By Michael D. Cohen, Ph.D
Exit pollsters are getting a well-deserved beating in light of the third straight fiasco in as many elections. But it is high time to leave the critiques of the individual pollsters aside and focus on ditching the early reporting of exit polling results, if not getting rid of the exit polls altogether. In a perfect world, exit polls would be used by news organizations to explain the vote rather than to predict the outcome on Election Day. We do not live in that world.
Mistakes Times Three
In 2000, Voter News Services, a group brought together by the major networks and other news organizations, failed spectacularly in Florida resulting in the early declaration of victory for George W. Bush. In failing to make the right call – that it was too close to call – VNS became the story rather than the election it was supposed to be analyzing. Networks and some newspapers jumped to the conclusion that Bush had won before the recounts showed that it was well within any survey margin of error. VNS was wrong but news shops were wrong with them.
In the 2002 mid-terms, VNS exit polling computers melted down, resulting in no data to report because it was so unreliable. The network talking heads were clueless and somber, like kids unwrapping boxes of presents filled only with tissue paper. No one knew what was going on and, frankly, that was a good thing. Undecided voters could not be influenced by Election Day momentum because no one knew who was winning and why. People voted, candidates won and lost and Democracy survived. Later when the data was released it was deemed no longer newsworthy by news editors. We understood why people voted but were not influenced at all the day of voting. Exit polling nirvana in my opinion.
Finally in 2004, Edison Media Research and Mitofsky International teamed up for what should be the final go-around for exit polling. Willful leaking and early reporting suggested that Senator John Kerry was going to run away with the election when in fact, George W. Bush won a majority on Election Eve. Fortunately, bloggers were way ahead of the curve questioning the male/female split in the data and other funky characteristics of the data being reported. Today, Mitofsky is the subject of columns by other pollsters and consternation by the public opinion research community.
The Aftermath
Reputations have been ruined or are in the process of ruin. It took two rounds to knock out VNS. Murray Edelman, who led VNS, turned on his network partners in early 2003 for making his shop the “scapegoat” for the exit polling problems.
The networks dissolved VNS and Edelman and company have since found other work. In truth, they were both at fault – Edelman for under-delivering an improved system from 2000 to 2002 and the networks for misusing exit polling data to try to predict the outcome on Election Day.
Today's punching bag is Warren Mitofsky who David Hill (a pollster) described as the “dealer of Election Day crack cocaine." I am sure that the pile of Mitofsky hate mail from the Right is rising. He might be getting more business inquiries from Democrats who believe he was right that Kerry won. Even if Mitofsky gets more business he still loses “street cred” in the polling business. The professionals have turned on him.
Neither of these two outcomes includes what should have been widespread discussion: the validity of exit polling. In short, how honest are we when we leave the polling booth? Is it possible to get a sample of voters that reflects the actual outcome on Election Day? There are dozens of academic articles and Wednesday morning pundits who would argue that it is impossible to get an accurate read the day of the election.
Voters are notoriously skittish about talking to anyone as soon as they leave the polling booth. There is some evidence that candidates from one side or the other want to talk while others want to be left alone. It is impossible to quantify the difference since each election is different but we know it to be true. Anti-Bush enthusiasts were apparently more likely to voice their votes for Kerry while the pro-Bush “quiet majority” just wanted to get to their cars.
What to Do Next
You can bash the exit pollsters or the methods they use but it is not very useful. While it would be easy to pass an unofficial “three strikes law” and strike out exit polling altogether, the more difficult and correct way to go forward is to limit reporting. Self-limits on the reporting of exit polling data would also save the reporters and talking heads from themselves.
Discretion is something that the media could use a lot more of lately. Clear thinking would have saved the New York Times from a reporter writing fiction and tipped off Dan Rather to problems with a high-profile story. In a world where bloggers sometimes get it first but wrong, it would be nice if our elite media raise the bar. It takes leadership, something the elite media should embrace.
This is how the elite media should do exit polling:If the elite media cannot commit to these five principles then they should not do exit polling at all. They do more damage by reporting only part of the story than the truth of what is really happening on Election Day.
-
Exit polling data should not be reported to the public at all before all election polls are closed – including Hawaii. This includes keeping the issue priorities and demographics quiet.
-
The right time to release results would be only after a candidate has conceded the election. Candidates know when they have won or lost.
-
Exit polling data should only be used to talk about why people voted and which groups voted one way or another.
-
Exit polling data is a great day-after/moment-after concession story.
-
Only hire an exit pollster that is committed to all of these.
Michael Cohen is President of Cohen Research Group of
Washington, DC. He can be
reached at Click here to contact this Author.
