Stop blaming exit polls for Election Night problems.

Stop blaming exit polls for Election Night problems.

Stop blaming exit polls for Election Night problems.

The conventional wisdom debunked.
Nov. 6 2002 7:42 PM

Don't Blame the Exit Polls

They didn't cause Election Night problems on Tuesday or in 2000.

Since the 2000 election fiasco, exit polls have been singled out as the chief villain in television's Election Night coverage. The crash of Voter News Service's exit poll computers on Tuesday only confirmed the bad reputation of exit polls. But let's get something straight about what went wrong on Tuesday and two years ago. In spite of what you may have read or heard, exit polls had hardly anything to do with either disaster.

Advertisement

The networks called Florida for Al Gore two years ago not based just on VNS polls of voters but largely on real election returns as well. When the networks reversed themselves late in the night and called Florida for Bush, exit polls were again irrelevant. The second call was also based on real votes, close to 6 million of them. Even so, exit polls have been front and center in the soul-searching about Election Night reform. (The source of these exit polls—and most other Election Night data—was and remains VNS, which is jointly owned by the networks and the Associated Press.)

So Tuesday night, three broadcast and three cable networks, having declared their commitment to reform and having put up more than $10 million dollars for improving VNS, put the 2002 model on display. Most of the money was spent to rewrite the computer programs for counting votes and estimating winners. The system in use in 2000 was old and creaky, and the rewrite was long overdue. By midsummer, however, it became obvious that the company being paid to write the new programs was far behind schedule. When VNS officials and the network managers were asked about this on the record, they offered only serene confidence that the new system would work. Off the record, fear of another disaster was easy to find.

On Tuesday night, VNS realized its spanking-new operating system didn't work as well as it hoped. Like a crew facing a possible shipwreck, it looked for cargo to throw overboard. The most obvious thing to jettison was some of the exit polling data. It takes a lot of computer hardware and operators to process the huge amount of data—why did black women in Arkansas over 35 vote for that candidate—that goes into an exit poll. Dumping that data would allow VNS to keep churning out basic horserace numbers from the polls.

For polling analysts at the networks and academics who thrive on postelection studies of the exit polls, this may be a huge loss (though one network executive is pretty sure the data can eventually be retrieved for the cause of scholarship).

But that lost data did not have any impact on the fundamental question of calling the election. For the purpose of declaring winners and losers, the networks appear to have had all the data they normally have.

That is not, however, the impression you got in reading the morning papers. Writing of "TV's Slow Motion Election," Howard Kurtz in the Washington Post recounts the dismay of network anchormen because a "computerized fiasco had blown up their exit poll data. They looked like their favorite toy had been taken away." The New York Times announced the return of an old "Election Ritual-Awaiting votes."

The networks may have been a bit slower than in the past to call close races, but that had nothing to do with a breakdown in technology. It had everything to do with a breakdown in the decades-old zeal to be the first to "call" winners. Not everyone in TV news management has cared that much in the past about being first, and since 2000, those who don't want the aggravation (and don't want to get it wrong) are in the ascendant.

More important, exit polls would never have been used to call any of Tuesday's tight contests. Given the large sampling error in the surveys, no one would ever consider calling an election on an exit poll reading that was closer than seven points: Five Senate contests were decided by four points or less. "If we had exit polls that were the most perfectly designed and executed possible," says Bill Wheatley, the NBC vice president who sits on the VNS board of managers, "it would not have enabled us to announce the shift in control of the Senate any sooner that we did last night."

Martin Plissner is former executive political director of CBS News and the author of The Control Room—How Television Calls the Shots in Presidential Elections.