THE POLLSTER; THE POLITICAL SCENE

LARISSA MACFARQUHAR. The New Yorker.

Shortly after his fifty-sixth birthday, several weeks ago, John Zogby, the pollster, could be found at seven-thirty in the evening walking rapidly back and forth among the brightly lit shops in the Copley Place Mall, in Boston, trying vainly to locate the skybridge to the Prudential Center. He had been told to "walk through retail," but, amid the confusing display of stores leading in every direction, he was finding this instruction insufficiently specific. He was dressed up in a tie and jacket, and he walked with his shoulders hunched over and pointed forward, as though he were trying to prevent a cloak from falling off. He seemed, as he usually does, mild, overcaffeinated, inquisitive, watchful, cautiously friendly, somewhat anxious, yet fundamentally optimistic.

He had been hired to address the American Road and Transportation Builders Association, one of his polling clients, and worried that he was late, but when he arrived the road builders were still in the middle of a preprandial auction (a bird hunt, a golf getaway), and so, after securing himself a plastic glass of wine and a plateful of cheese cubes, Zogby retreated to a distant table to await the commencement of dinner.

"I can't help but remember the early years," he told the road builders when it was time for him to speak. "One Friday, I was at a happy hour and ran into an old chum from high school. I told him I was a pollster, spread the word. Lo and behold, first thing Monday I got a call from his aunt, and she said, 'You're a pollster?' and I said yes, and she said, 'Well, I have a sofa and a chair.'

"Welcome to the Armageddon election," Zogby said when the laughter had died down. "We have divided ourselves into two equal warring nations. We did a poll last December on this fiction we've created, red states and blue states. It's a pure fiction, because Florida's a red state by a few hundred votes, and New Mexico--a couple of hundred votes made it blue. But listen to what we found. Fifty-four per cent of red-state voters said they attend a place of worship at least once a week--that's a very important conservative voting indicator--thirty-two per cent of blue-state voters said they did. Seventy-five per cent of the reds said they want a President who believes in God, fifty-one per cent of the blues. Fifty-six per cent of red-state voters said they keep a gun, thirty-five per cent of blue-state voters said so. In a blue state, you are seven points more likely to be single, never married, and, let me tell you, we talk about a gender gap in politics, but it's minuscule compared to the married-single gap." In fact, Zogby has since discovered in a poll of women that the gender gap in this election has disappeared altogether. (He also discovered that women pick Mel Gibson's "The Passion of the Christ" over Michael Moore's "Fahrenheit 9/11" by a nearly two-to-one margin as a movie they'd like all Americans to see.) "On most issues married and single voters can be twenty-five or thirty points different, married ones being more conservative," he continued. "So that's the context in which we have this election."

John Zogby has been polling for two decades now, but he made his national reputation eight years ago, after the Clinton-Dole race of 1996. Most pollsters predicted a double-digit victory for Clinton, while Zogby predicted a narrow margin of eight per cent--the actual figure was 8.5 per cent. Zogby's uncanny accuracy, combined with the dramatic failure of the other polls, won him a lot of attention. In 2000, he repeated his performance, being one of only two major pollsters to predict that Gore would win the popular vote. He has had his share of humiliations, too, of course. When Hillary Clinton was deciding whether to run for the Senate in New York, he wrote an Op-Ed piece for the Times explaining why she wouldn't win; his final poll predicted a tie, when she actually won by twelve points. Nonetheless, this year he is conducting national election polls for Reuters and the Wall Street Journal, and smaller polls for the Miami Herald and the Atlanta Journal-Constitution. Nor is his ambition confined to the United States: he has polled in the Middle East, Europe, Africa, and South America. His group was one of the first to publish survey results from Iraq. Zogby wants to be the Gallup of his generation--the brand name in polling all over the world. Lately, in fact, he has derived some satisfaction from observing the wild vacillations of the Gallup election polls--Bush thirteen points ahead in mid-September, eight points ahead at the end of the month, dead even with Kerry on October 4th--because he finds it implausible that such enormous swings reflect actual changes in public opinion. "I mean, good debate, but it wasn't that good," he says. He concludes that Gallup's polls are less reliable than his own. "Zogby is more volatile, but Zogby polls are less volatile," he says. It's all a matter of technique.

"How do I get a handle on this election or any other?" he asked the road builders. "I asked one question the Saturday before the election in 2000. I called my call center in Utica and said, 'Put this in the poll: "You live in the land of Oz, and the candidates are the Tin Man, who's all brains and no heart, and the Scarecrow, who's all heart and no brains. Who would you vote for?" ' The next day, I called Utica and said, 'Whaddaya got?' They said, 'Well we've got Gore--,' I said, 'I don't care about Gore. What's Oz?' It was 46.2 for the Tin Man and 46.2 for the Scarecrow. It was right there that I knew I wasn't going to know what was going to happen. But I asked this question again two weeks ago and the Tin Man led by ten points."

Utica, New York, where Zogby grew up and now has his headquarters, is a small, depressed, formerly industrial city about fifty miles east of Syracuse. Zogby's office is situated in a particularly desolate part of town, full of vacant lots, boarded-up buildings, and closed-down pizza parlors. The office is on the third floor of a giant old factory building, on a road of what used to be textile mills. It shares the place with several small businesses--the Trim-A-Lawn Corporation, Variable Speed Motor, U.S.A. Sewing. The building's ancient elevator still has a sign in it instructing its patrons how to push the buttons by themselves, along with a newer, paper sign warning that employees caught writing on its walls will be immediately dismissed.

Most of Zogby's office space is taken up by the call center where the polling is done--a large, low-ceilinged room divided into rows of cubicles, each of which is equipped with a phone and a computer. There are no personal items in the cubicles, because they are occupied by several shifts of callers, which start at nine in the morning and continue until it is evening on the West Coast. The day shift at the call center consists mostly of elderly women (the oldest is eighty-six), who dress in casual pants and tops and drape cardigans over the backs of their chairs for protection against the air-conditioning. Most seem to like their work, chatting with their respondents in a grandmotherly way and commenting on their answers.

Zogby's own office has the look of a rec room from the seventies, with wood panelling and wall-to-wall blue carpet. Above his desk he has hung a large photograph of the New York City skyline at night; nearby hang his college diploma and charts detailing "Leadership Succession in the Middle East since 1945" and "Middle East Selected Socioeconomic Indicators." Over the years, Zogby has often asked himself why he stayed in Utica. Sometimes he says it's because living there keeps him in tune with what real Americans are thinking. "If folks aren't talking about it when I pick up the paper, it's not a real issue," he says, citing as an example the Gary Condit scandal, which nobody in town cared about. But he will also say that he is there because life gave him lemons and he made lemonade: his wife is from Utica, too, and didn't want to leave her family, and it was a good town to raise kids in. (They have three sons.) Whatever the reason, living in this far-from-Washington place has reinforced his sense that he is an outsider in the polling business.

Zogby thinks of himself as a natural maverick who stands outside the clubby world of the other pollsters because he finds it pompous and stuffy and because he isn't a joiner anyway. But it is also true that he uses techniques that are frowned upon by aapor, the American Association for Public Opinion Research, as unscientific or unethical. For instance, in order to save time and therefore money, Zogby uses only listed phone numbers; most pollsters program their phones to dial digits randomly, in order to capture unlisted phone numbers as well. Zogby says that there is no demographic difference between people with listed phone numbers and people with unlisted ones; other pollsters say studies have shown that people with unlisted phone numbers are disproportionately rich or poor.

Zogby is also willing to guard the identity of his clients if they want to remain anonymous, while aapor insists that an ethical pollster will always make the sponsor known if a poll's results are published, so that the public can judge whether or not he had a vested interest in the outcome. Zogby argues, reasonably enough, that if his methods are transparent and correct, the sponsor should be irrelevant. He tells clients that if they wish to publish any part of a survey they have to publish the whole thing, including the questions. But if a client wishes to publish only part of a survey--the results that he likes--Zogby offers another, rather underhanded option: he will conduct a second survey, using only the questions from the first one whose answers served the client's purposes, and the client can publish that and keep the first poll private. Zogby is not always as forthcoming about his methods as he claims. When the Pew Research Center requested information about the statistical adjustments he had used in his polls of state races in 1998, he refused to give them up. But there are things he won't do: he polled for Tom Golisano's 2002 campaign for governor of New York, and when his results showed Golisano coming in second, he says, the campaign's advisers wanted him to fiddle with the numbers lest they persuade Golisano to drop out; Zogby said no, and the campaign hired someone else.

Zogby's questions are mostly straightforward, but sometimes they are more leading than necessary. For instance, one question from a recent survey of American Muslims asks, "If you had to choose one of the following ways to wage the war against terrorism, which would you choose?" and offers these options: "Changing America's Middle East policy," "Use of strategic nuclear weapons," "Attacking Iran," "Contracting with mercenaries," "Using U.S. Military Covert/Special Forces," "U.S. Air Force bombing," and "Biological warfare." If there exist choices beyond Arafat or war, they are not offered. Zogby questions can be equally dubious in the service of a right-wing point of view. (At least in his paid work, Zogby is genuinely nonpartisan.) A 1999 poll that Zogby conducted for Newsmax, a conservative Web site, stated, "Allegations have risen"--they were totally unfounded, of course--"that Bill Clinton traded U.S. nuclear and ballistic missile technology to the Chinese for campaign contributions," and then asked whether impeachment proceedings against him should be considered.

Some Zogby surveys can sound almost like push polls--fake polls designed not to collect data but to persuade respondents of a point of view (the prototypical push poll being one in which an interviewer calls voters and asks them whether they would be less likely to vote for Candidate A if they knew that he was a bigamist or had killed his mother with an axe). For instance, a Zogby survey conducted for People for the Ethical Treatment of Animals, or peta, asked respondents whether they would be likely to give up eating meat or dairy products if they were to be apprised of certain facts: "If you knew that within days of birth, chickens have their beaks seared off with a hot blade to keep them from pecking each other in their overly crowded cages?"; "If you knew that male cows and pigs are castrated without painkillers?"; or "If you knew that within days of birth, in order to make veal, male dairy calves are taken from their mothers, chained inside tiny crates to keep their muscles from developing, and fed an iron-deficient diet so that their flesh remains pale?" Provocative as these questions are, though, they do serve a legitimate polling purpose--peta's desire to know which grisly statement about meat will most effectively persuade the public to turn to tofu--and aapor allows that it is the failure to collect data, not shocking questions, that defines a push poll.

The chief reason that the polling establishment disapproves of Zogby is his practice, in political polls, of weighting for party identification. That is, once he assembles the results of a poll, if he discovers that there are, say, more self-identified Democrats and fewer Republicans than he would expect to find in a random sample, he gives each Democratic vote slightly less weight--he might count it as nine-tenths of a vote, for instance. This year, he is basing his expected party percentages on the percentages known in 2000. Weighting per se is not controversial: all pollsters adjust their totals so that their samples reflect the distributions of age and gender known to exist in the population. (And this is often necessary--women tend to answer the phone more than men, and young people are underrepresented in polls, because they spend less time at home.) But to most pollsters party identification is the very variable that a poll of political attitudes is trying to measure in the first place, so the idea of assuming a certain distribution from the outset is preposterous. "Traditional research people trained in statistics just roll their eyes," Andrew Kohut, the former head of Gallup and now the president of Pew, says. "Party is a loose attitude, not a demographic."

Zogby defends his method by arguing that although attitudes toward transitory issues and candidates are indeed unpredictable, a person's relationship with a party is a fairly stable thing. "We inherit it," he says. That's why terms like "Reagan Democrats" come into use: people may be persuaded by a candidate of particular charisma to vote against type, but that doesn't mean they'll identify themselves with the other party from then on. Of course, people do sometimes change parties, and so when Zogby weights for party identification he is basing his calculation on an informed guess; but that's O.K. with him. He believes that polling is an art as much as a science, and if informed guesses will bring him closer to the correct numbers he will use them. He believes it was his weighting by party that enabled him to get so much closer to the final electoral count in 1996 than anybody else and that is making his polls less volatile now, and it is likely that this is true.

Establishment pollsters are concerned above all with trying to eliminate guesswork from their research, with trying to make it as scientific as possible, so they stick to weightings that they can know with near-certainty are justifiable. If some real transformation is taking place in the electorate, after all, they want to discover it, and science is the way. To guess, based on the past, is to preclude discerning change in the future. But Zogby doesn't want to be scientific: he wants to be right.

In the mid-nineteen-thirties, at the dawn of polling's modern era, the biggest and best-known poll in America was the one conducted by Literary Digest. Before each Presidential election, the magazine would send out as many as twenty million postcards to people all over the country whose names it gleaned primarily from automobile registries and telephone books, asking them how they planned to vote. It received as many as five million postcards in response, and so the sample size was enormous. The poll was famous because it had correctly predicted the outcome of three successive elections, from 1924 to 1932.

In 1936, Roosevelt was running for reelection against the Republican Alfred M. Landon. That year, a young pollster named George Gallup made a bold public bet: not only would he predict the election result correctly but Literary Digest would get it wrong. He forecast a Roosevelt win; Literary Digest, he said (with amazing effrontery, since the magazine had not even started polling yet), would predict a fifty-six-per-cent victory for Landon. Sure enough, Literary Digest's results forecast a Landon win (indeed, within a percentage point of Gallup's prediction), and the magazine was humiliated. The trouble was that Literary Digest's sample, large as it was, was rotten, because people who owned cars or telephones were disproportionately affluent and disproportionately Republican. After 1936, the new breed of pollsters such as Gallup, Archibald Crossley, and Elmo Roper canvassed random samples of a few thousand in person rather than non-random millions through the mail.

Zogby is not wrong to talk about the polling establishment as a club. It is a club--in fact, it set out consciously to become one. George Gallup used to invite pollsters out to his farm in New Jersey once a year to meet each other and sit at the feet of the master. aapor, which was founded in 1947, is a much larger group now, but it retains its old sense of camaraderie. It recently put together a book, which consisted in part of social reminiscences ("First experience in the Poconos," the book relates of its second annual conference, in 1948. "They don't appreciate our singing, drinking, late hours") and the lyrics to songs that were sung on such occasions ("Je vous aime, aapor!"). Indeed, several of the association's current members are the sons and daughters of its founding generation.

Nancy Belden, for instance, the president of aapor, is the daughter of Joseph Belden, who founded the first statewide poll, in Texas. She grew up helping out in the extraordinarily labor-intensive process that polling was then: typing up questionnaires, mimeographing, collating, waiting for interviewers to conduct interviews in person, validating their findings (seeking out a certain portion of the respondents to make sure the interviews hadn't been faked), keypunching the results onto cards by hand, and sending the cards off to a mainframe to be analyzed. And that wasn't counting hiring interviewers in the first place, which Belden took very seriously: candidates had to submit to a psychological evaluation consisting of word associations (rhymers--people who picked "rat" rather than "feline" to go with "cat," for instance--didn't get hired, because Belden felt that they seized on superficial similarities rather than thinking things through). The whole process took months.

Back in the forties and fifties, polling was still new and glamorous, which is doubtless one of the reasons that there are so many second-generation pollsters. "When people found out I was related to the Texas Poll, there would be a 'wow' reaction," Nancy Belden says. Then, as now, most pollsters earned their bread and butter by conducting polls for businesses, surveying consumer preferences, and the like. (Two-thirds of Zogby's revenues come from commercial polls.) But many of the early pollsters, Gallup in particular, were idealistic about their calling, feeling that polling would bring democracy to a higher level, enabling and compelling government to be more responsive to the will of the people. Even now, many members of the polling club believe strongly in this idea. "Polling is a central part of why our democracy works so well," Nancy Belden says. "We know what people think and what they need." Frank Newport, the editor-in-chief of Gallup, defends polling against critics who claim that it affects elections--which it certainly does, pollsters' protests notwithstanding--and undermines leadership. "A lot of people don't know where Bosnia is on the map and wouldn't win on 'Jeopardy!,' " Newport says, "but instinct distills into great wisdom with a lot of people voting." It upsets him that politicians claim never to listen to survey results, because he feels that to ignore polls is to turn your back on democracy.

Zogby tends not to talk about his business in such grand terms. He loves polling because he's obsessed with the latest political news, obsessed with information and with being the first to detect shifts of opinion. He remembers waking up each morning during the Watergate hearings feeling a glow of happiness, immediately switching on the radio, and listening to the hourly summaries throughout the day. When he was four, his brother, Jim, took him to school to exhibit him for show-and-tell because he had memorized all the American Presidents forward and backward, the names of their Vice-Presidents and of their wives, and the numbers of children they'd had.

Zogby and his brother are still close--they talk on the phone about five times a day. Jim Zogby, who lives in Washington, founded the Arab-American Institute, a lobbying group.Their parents were Catholics from Lebanon; their mother's maiden name was also Zogby--her parents came from the same village, northeast of Beirut, from which their father emigrated (the village consists, still, almost entirely of Zogbys). When they were growing up in Utica, their father owned a grocery store and their mother was a schoolteacher. John Zogby left town after high school to attend Le Moyne College, a small Jesuit school in Syracuse, intending never to return, but when he graduated, in 1970, he came back to Utica to teach in a Catholic school in order to avoid the draft. He did graduate work in history at Syracuse University and started teaching, and he met his wife, Kathy, when she was a student in his African-American history class.

In college and afterward, Zogby was a "madman on the left," as he puts it. He subscribed to the Socialist Workers Party newspaper, and began a Ph.D. dissertation on Earl Browder, once the general secretary of the American Communist Party. He helped found an alternative "free university" for the community which offered classes on peace and civil rights and Lenny Bruce. He started a consumer lobbying group. But little by little he moved toward the center. He was annoyed when two Marxist professors came to town and criticized his consumer group for not being politically pure enough. He read a biography of Earl Browder and was stunned to discover that there were four people at his funeral, because he'd been thrown out of the Party and declared a nonperson, and he said to himself, "Are these the kind of fucks that I want to be a part of?" Eventually, he abandoned academia and decided to run for mayor of Utica. He lost; but, having polled the public in the course of the race, he knew before anyone else did exactly how much he would lose by. He decided that polling was more fun than running for office. "I realized I'd rather be right than be President," he says.

Even so, Zogby finds it one of the odder twists in his career that he has been adopted by Republicans. Because he predicted correctly in 1996 that Clinton would win by a far smaller margin than most polls were forecasting, many on the right decided that here, finally, was a pollster who saw things the way they were. He was hired by the conservative New York Post, and appeared on Fox News programs and the Christian Broadcasting network. Rush Limbaugh praised him. He is still enough of a lefty to feel out of place in such circles, but he has learned to appreciate work from all quarters. "I'm a vender," he says. Jim Zogby, who is on the executive committee of the Democratic National Committee, used to chastise him for this promiscuity and once asked him whether there was anyone he wouldn't work for. Zogby told him big tobacco. Half an hour later, Philip Morris called requesting his services and he said yes. It was some time before he worked up the nerve to tell his brother.

Pollsters other than Zogby may stick to the same list of methodological rules, but nonetheless establishment polls like those of Gallup, Pew, and Harris often differ substantially from one another. This happens for a variety of reasons, none of which are accounted for in the polls' stated "margin of error." (Margin of error takes into consideration only the size of a sample, which is usually the most trivial source of inaccuracy a pollster has to worry about.) A Pew poll released on September 16th, not long after the Republican Convention, for instance, found Kerry and Bush effectively tied among likely voters, whereas a Gallup poll of likely voters released the next day found that Bush commanded a thirteen-point lead. Representatives of Pew and Gallup explained that this difference was due chiefly to extreme voter volatility. The polls, after all, purported to be measuring the state of voter opinion, so, presumably, if results were varying wildly from day to day, that meant opinions were also. But in fact the results could have been that far apart even if not a single person changed his mind between the times when Pew and Gallup conducted their research.

For one thing, the slightest variation in the wording of a question can significantly affect the result, even in a simple horse-race poll. A poll that asks whom a person would vote for if he had to vote today is more likely to favor the candidate currently ahead (as opposed to a poll that asks whom a person will vote for in November). The order in which questions are asked is also important. If a person approves of Bush's war in Iraq, say, but disapproves of the deficit, his response to a question about whom he plans to vote for may be affected by whether previous questions reminded him of either of those issues.

Stanley Presser, a past president of aapor and a professor of sociology at the University of Maryland, has spent much of his career studying the effects of question wording, and knows that they can be complicated and hard to predict. Back in 1974, he tested what he thought was an easy example--a Harvard study by the eminent pollster Samuel Stouffer, conducted in the fifties during the height of McCarthyism. Stouffer had polled with questions that to Presser seemed obviously leading, such as: "There are always some people whose ideas are considered bad or dangerous by other people. For instance, somebody who is against all churches and religions. Should such a person be allowed to speak in your town?" Presser and a colleague conducted a survey of their own in which they asked half their sample the question as Stouffer had written it and the other half the question without its leading first sentence. To his surprise, Presser discovered that the change made no difference: sixty-six per cent of both halves of the survey felt that an atheist should be allowed to speak.

Presser realized that when a question was so blatantly leading, most people saw through its intent and were not affected by it. The same year, though, he conducted another poll, in which he tested responses to a more subtle difference. He asked half of his sample, "Do you think the United States should allow public speeches against democracy?" and the other half, "Do you think the United States should forbid public speeches against democracy?" This time, the difference between the two groups was very large indeed: fifty-six per cent said no to the first question, but only thirty-nine per cent said yes to the second. Presser saw that, while "not allow" and "forbid" might literally mean the same thing, the difference in the connotations of the words--the intimidating, legalistic tone of "forbid," compared with the gentler, parental tone of "not allow"--was significant enough to reverse a person's stated opinion on an issue and alter the majority result of a survey. He concluded that, because such effects were so unpredictable, if a survey was really to get at the truth of a matter it had to ask about it in several different ways. It is a source of continual irritation to him that so few surveys do this.

For this reason, ever since his studies in the seventies, Presser has felt that polls that ask people whether they're for or against things are not to be trusted. Absolute percentages, he believes, are pretty much useless. But he does place considerable faith in comparative percentages. Do seventy-six per cent of American Ph.D.s believe that dogs are cuter than ferrets? He doubts that we can ever know for sure. But were a poll to compare the responses of Ph.D.s and high-school dropouts to the same question, or were a poll to compare the responses of Ph.D.s ten years ago with the responses of Ph.D.s today, then, he believes, a meaningful result can be obtained: because the same question is used for the two separate groups, whatever prejudicial effect its wording might have will be cancelled out. When Presser was in graduate school, at the University of Michigan, he used to run into an emigre statistician named Leslie Kish in the elevator. Presser would ask how he was doing; Kish would reply, "Compared to whom?"

Timing is another reason that election polls often differ from each other. Polls that come out after a convention, for instance, are likely to be compared for their estimate of a convention bounce, but if one polling company did its calling during the convention and another just a day or two afterward they may show different results. What's more, a poll that is conducted over three or four days rather than one or two is likely to be more accurate because it will have a better chance of reaching people who were not home the first time they were called. Survey samples are ideally supposed to be random, and everyone in the sample should have an equal chance of being questioned. A survey, then, that registered only the responses of people who were home on a Saturday night would not be as close to that ideal as one that polled on Sunday and Monday as well. There is also the matter of interview protocol. All pollsters include options such as "not sure" and "don't know," but not all interviewers are instructed to read those options to the respondents. (Zogby's interviewers almost never do.) If it is not explicitly suggested to a respondent that "not sure" is a valid reply to a question, he may well come up with an opinion, so as not to appear stupid or unhelpful, but this opinion may be hesitant and unconsidered, and he may change his mind about it the next day. The pollster, however, will register it as having the same validity as the response of someone with a firm conviction on the issue.

Probably the single most important factor in generating the differences in establishment election polls, though, is the nature of the sample: some polls survey all adults, some polls survey only registered voters, and some polls survey only likely voters. The first two categories are relatively straightforward (though follow-up research has determined that people lie about being registered to vote); it's the third one that's tricky. Figuring out who is likely to vote is difficult, especially several months before an election; every polling firm has its own algorithm, and the nature of the algorithm has a large effect on a poll's outcome. Gallup, for instance, asks respondents seven questions, such as: "Do you happen to know where people in your neighborhood go to vote?" and "In the election for President in November of 2000, did things come up that kept you from voting, or did you happen to vote?" (Each firm asks slightly different questions, and that makes for discrepancy: questions that ask about past voting behavior, for example, may affect results because regularity of turnout is associated with high socioeconomic status, which, in turn, is associated with voting Republican. Almost every firm will ask directly how likely the person is to vote; but responses to that question are unreliable, because they will depend on the person's feelings about the election at that moment, which may change the next day if an event or piece of news stirs up his anger or enthusiasm.) Gallup then assumes that only fifty or fifty-five per cent of the population will vote in the election, since that is historically the case, and counts only the fifty or fifty-five per cent of its sample that scored the highest on the likely-voter questionnaire. This method of calculation means that Gallup is throwing out half of its sample (rather than, say, weighting the whole sample), and that is risky for the same reason that weighting for party is risky: it means that Gallup is relying on a key variable in the election--in this case turnout--remaining unchanged.

Polling isn't all statistics, though--if it were, Zogby wouldn't want to be in the business. Sometimes in the course of work he gets to do one of the things he loves most, which is to chat with someone nearly as obsessive as he is about the intricacies of political races, past and present. One afternoon, he had an appointment to talk on the phone with such a person, the novelist Richard North Patterson. Patterson was writing a political thriller about a Presidential candidate named Chad Palmer, inspired by John McCain, and, since he wanted the book to be realistic, he had arranged to consult with Zogby on the details of the race. Palmer had already won the New Hampshire primary, as McCain did in 2000, and the question now was how well he could plausibly do in other states, given that he was opposed to an amendment to ban gay marriage and was dating a liberal black Hollywood actress. Zogby is a big fan of thrillers and mysteries, and a particular fan of Patterson's books, so he found the idea of contributing to one extremely exciting. He sat at his desk in his office and leaned eagerly into his speakerphone.

"I think Michigan is made to order for Palmer," he said. "I don't think it's far-fetched to have him winning South Carolina, but a lot of mischief can be done in Michigan. The autoworkers came out to vote to defeat George W. Bush, but this is the home of the Reagan Democrat, and it's easier for Reagan Democrats to cross over for a moderate conservative. I think also there are very good opportunities in Pennsylvania."

"It's Fasano's home state," Patterson reminded him. Fasano was Palmer's rival from the religious right.

"Oh, that's right. I close my eyes and think Rick Santorum."

"You're supposed to. Ugly thought, but the son of a bitch has had more survival power than I ever thought he would."

"I know. He's also a foul, foul kid. The Fasano novel is next, right?"

The conversation with Patterson drifted from Chad Palmer to his real-life counterpart, McCain, whom both Patterson and Zogby knew and admired. (He is a former Zogby client.)

"McCain is one of those rare people like Jimmy Carter that, when you like them, you project yourself on them," Zogby said. "I always remember that Jimmy Carter won the New Hampshire primary in 1976 by running to the right of all the liberals, then went down to Florida and ran to the left of Scoop Jackson and George Wallace. A poll came out and forty-three per cent of Americans thought Jimmy Carter was pro-life and forty-four per cent thought he was pro-choice. They just liked Jimmy Carter. And that's true of John McCain: 'He's a great man and I love him, he couldn't possibly be . . . '--fill in the blank."

"I don't know how many fanatical pro-choicers I know who love McCain," Patterson said.

"Well, this whole bear hug is a very intriguing thing." Zogby had been horrified to open the newspaper a few days earlier to see McCain clinging to the President from the side, his arms wrapped all the way around Bush's torso, like a just-saved refugee.

"Oh boy, it made me squirm, because I know what John feels about Bush," Patterson said.

" 'Disgust' is probably too mild a word."

"He's got his admiration under tight control, that's for sure," Patterson said. "I'm pretty close to John, but I'm unable to ask him about it."

Zogby and Patterson agreed to speak again soon, and Zogby hung up the phone smiling. He was still thrilled by the thought of playing a role in plotting the novel. He had suggested to Patterson that Palmer should hire Zogby as his campaign pollster. He had been joking, of course, but you never knew.