PERMANENT FATAL ERRORS; POSTCARD FROM STANFORD

LOUIS MENAND. The New Yorker.

"You can't imagine how much time it takes to lie on the floor in a fetal position," Mark Mellman explained to a gathering of political scientists and polling experts at Stanford a week after the Presidential election. Mellman is the C.E.O. of the Mellman Group and, until his services were rendered unnecessary by events, was the lead pollster for the Kerry campaign. He is a friendly and outspoken man, with the look of dishevelment that is associated with brainy enthusiasm, and it was entertaining to imagine him curled in a ball somewhere after November 2nd, trying to forget about a very bad night for John Kerry and for the art of polling.

One of the Bush campaign's leading pollsters was at the conference, too--Jan Van Lohuizen, the president of a group called Voter Consumer Research. Van Lohuizen is a figure of more businesslike mien, a man whose reaction to disappointment would not, one feels, include lying on the floor. He and Mellman had been on the radio together at 7 p.m. the night of the election, a time when there was every indication, from the exit polls, that the Presidency was going to change hands. At Stanford, Mellman admitted to "the cardinal sin," as he put it, "of gloating, and gloating early and inappropriately," on the air, and he apologized to Van Lohuizen. Van Lohuizen was gracious, in the spirit not so much of a good winner as of a professional who has been there, too, and who fully expects to be there again one day. The pollsters are like the pundits and the press: they are the survivors. Politicians come and go, but there is always another campaign.

"How did people decide for whom to vote?" was the question that the Stanford postmortem was organized to answer. There were, by November 9th, plenty of answers out there already, of course. Everybody had an opinion. You can't not have an opinion, and that's the problem. A consensus that George Bush won because voters cared about x easily turns into a consensus that voters were "sending a message" about x, and this determines the story line of the next four years. Ultimately, it can determine the three paragraphs on the election of 2004 in every American-history textbook of the future. But picking out the x, if there is an x, from the dozens of available reasons that Bush won and Kerry lost is a delicate piece of statistical calculation. This is why polling is an art.

Why did you order the cheeseburger? "Because I always order the cheeseburger" is an acceptable answer. "Because I felt like a change" is also acceptable. But those (or their electoral equivalents) are generally not considered good answers to the question "Why did you vote for John Kerry?" Most people feel that a civically responsible answer must name an issue. So they say something like "Because I am very concerned about the rising cost of health care." In fact, though, millions of people vote the way they do on the theory of "I always order the cheeseburger." This is not because of ignorance or indifference; on the contrary, many people who follow politics closely are partisans who choose the Republican or the Democrat no matter what. They have already picked the party, and they let the party pick the candidate. It's a reasonable and respectable labor-saving device.

And, in some elections, "I felt like a change" is the only reason many people have for voting against an incumbent. "Leadership, for a change" was a slogan that Bill Clinton campaigned on against George Bush in 1992. It worked, because Bush's approval rating went as low as thirty-three per cent--about the same as Jimmy Carter's in 1980, when Carter lost to Ronald Reagan. The younger Bush's approval rating was fifty per cent, which is low for an incumbent seeking reelection. But he was less vulnerable to a need-for-change strategy because this year, as Mellman explained at the Stanford conference, voters reported that they wanted stability. Kerry therefore tried to work "change" into his message without using the actual word--for example, he called for a "new direction." Evidently, the voters were not fooled.

The potential story line that floated fastest to the surface after the election was that voters were sending a message about moral values. This is a little strange, because the "moral values" peg is entirely an artifact of the very exit polling that led Mellman to his brief and illusory moment of Election Night triumph. Those polls, as everyone knows, were a fiasco. Exit polls conducted by the National Election Pool (a consortium created by the networks, CNN, and the Associated Press to replace the service that was responsible for the disastrous miscalls of the last Presidential election) and the Los Angeles Times, along with election predictions by Zogby International, showed Kerry pulling away. CNN, on its Web site on Election Day, posted exit polls that, according to Steven Freeman, at the University of Pennsylvania, had Kerry winning Ohio by 4.2 percentage points (he lost by 2.5 points), Pennsylvania by 8.7 percentage points (he won by 2.2), and Minnesota by 9 points (he won by 3.5). At 5:30 p.m. on Election Day, Zogby's web site had Kerry with a projected total of three hundred and eleven electoral votes, a blowout. (Kerry ended up with two hundred and fifty-two electoral votes to Bush's two hundred and eighty-six.) At 8 p.m., journalists were looking at national exit polls showing Kerry winning the popular vote by three points. And even after the polls had closed in Virginia the networks designated the race there too close to call, based on the exit data. Bush won Virginia by nine points.

Traditionally, polls that ask people whom they just voted for have been regarded as about as sound as polls can get--which is why their failure this time gave so much anguish to Kerry supporters. People don't forget, in the few minutes after leaving the booth, whom they chose; they have little reason to lie; and their answers aren't affected by the I-was-at-Woodstock syndrome, which leads more people to claim, after they know the outcome, that they voted for the winner than actually did. But polls that ask people why they voted for a particular candidate have usually been regarded with a heavy drip of saline solution--not because people do not tell the truth but because they often don't know the truth. If a poll has trouble naming the winner within the margin of error, how reliable are its answers to the question why people voted the way they did?

The National Election Pool questionnaire asked people leaving the polls to pick "the one issue that mattered most in deciding how you voted for president." These were the reported results:

Education 4%
Taxes 5%
Health care 8%
Iraq 15%
Terrorism 19%
Economy/jobs 20%
Moral values 22%

Eighty per cent of the respondents who picked "moral values" as the issue that was most important to them voted for Bush, but even the Bush pollster dismissed the significance of the poll. The "moral values" number, Van Lohuizen said at Stanford, is "entirely determined by what else is on the list." Voters weren't asked to name an issue that mattered to them; they were asked, in what is known in polling as a "closed-ended question," to pick one answer out of seven. "And, if you look at the list, there are a lot more places for a Kerry voter to park himself than for a Bush voter to park himself," Van Lohuizen said. "That's point No. 1. Point No. 2: if you give people a list of seven and you ask them what's their top concern and the highest number is twenty-two, that means there is no consensus. It means that there was no one issue that drove the election."

The belief that the issue of "moral values" was somehow decisive is tied to the belief that a greater proportion of voters this year were highly religious. This, too, is a belief unsupported by the data. "As a conservative, you love to see the liberal media twinge and say, 'Oh, my God, it's these moral values,' " Van Lohuizen said. "It was an important factor, but I have seen no data that it was more important in '04 than it was in 2000 or '96. I've seen no data that, in the composition of the electorate, the religious voter was more heavily represented." More churchgoers turned out to vote in 2004 than in 2000, but only because more people turned out to vote.

According to a report on Beliefnet, a religious Web site, by Steven Waldman, the editor of the site, and John Green, a political scientist at the University of Akron, forty-three per cent of Ohio voters in 2000 attended church once a week or more often. In 2004, the figure was forty per cent. In Florida in 2000, forty-one per cent of voters were regular churchgoers, but only thirty-five per cent were this year. Bush even seems to have won a greater percentage of voters who never attend church than he did in 2000. "Moral values" has been tracked as an issue in Presidential elections at least since 1992. Many voters care about moral values, but there is no reason to conclude that more voters care about them now than ever. At Stanford, Kathleen Frankovic, the director of surveys for CBS News, pointed out that in the Los Angeles Times exit poll voters were asked to identify, from a closed list, the issues that were most important to them. In 1996, the year Clinton was reelected, forty per cent chose "moral/ethical values"; in 2000, thirty-five per cent chose it. In 2004, it was back to forty per cent. And a recent report from the Pew Research Center shows that on open-ended surveys, which allow respondents to name any issue that matters to them, items associated with moral values finish behind the economy and the war in Iraq.

The phrase "moral values" is open to interpretative license. Peace and social justice are moral values; they just happen not to be values associated with the Bush Administration. Most commentators assume that voters understood the phrase as code for two issues: gay marriage and abortion. So-called "marriage initiatives," designed to prevent same-sex marriages, were on the ballot in eleven states and carried easily in all of them, even Oregon, where fifty-two per cent of the voters voted for Kerry and fifty-seven per cent voted against gay marriage. (Kerry, after all, is not a supporter of gay marriage.) Still, there was agreement among the experts at Stanford that the presence of a marriage initiative on the ballot did not measurably increase turnout for Bush, and did not help him win states that he otherwise would have lost. If the National Election Pool's list had offered "gay marriage" instead of "moral values," it is hard to imagine that twenty-two per cent of the respondents would have pointed to it--an assumption that seems to be confirmed by the Los Angeles Times exit poll, where voters were asked specifically about gay marriage and abortion. A smaller proportion of voters picked those as important issues than picked "moral values" on the National Election Pool questionnaire, and they split almost evenly between Bush and Kerry.

"Why did President Bush win this election?" Gary Langer, the director of polling at ABC News, said at the Stanford conference. "I would suggest that the answer can be expressed in a single phrase: 9/11." No one there disagreed. "Fifty-four per cent of voters on Election Day said that the country was safer now than it was before September 11, 2001," Langer pointed out. "And perhaps, I would suggest, more important, forty-nine per cent of voters said they trusted only President Bush to handle terrorism, eighteen points more than said they trusted only John Kerry." He went on, "Among those who trust only Bush to handle terrorism, ninety-seven per cent, quite logically, voted for him. Now, right there, if forty-nine per cent of Americans trust only Bush to handle terrorism and ninety-seven per cent of them voted for him, those are forty-eight of his total fifty-one percentage points in this election. Throw in a few more votes on ancillary issues and that's all she wrote." Langer thinks that a key statistic is the change in the votes of married women. Gore won the women's vote by eleven per cent; Kerry won by only three per cent, and he lost most of those votes among married women. Bush got forty-nine per cent of the votes of married women in 2000; he got fifty-five per cent this year. And when you ask married women whom they trust to keep the country safe from terrorists fifty-three per cent say "only Bush." (The really salient demographic statistic from the election is one that most Democrats probably don't even want to think about: If white men could not vote, Kerry would have defeated Bush by seven million votes.)

Assuming that the election did come down to a referendum on terrorism, there was very little that a Democrat could have done to win it. Kerry could not change the subject; war and terrorism were in the news every day. According to Mellman, polls showed that although only thirty-three per cent of voters thought that the invasion of Iraq was worth it, fifty-two per cent thought that it was the right thing to do. Those are tough numbers from which to devise a campaign strategy. They are the numbers behind Kerry's notorious trouble in parsing the matter of his own position on the invasion--his attempt to criticize the outcome but not the decision. It's getting hard to remember now, when Iraq has become a violent and ungovernable mess, but the invasion of Iraq had overwhelming public approval, and people don't like to admit that they were wrong. Neither does the President. It's one of the attributes that voters seem to identify with.

Of course, it doesn't matter what the science of public opinion concludes. It only matters what the politicians conclude. If Democrats believe that the lesson of the election is that the Party needs to move to the right, then, if it moves, that will be the lesson. It might be wiser for the Democrats to chalk Bush's reelection up to 9/11 and stick to their positions. The Democratic candidate did not lose votes in 2004: Kerry got five million more votes than Al Gore got in 2000, when Gore won a plurality. Unfortunately for the Democrats, Bush got nine million more votes than he did four years ago. But it wasn't because the country moved to the right. The issue that seems to have permitted an incumbent with an unimpressive approval rating to survive reelection was not an ideological one. The country did not change radically in the past four years. Circumstances did.