Political Polling: Art & Science Merge in Many Variations

Pollsters, campaign strategists aim to stay within 'margin of error'

Campaign news stories driven by political polls tends to be looked down upon as "horse race" coverage.

But polls are necessities -- not luxuries -- for candidates, ballot causes and their political consultants who use results to help shape strategy.

They all caution against reading too much -- or too little -- into those surveys and related numbers.

Take, for instance, the Democratic Party's registration surge among new and young voters.

"What happens if those new voters decide, 'I don't need to vote, Obama's got it in the bag'?", says John Nienstedt, president of Competitive Edge Research & Communication, a leading San Diego polling firm. 

 "What if they don't know where their polling place is, or don't want to take that extra hurdle to get to their polling place?" Nienstedt added in an interview.  "Those sorts of factors can wreak havoc with anybody's 'likely voter' model."

Political polling can be a merger of objective mathematics, sociology and political science with the art of subjective interpretation.

Methodology matters; raw numerical results are "weighted and adjusted", analyzed for "social desirability bias" on the part of respondents who may be surveyed via "robo-calls" or live interviewers, "pushed" for answers or probed further to determine whether "undecideds" have, in fact, made up their minds.

"There's still a lot of undecideds out there, and a lot of even hard-core Obama supporters that may not go to the polls," says John Dadian, a San Diego political consultant.  "You dovetail that with one of the largest absentee blocs we've ever seen -- everybody's scratching their heads and trying to figure things out just a week before the election."

Some of the many issues surrounding political polls include: how closely does a survey's demographic sample mirror the actual electorate -- likely voters or not? 

And what to make of would-be respondents who refuse to cooperate, and hang up on surveyors -- or have already voted?

"Usually, we see that there's more of a drop off in cooperation among those people who have already voted," says Nienstedt.  "Those are issues a pollster needs to be aware of."

According to Nienstedt, exit polls have found that somewhere between five and 10 percent of voters who cast their ballots at precinct poling places make up their minds in the voting booth, or just before getting there.

As for exit polls, Dadian says the "refused to cooperate" factor is higher than for pre-election polls, because voters tend to be in a hurry to leave the precinct place -- "and they think they've already done their duty just by voting."

Either way, the professionals say polls can't quite be ignored -- or taken for absolute gospel.

"One of the dirty little secrets in polling that the pollsters don't like to talk about is, they never show you what  their "hang up" rates are," Dadian says.  "That could tell you volumes.  Now they use that internally to determine a lot of things.  but they usually don't make that part of the polling results."

Nienstedt responds by saying that if people who refuse to cooperate with a survey "are no different, or minimally different, from people who DO take part, it's not an issue -- as long as you do a random sample."

But how does a pollster KNOW whether or not they're different, or minimally different, he was asked.

"You don't," Nienstedt replied with a sly smile. 

So political polls may be quite educated, but there's still guesswork involved.

And much of what's unknown, scientifically, is left to artful imaginations. 

Contact Us