News your connection to The Boston Globe

Real-time poll aims to feed survey appetite

Meter to chart voters' reaction to TV debate

In the latest bid to satisfy the public's seemingly insatiable demand for campaign polls, CBS News is unveiling a real-time, Internet-based survey of 200 undecided voters watching tonight's presidential debate.

As George W. Bush and John F. Kerry face off, bar charts on the CBS website will record immediate reaction to the candidates, producing a bouncing thermometer graph much like the volume meter on a home stereo system. ''We're going to create a whole new class of political junkie," boasted Mike Dennis, the vice president of Knowledge Networks, the California polling firm hired by CBS to run the survey.

CBS News' real-time meter is the newest twist in a campaign year already brimming with horse race data. More opinion polls have been conducted in 2004 than in any other presidential election up to this point -- and by more varied pollsters. When amplified through the Internet and partisan supporters, these polls can have more power than ever to influence public perceptions of the race.

Yet today's opinion polls are potentially less reliable, and often contradict each other.

Survey takers say that the declining number of Americans willing to talk to pollsters is making their job more difficult than ever, and that some of the public's demand for numbers is being met by untested pollsters who post their results on dozens of political websites.

''The polling business is a pretty low-cost business to enter, and that partly explains the uptick in polls," said Michael W. Traugott, a communications professor at the University of Michigan. ''Many firms only operate in a state or two."

Even among established firms, assembling reliable samples has become more difficult. Fewer Americans answer surveys, either because they use a cellphone, they aren't at home when the pollster calls, or they simply don't want to talk to a stranger. So opinion polls are more expensive, take longer to compile, and theoretically carry more of a risk that they don't reflect what will actually take place in the voting booth.

''It's a lot harder these days than it's ever been before," said pollster John Zogby. ''It's not impossible, it's not a crisis, but it's more difficult."

Reaching a reliable sample of voters, though, is just the first problem; turning those responses into the final poll numbers presented to the public can be much trickier, and more subjective. A series of polls that seemed to show widely disparate numbers in the presidential race over the past two weeks moved long-standing disagreements over methodology out of statistics journals and into the public debate.

While major polls showed Bush opening a lead against Kerry, most concluded that the gap was in the single digits. But a poll by the Gallup Organization two weeks ago infuriated Democrats by reporting a 13-point lead for Bush among likely voters. (A more recent Gallup Poll put the gap at 8 points.)

In response to the 13-point poll, the liberal group took out a full-page ad in the New York Times Tuesday attacking Gallup's methodology. The Kerry campaign was forced to go into damage control in an effort to prevent the poll from damaging public perceptions of the candidate.

''Our poll is completely objective and nonpartisan," said Frank Newport, Gallup's editor-in-chief. ''We're objective scientists doing the best we can. . . . If Kerry suddenly moves into the lead, you would no doubt hear from the Bush campaign right away. That's par for the course."

Gallup, polling experts say, uses a more demanding method to identify ''likely voters" than many of its competitors. During surveys, respondents typically are asked a series of questions to gauge their likelihood to vote, including if they voted in the last election and know where their polling place is this year.

In polls that include all registered voters, or use looser definitions of ''likely voter," Kerry often polls more strongly.

His supporters expect Kerry to do better on Election Day, drawing new and infrequent voters who turn out as a result of the Democrats' rigorous get-out-the-vote efforts this year.

''For me the big story is how the methodological minutiae is so important, and inadvertently can create momentum for one candidate and a problem for another," said Dennis of Knowledge Networks. ''These seemingly arcane conversations are having an impact on the public perception of which candidate has momentum or not."

Despite the criticism from Democrats, pollsters say that since only votes that are cast actually count, it makes sense to at least try to weed out nonvoters. This process of doing so is subjective, though, and produces some of the sharpest disagreements between polling firms.

''There is no standard methodology in the polling profession for identifying likely voters," said Traugott. ''Each firm uses a calculus of its own. . . . This is part of the art of polling, rather than the science."

Polling researchers say that with so many firms and news organizations offering poll numbers produced with different methodologies, one of the dangers is in comparing their results to each other.

Many political sites average out poll numbers from different organizations that use different methodologies, which Traugott says is a bad idea. ''The problem occurs when people begin to compare results from one survey organization to results from another survey organization," Traugott said. ''What appears to be volatility is often actually in-house differences." Traugott suggests that at this stage of the campaign, polls are better indicators of trends in public opinion, rather than fixed predictions of share of the vote the candidates will win.

Despite potential flaws, pollsters say the public is increasingly showing a direct interest in their numbers, rather than being content to receive them through media filters. Gallup, for example, has started a blog to explain -- and, lately, to defend -- its methods to the general public.

''It's easier now for average citizens to read about polls," Newport said. ''In the past journalists did some vetting, but now readers can read almost any polls."

Pollsters also say that the media oversimplifies poll results, reducing them down to two numbers and removing some of the carefully worded caveats.

''I think we would all be better off if journalists had more time to write more thoughtful poll stories, rather than just rush to print with a story that says a poll says x," Newport said. ''The quickness of the news cycle sometimes makes the differences between polls look greater than they are."

Today (free)
Yesterday (free)
Past 30 days
Last 12 months
 Advanced search / Historic Archives