Ed Morrissey calls them “worthless”:
Exit polling data gets collected all day long to find the eventual turnout model for elections, especially in demographics such as age, gender, ethnicity, affiliation, etc. That data only becomes valid when it is fully compiled. Partial data sets for exit polling do not provide predictive outcomes because the turnout models can change significantly during the day, perhaps especially because of early voting. That is exactly what happened in 2004, when media outlets used non-predictive data in predictive ways, and while the data sets were still being compiled.
That isn’t to say that completed exit polls are meaningless. The networks will use the data in part to plug into their election models in order to call races — but that takes place while the results of actual voting are being published, after the polls have closed.
Nate Cohn chips in his two cents on exit polls:
They’re not designed to measure the results perfectly or measure the composition of the electorate. I find myself surprised by how just how accurate the exit poll figures can be, despite the obvious issues with the raw responses and the inability to weight to population targets. Unfortunately, most analysts and reporters jump on the surprising, outlying, newsworthy findings. Often, those figures are the ones most likely to be wrong.
Dana Lind identifies another problem with exit polls, their “tendency to oversample a particular kind of voter of color — the kind who lives in majority-white areas”:
Even though the public doesn’t know exactly how the exit poll chooses where to go, it’s possible to make some educated guesses. The exit poll is trying to predict the margin of victory for one candidate over another across the state. So when it decides which polling places to put interviewers outside of, it’s reasonable to assume that it’s choosing lots of swing precincts — precincts that are harder to predict and likely to affect the outcome. Those are going to be largely white precincts. …
Here’s why this is a problem: the voters of color pollsters run into in majority-white precincts might not be representative of the voters of color across the state. In particular, according to Latino Decisions, voters of color living among whites are “more assimilated, better educated, higher income, and more conservative than other minority voters.”
The Dish 