How do Exit Polls work ?

It sounds almost like you’re just asked who you voted for. I’m pretty sure that’s not the case.

Can someone please tell me how it works ?

Thanks.

You should have stopped with the first sentence after the thread title, you hit it on the nose. They ask after someone leaves the voting building.

Yep, it’s that simple.

The difficult bit is making a smple from a few polling stations applicable to the whole country - you need to find a demographic and politically-representative cross-section. An added difficulty is different people’s likelihood to respond to the pollster. In the UK, some polls are weighted towards the Conservatives, because Tory voters are more likely to refuse to answer the question. (The abscence of this measure was one of the reasons the 1992 polls were so very wrong.)

I haven’t seen anything specific about how pollsters are handling the early voting that is being allowed this year in 30+ states. Intuitively, those who vote early wouldn’t be representative of the whole, so they can’t be ignored. Anybody have any info on this?

They don’t just ask who someone votes for; they ask about their age, income, party registration, and so on. Exit polls are targeted at certain voting districts that may be a bellweather for the state as a whole, such as precincts that have a heavy registration for independents.

Once the data is sorted, one can get a pretty clear idea of who is turning out to vote and which way they are going. For example, lets say that the exit polls show that independents are voting for, say, Bush, that then has to be weighed against the turnout of Republicans. It’s not an exact science, but it can be quite accurate.

Part of the science of polling is to mix your sample between groups that have historically voted consistently and groups that have not.

So over the years, the pollster will have figured out that Precinct A almost always votes for one party. The pollster wants to interview people from that precinct to learn whether they’ve voted in the same percentages they always have, and whether voting turnout is as high as it’s traditionally been.

Then they’ll look at Precinct B, which almost always votes for the winner. They’ll ask mostly the same questions, but look at the answer through a different filter.

The good thing about exit polls is that they absolutely, positively only include actual voters. No matter how well designed, pre-election polls wind up with some people who change their mind at the last minute, or who say they will vote and then don’t.

This year, the pollsters are tearing their hair out over first-time voters and people who only have cell phones (pollsters don’t call people with cell phones because the person would have to pay for the call, which invalidates the response.) First-time voters, naturally, don’t have a track record and no one knows how many of them will actually vote. And for the first time, there are enough voters who only have cell phones (a lot of them also happen to be first-time voters) that it’s becoming more difficult to get a valid segment of that group. They’re hoping the exit polls will be able to show some sort of pattern with those people.