Pollsters Offer Insiders' Advice on Appraising the Races

ORLANDO, Fla. — With the two major-party presidential nominees this year all but certain, pollsters gathered in a hotel conference room here on the third day of the American Association for Public Opinion Research's annual symposium with a colorful yet pointed message for their peers, the news media, and consumers of political news: Don't pay so much attention to the horse-race polls that dominate coverage of the campaign.

"They give you the score of the game," said Gary Langer, who produces polls for ABC News, "but they don't predict the outcome."

Langer, during an early Saturday morning panel discussion, emphasized other polling measures as better reflective of the current state of the race. He pointed to the wide favorability gap between President Obama and Mitt Romney, the presumptive GOP nominee. The most recent look from ABC News/Washington Post at the favorability ratings for both men came in mid-April, when the divide between the percentages of Americans who had a favorable opinion of Obama and Romney was a staggering 21 points.

A USA Today/Gallup poll released earlier this week showed Romney largely closing that gap -- and sporting a better net favorability-to-unfavorability rating. "I expect this to turn" in ABC News polling as well, Langer said, noting the USA Today/Gallup finding.

Langer also cited other troubling numbers for Obama in recent ABC News/Washington Post polling: Obama's low approval rating on the economy — 44 percent — makes him "clearly vulnerable," and the 64 percent of Americans who believe the nation is on the wrong track is "uncomfortably high for an incumbent."

Looking back to the Republican primaries this year, Langer compiled data from exit polls conducted prior to former Pennsylvania Sen. Rick Santorum's exit from the race and found that throughout the process, Romney struggled among the most conservative members of his party and ran best among those who prioritized a candidate who could defeat Obama in the general election.

Andrew Smith, the director of the University of New Hampshire Survey Center, which conducted polls in the Granite State during the primary season for the Boston Globe, CNN and WMUR-TV, brought his unique experience from polling the first-in-the-nation primary state.

"If you poll accurately in New Hampshire, it's luck," Smith quipped, noting, as did Langer, that members of the political news media make too much hay of the horse-race numbers and ignore the fundamentals of the race.

Smith joked that he should tell reporters who contact him at the early stages of the primary campaign, "You really need to get a hobby, because the voters aren't caring as much about this stuff as you think they are right now."

Unlike the general election, Smith noted that primary voters are more volatile because those races lack party cues — that is, partisans will largely vote for their party's candidate in November, but will be mostly satisfied with any of the primary candidates. "They're choosing flavors like ice cream," Smith said. "They'll be happy with whatever they end up with."

"The media have great incentives to report the horse race," Smith added. "It's an easystory to write. ... The viewers want this, the readers want to see this."

But Smith said that he and his colleagues "share some blame too" for this overemphasis. "We don't emphasize the fundamentals enough," said Smith. Smith implored his fellow pollsters to highlight what he says are the more important poll results in their press releases, pointing to the large percentages of respondents to UNH polls that said they were still trying to decide for whom to vote in the GOP primary.

"Make sure that you include the other measures of what's going on in the campaign," Smith said. "We have such great control over what the media print. They're lazy. They'll take what you put in your press release verbatim, almost."

Jeffrey Jones from Gallup described the results of a new experiment his firm launched this year. Gallup applied what it calls Q-score methodology — previously used to evaluate how people feel about news media personalities — to measure how Republicans and Republican-leaning independents felt about the candidates for the GOP nomination.

The metholdology involves first asking poll respondents if they recognize a candidate by name (without stating that he or she is a candidate for president). If the respondent says they know the individual, they are then asked if they have a favorable or unfavorable opinion of that candidate. If they have a favorable or unfavorable opinion of the candidate, interviewers then ask if they feel "strongly" or "somewhat" favorable or unfavorable.

The measure, called the Positive Intensity Score, is derived by subtracting the percentage who have a "strongly unfavorable" opinion of the candidate from those who have a "strongly favorable" opinion. The general trend was for intensity scores to go down over time during the GOP nominating process, Jones said.

Romney and Santorum ended up higher on the measure than when they started, but their numbers also fell until December and only rebounded as they racked up wins in the primaries. Both candidates also found their fortunes rise among Republicans according to this measure after Santorum dropped out of the race and Romney became the presumptive nominee, Jones said.