Election 2010: The dangers of exit polls

Video
The Washington Post takes a look back at some of the more memorable moments from the 2010 campaigns.
By Jon Cohen
Washington Post Staff Writer
Tuesday, November 2, 2010; 5:35 PM

It's unavoidable: Exit polls will be broadly misreported and overanalyzed Tuesday evening, leading to severe confusion about what's happening, and why.

Months of painful hand-wringing and hype cumulate on Election Day with an unstoppable, insatiable demand for information. Any information. Rain, for instance. Lines at the polls. A lack of lines.

Entering centerstage sometime after 5 p.m. will be a raft of early exit poll numbers. You might see them on Drudge or elsewhere, purporting to show who is winning. As anyone who contemplated a President Kerry midday Nov. 2, 2004, knows, these first numbers are not always good predictors of final results. They are not supposed to be.

Exit polls are terrific analytical tools on election night, and beyond. But they are not magically predictive.

At 5 p.m. when exit pollsters emerge from quarantine to share their numbers with subscribers - including The Washington Post - the data will include interviews only through the afternoon. In the case of California and other West Coast states, only morning interviews are included. Also, until results start to pour in after the polls close, exit polls are closely linked with pre-election surveys. If those surveys prove valid, there might be little issue. But if they don't, watch out.

The early exit poll numbers will provide important clues about the types of voters who are showing up, and preliminary information about the issues that are resonating. We will report on these nuggets, but not the potentially misleading estimates of who might be ahead in initial exit poll data. There's enough confusion out there already.


© 2010 The Washington Post Company