Every election, you get bombarded by data points from exit polls dissecting what issues voters said mattered, what share of the vote came from certain demographic groups and how those groups voted. Now, exit polls are useful in that they attempt to reach people who actually voted, so their results can be valuable when it comes to understanding trends and the importance of various issues. However, we’re here to caution you to take a wait-and-see approach with exit poll data tonight.
For all the attention they get, exit poll data is best used after the election is over. That’s because exit poll data is reweighted after results start coming in to better match the vote percentages from the location they were sampled and more accurately reflect voter preferences. As the night wears on, you may see the exit poll figures change in meaningful ways as further reweighting occurs. Exit polls will stop changing only after we have most or all of the results, but that’s typically very late on election night or even a day or two later, especially in an era with more mail voting. As a result, exit poll numbers — particularly the early results — are not a good measure of a candidate’s support, either in the overall race or among a certain demographic.
Moreover, an exit poll is still a poll, so there is some degree of potential error from the sampling process, including the risk that certain groups of voters will be less likely to respond to an exit poll worker at a sampling precinct. And while exit polls aim to talk to actual voters, the rise in alternative forms of voting, such as voting by mail or early in-person, has complicated how exit polls are conducted — and raised questions about how well they can gauge the actual makeup of the electorate. For instance, FiveThirtyEight largely avoided using exit poll data from vote-by-mail states like Colorado in the 2020 Democratic presidential primary because there was no way of being certain that respondents, reached by phone, had actually voted. And in the 2020 presidential election, the national exit poll may have exaggerated the share of the electorate that was nonwhite, which has been a long-term challenge for exit polling.
This is not to say that exit polls don’t have value — they do! But they’re most useful after the election, not on election night.