If political campaigns are horse races, then consider public opinion polls one way to set the odds.

But campaigns create and use polls for much more than the neck-and-neck numbers you hear on the news.

Maybe you’ve already gotten one of these calls this election season: asking you to “press one” if you’d like to vote for such-and-such a candidate, or “press two” for another.

Public polling is a voter’s chance to weigh in. But what happens with this information - and exactly who is behind all this polling?

Gregg Durham heads up the suburban Oak Brook-based We Ask America polling, which has done work for politicians, news outlets and interest groups in Illinois and around the country.

It’s his job to call up registered voters - some 12 million in 2012, Durham says - and take their temp on the candidates and issues of the day.

Good audio, believe or not, is important, lest people hang up. And asking questions in a specific order, as not to taint the polling pool, is key.

Getting people to stay on the phone has become a pretty big part of our democratic process. Public opinion polling isn’t just used to predict who will win an election. It oils the modern campaign machine, helping it test different talking points, and form the messages most likely to influence voters on election day.

But all of that depends on the accuracy of the poll.

Durham points to Illinois’ super-tight 2010 Republican gubernatorial primary, where State Sens. Bill Brady and Kirk Dillard were neck-and-neck near the end.

“I was Mr. Dillard’s pollster, and I had to make that call and say, ‘You’ve got a problem here. This guy’s catching you,’” Durham said. Durham predicted then that the election would be within 200 or 250 votes. Brady ended up winning by 193, only to narrowly lose the general election.

So how do pollsters get so dead-on?

Tom Bowen, a Democratic campaign strategist, says the absolute most important thing for accuracy is that the sample in the poll mirrors the make-up of the larger electorate - ideally, of the people who will actually vote on election day.

“Think about how a pond would look with a bunch of fish in it,” Bowen said. “If you grabbed a whole bunch of fish out of the pond, you’d have a pretty good idea of what the fish look like.”

But, Bowen explains, that would not be a statistically accurate sample, “because some fish are on the bottom. Maybe they’ve just eaten and are resting, and some fish are hiding.”

So before they blast out any phone calls, pollsters spend big money on demographic data to learn as much as they can about voters, based on where they live: whether they rent or own, whether they have health insurance or enjoy going to the movies.

After the poll, they run their results through a complex math equation to account for the inevitable imperfections in the sample. This process, called weighting, accounts for the over- or underrepresentation of certain folks who happened to answer the phone.

But the trophy for campaigns is not the horse-race number they may release to the public. It’s the drilled-down data the rest of us usually don’t get to see - the stuff that’s used to craft the all-important campaign message.

“It’s not about telling a voter something you want them to know. It’s about reminding them about something they already know,” Bowen said.

For example: In 2009, Bowen was running the congressional campaign for County Commissioner Mike Quigley, when he saw some surprising poll numbers.

They showed voters didn’t really recognize Quigley by name, but they did recognize County Board President Todd Stroger - and they didn’t like him.

So Bowen put together an ad that touts Quigley as someone who had been “taking on” Stroger and his unpopular penny-on-the-dollar sales tax increase.

“One thing you’ll notice about that ad, besides the fact that Todd Stroger was right in the front of it, was that Mike Quigley’s name was used six times,” Bowen said. “So in order to stand out, this was sort of what the poll told us to run.”

Quigley won handily. But sometimes, winning means knowing what not to talk about.

‘Explaining is losing’

Democratic Campaign strategist Terrie Pickerill recalls a race where her candidate (she declined to name them) was late in paying property taxes, but the opponent had some ethical problems of his own. So she polled to see which would hurt more.

“People just didn’t care as much about just paying property taxes on time, but they really cared that this guy had ethical issues,” Pickerill recalled.

So when her client was attacked over the property tax thing - and wanted to explain it by holding a press conference - she told them to stay quiet.

“It’s like, ‘Look at the poll!’ This is much worse for him than it is for us,” she said. “Explaining is losing, so what we wanna do is say, the real issue is his ethics.”

But there are also ethical issues for the pollsters, says Jason McGrath, a Democratic pollster who’s worked for Chicago Mayors Rahm Emanuel and Richard M. Daley, among others.

“Good pollsters don’t tell a candidate what to say,” McGrath said. “The political graveyard is scattered with failed candidates who try to be something they weren’t. And it’s not in our interest to use a poll to tell somebody to be something they’re not.”

McGrath says voters can sense when candidates are faking it. And dishonesty  doesn’t poll very well.

Alex Keefe is political reporter at WBEZ. You can follow him on Twitter and Google+.