2000 and, to a lesser extent, 2004 election nights were big debacles for exit pollsters. Will 2008 be any different? Joe Lenski, co-founder and executive vice president of Edison Media Research, will conduct the exit polling for the TV news networks and the AP this year. He discusses how the process has changed.
BROOKE GLADSTONE: From WNYC in New York, this is NPR’s On the Media. I'm Brooke Gladstone. BOB GARFIELD: And I'm Bob Garfield. Right now we're bracing ourselves for another exit poll debacle. It’s been a terrible decade for exit pollsters. In 2000, shame rained down on the now-defunct Voter News Service when it called Florida for Gore. In 2004, leaked poll data showed Kerry surging – until he lost.
This year, the exit poll data will come from the National Election Pool, a consortium of the five major networks and the Associated Press, each of which will have people sequestered with the incoming information.
Joe Lenski is one of the founders of Edison Media Research which, along with Mitofsky Research, will conduct the polls. Lenski says they figured out how to cure the two biggest problems plaguing exit polls – one, the over-counting of Democratic voters because young Democrats are more responsive to the usually young exit pollsters and, two, all those awful leaks. JOE LENSKI: The first thing we did, starting in 2006, the news organizations established what they call a quarantine room. And that means there are three representatives from each of the six news organizations in that room and they are the only people allowed to review the exit poll data before 5 p.m.
Those people give up any ability to communicate to the outside world, so their cell phones, their BlackBerrys, any other communication devices are confiscated, and their jobs are to review the data before it’s released to the rest of the news organization at 5 p.m.
That has helped, both in 2006 and the 2008 primaries, to keep a handle on the distribution of the exit poll data until much closer to poll closing time. BOB GARFIELD: And the overrepresentation of Democrats problem? JOE LENSKI: We've done several thing here, mainly in the hiring and training practices of the more than 1,000 interviewers that we have around the country in all 50 states upon Election Day. We found that older voters were much less likely to fill out a questionnaire that was handed to them by younger interviewers.
So we have increased the average age of our interviewers, and we find that we get a higher response rate from older voters when the interviewer is older, as well.
And our interviewers are better trained to select voters as they leave the polling place in a random way, so that that sample best reflects the voters at that polling place. But we always have to keep in mind this is a survey like any other survey, and about 50 percent of the people that we approach still refuse to fill out the survey. BOB GARFIELD: I want to ask you about the Bradley effect. If it is true that prospective voters misrepresent their intentions to pollsters before they vote, because they don't want to portray themselves as racists, is there any reason to think that the Bradley effect wouldn't apply for exit polling, as well? JOE LENSKI: It shouldn't apply, mainly because these are self-administered questionnaires. Our interviewer basically hands the questionnaire to the respondent and the respondent confidentially marks their responses, folds the questionnaire and puts it in a ballot box outside the polling place.
We have never found any evidence that voters who fill out our surveys are lying to us about any of their answers. BOB GARFIELD: This time around there were a number of factors that could frustrate you. In some states, as many as a third of the votes will be cast before Election Day.
JOE LENSKI: Let me correct – that one-third number you quote is probably a national number. There are many states where the percentage of voters voting early is well over 50 percent now.
What we do in those cases is we do telephone surveys of early voters and we merge that data with our exit polls on Election Day. So that’s how we deal with the early voters. BOB GARFIELD: Wow, never mind 5 p.m. on Election Day. Do you have enough data to project right this minute? JOE LENSKI: [LAUGHS] No. That data isn't even put in our system until Election Day. It has the same quarantine practices as the exit poll data.
BOB GARFIELD: So if 2000 was the ultimate nightmare for people like you actually giving the nod to the [LAUGHING] wrong candidate, what is your nightmare? What can happen to just ruin your Election Day? JOE LENSKI: Basically, not having a clear decision. There are many states where people fill out provisional ballots, and those ballots aren't counted on election night. There are many states that don't even count the mail-in ballots until days after the election.
If we have an election that’s really close in key states, we just might be sitting there without a clear-cut result for several days. BOB GARFIELD: Hmm, so you don't professionally care who wins but you’re hoping for a landslide. JOE LENSKI: That does make our job a little easier, yes. BOB GARFIELD: [LAUGHS] All right, Joe. Thank you so much. JOE LENSKI: Okay. Take care. BOB GARFIELD: Joe Lenski is the co-founder and executive vice-president of Edison Media Research.