The approaching end of the primary season prompts a terrific question: What the heck happened?
You’ll hear answers from a lot of sources. But a beautiful thing about the internet is that you can also dig out some of your own conclusions – or fact-check what you’re hearing from others – by going directly to the source. With that in mind we’ve posted full exit poll results from each state contest in which they were conducted. Click here to get there.
We’ve been dissecting these polls in detail all year; they’re an absolutely invaluable source of intelligence on the election outcomes. Our fundamental understanding of the contours of the Democratic and Republican nominating contests alike relies on them.
There’s some apparent misunderstanding of how these polls are done. Interviews are conducted in person at randomly selected polling places in each state; voters leaving the polling place are randomly selected to participate. Interviewers note the sex, race and estimated age of those who decline; these are used in a non-response adjustment. In states where it's estimated that about a quarter of the vote, more or less, will be by absentee ballot, telephone polls supplement the exit polls.
There can be differential non-response in exit polls; for example, in many of this year’s Democratic primaries Barack Obama's supporters have been disproportionately apt to participate. In addition to the sex, race and age adjustment, the exit poll is adjusted for non-response in two other ways: Preliminary results are partially weighted to an estimate derived from pre-election polls, and later results are weighted to the actual vote as it becomes available.
One use of the exit polls is to assist efforts by news organizations to project winners. Another use – the one I find most valuable – is to enable us to make an independent, valid and reliable assessment of who voted, how and why. You've seen our analyses. Have a look at the data.