Will early voting skew exit polls?

Will early voting skew exit polls?

Will early voting skew exit polls?

Answers to your questions about the news.
Oct. 28 2008 4:47 PM

Will Early Voting Skew Exit Polls?

No.

More than 30 states allow no-excuse early voting, either by mail or in person, and this year one-third of the electorate is expected to cast a ballot before Election Day. (That's up from 22 percent four years ago.) Will all these trigger-happy ballot casters screw up exit polls on Nov. 4?

Nope. In states with lots of early voters—like Georgia, where one-fifth of registered voters have already recorded their decisions—the same outfit that conducts exit polls for the National Election Pool starts telephone surveys about a week before Election Day. Phone numbers are generated by computer with random-digit dialing, but only respondents who have already voted, or who plan to ahead of time, are questioned fully. Then analysts merge the data collected by phone with results from interviews conducted at polling stations, keeping the early/day-of ratio pegged to that of the actual vote total.

Advertisement

Estimates from absentee polls have been quite accurate historically. Oregon, for example, has had a vote-by-mail system in place since 1998, and pollsters haven't found big discrepancies between official results and survey answers. This year, some may worry that a Bradley effect will show up on telephone interviews but not in exit polls, since the latter are more anonymous. (For an exit poll, you just fill out a piece of paper and drop it in a box.) That would be the case only if the Bradley effect turned out to be real, something hotly contested during this campaign, and if the effect were the result of a conscious decision to dissemble rather than an unconscious one.

Got a question about today's news? Ask the Explainer.

Explainer thanks Mark Blumenthal of Pollster.com, Michael Dimock of the Pew Research Center, and Joe Lenski of Edison Media Research.

Juliet Lapidos is a staff editor at the New York Times.