter, who eventually defeated him.
During a panel discussion at UNH, Smith said the biggest problem with polling in any election is not determining voter sentiment, but rather, figuring out which voters are going to show up at the polls.
Polls are also subject to variability because many people don’t make up their minds until the day of the election. Voters can change their opinions in line with the day-to-day fluctuations in the campaign, he said.
For pollster Doug Kaplan, of Gravis Marketing, some of the criticism from conservatives this year is understandable because it was difficult to predict this year’s turnout.
Democrats turned out in droves in 2008, but many conservatives believed the phenomenon would not repeat in 2012, and that Democrats were being “oversampled” in political polling.
“I think the key is that everyone has good intentions…,” Kaplan said. “I just don’t think people understood the turnout. People just thought 2008 was a phenomenal year, and it wasn’t. Everyone now knows the electorate has changed on a presidential level.”
One big winner in Tuesday’s election was so-called “polling aggregators” — websites and experts who compile polling data to make predictions on political contests. One example is the website RealClearPolitics.com, which maintains a running average of the latest polling available.
The most high-profile example of a polling aggregator is the FiveThirtyEight blog at The New York Times, which is operated by 34-year-old statistician Nate Silver.
Silver’s election forecasts have proved uncannily successful in the past, and this year he correctly predicted the presidential winner in all 50 states, and almost all the Senate races.
In the run-up to election day, Silver predicted 90.9 percent certainty that Obama would win, and forecast him getting 313 electoral college votes. With the final vote tally now complete in Florida, Obama’s final electoral college tally is 332 votes.
Silver was already in the spotlight at the start of this election season, after correctly predicting the outcome of the presidential race in all but one state during the 2008 election.
This year, Democrats flocked to Silver’s blog and took daily solace in his consistent prediction that Obama would win the election by a small majority.
However, commentators on the right were critical of Silver, who was accused of weighting his forecasts too heavily toward Democrats.
Although Silver was vindicated in the election, Smith said he still has reservations about Silver’s methodology. Smith also pointed out that polling aggregators like Silver must still rely on the work of others to gather data. “They’ve done alright,” he said. “But you know what? I don’t give them so much credit as the pollsters who put the final numbers in.”
Bennett said taking a simple numerical average of the latest polls would produce about the same predictions that Silver came up with. Bennett said he questions how much value people like Silver add by applying complex formulas to make their prognostications.
Nevertheless, it appears voters are paying more attention to people like Silver during election season, Bennett said.
American Research Group has sponsored a contest during the last two presidential cycles asking voters to predict the outcome of all 56 electoral college races. Last time around, thousands entered the contest, and not a single person guessed every race correctly. This year, 13 people won.
“I think probably the aggregation is helping, because … we had basically ordinary people who were able to pick it,” Bennett said.
The Associated Press contributed to this report.