The Flaw of Averages

How polls obscure America’s many social patchworks
June 25, 2007

The first American survey was ordered by the Constitution, and Article 1, Section 2, even specified the cross-tabs: “Number of free Persons, including those bound to Service for a Term of years”; “Indians not taxed”; and, infamously, “three fifths of all other persons”—that is to say, slaves. The new nation may have taken as its motto E pluribus unum, “From many, one,” but differences among Americans were written into the original political charter.

One hundred forty-nine years later, modern political surveying was born. The most famous presidential “straw poll” of the day was published for years in a magazine called Literary Digest. Its 1936 survey said the Republican Alf Landon would beat Franklin Roosevelt. Instead, Roosevelt ran up the biggest landslide in history. But George Gallup, Archibald Crossley, and Elmo Roper, competing market researchers who did political work on the side, had called it for Roosevelt. They used, Sarah Igo explains in her very thoughtful book, “statistical techniques that permitted a tiny cross-section of citizens of different regions, classes, and races to stand in for the whole.” It seemed to bestow on them an almost magic power: to divine how the average man thought.

By the 1940 presidential election, the three researchers claimed a grandiose civic function. “The political wisdom of the common people can now be settled,” Gallup pronounced, “on the basis of a mountain of factual data.” They gave their articles titles like, “We, the People, Are Like This,” and made confident assertions like, “The average American takes the attitude that conscription is . . . necessary.” This “average American” took on an almost living, breathing presence.

Such rhetoric obscured a paradox. Producing that one essential American required an understanding of the nation as a patchwork: so many people in this income group, so many in that; x number of Protestants, y number of Catholics. Only then could the pollsters know which two thousand Americans could, when extrapolated, be taken as a reasonable facsimile of the whole. The work required as finely tuned a sense of the gradations of class as the compiler of a social register; “both Roper and Gallup preferred interviews to take place in the home rather than on the street,” Igo observes, “because a glance around the domestic interior allowed a better assessment of the interviewee’s financial status.”
Deepening the paradox was the fact that not all Americans were equally worth being counted toward the average. Predicting elections meant counting likely voters. Since minorities, the poor, and women were less likely to vote, they were deliberately undercounted—though they, too, might have opinions about whether conscription was necessary. In a sense, the poll impresarios were in a better position than anyone else to understand the pitfalls of averaging Americans. Yet that didn’t keep them from making extravagant claims about the pristinely value-free nature of their work.

Nor did it keep them from making extravagant value claims: that their average American did not merely exist, he happened to be wise and independent-minded. Intellectuals like Walter Lippmann had long warned that a mass-media-driven public would be easily manipulated. Gallup said his polls showed “not one bit of scientific evidence” to support the idea. It was, in fact, the opposite: “the political wisdom of the common people”—Gallup specified their readiness to submit to conscription—was “ahead” of their leaders.

Not surprisingly, once thus enshrined, the “average American” argued back at Gallup. Igo quotes the angry letters: “If you took a true poll of the American people . . . you would find them over 95 percent for Senator McCarthy,” read one; reported another, “Clamp your eyes on something true right from our plant. This straw vote was taken yesterday: Ike 3–Democrat 24 . . . Ike can not get elected.”

Sign up for CJR's daily email

The two letters make opposite partisan claims. But they partake of the same assumption: that there is an average American, and that his politics can be deduced by empirical observation. That was a fairly new idea, and Igo devotes the first third of her book to the phenomenon that did more than anything else to get the public thinking that way: not a survey, but a study of a single community published in 1929, Robert and Helen Merrell Lynd’s Middletown.

Like Gallup and the other political pollsters Igo writes about, the Lynds promoted the scientific status of their claims—and simultaneously slipped in a moral subtext. The Lynds had been hired to undertake “a Basic Religious Study of the people in a small industrial community.” Their sponsors wanted a town “near the centre of life in the U.S.” and home to “all the chief denominations.” The Lynds further refined those conditions in the interest of an emerging social science ideology. They rejected South Bend for its “cultural and religious heterogeneity”—too many immigrants, for one thing—preferring instead Muncie, Indiana.

Blessed with what they regarded as so perfect a laboratory, their ambitions expanded. The Lynds reinvented themselves as anthropologists of the average American. Their annoyed patrons cut them adrift. Released commercially, however, their book became a surprise bestseller. Thus did the Lynds learn what Roper, Crossley, and Gallup were to learn a decade later: there was a hunger among ordinary Americans for being told what it was they were supposed to already believe.

“This was looking at yourself in the mirror,” Good Housekeeping’s reviewer enthused, and, as another review suggested, people liked what they saw: “More cities like Middletown are needed here—good, sane, substantial, hard-working communities that breed the best citizens.” The only problem: that was the opposite of what the authors intended to convey. The Lynds worried that the typical Middletowner was shallow, irrational, and greedy—and yet their book was systematically misread through a prejudice: if they’re writing about the typical American, they must mean to describe a decent American. But Middletown was larded with Veblenesque scoldings: “More and more of the activities of living are coming to be strained through the bars of the dollar sign.” Even religion “served the instrumental function of furthering social status.” What’s more, whenever the Lynds revealed the Middletowners’ core values as inadequate or untrue, there would be “a redoubling of emphasis upon the questioned ritual and a cry for more loyalty to it.”

You might say, if you were being ungenerous, that the Lynds stumbled into a mess of their own creation. They had found in Muncie what they thought was the typical U.S. city, even if, as Igo points out, it was a “demographic curiosity,” “populated largely by farm-born factory workers . . . more ‘old stock’ . . .than any other city in the Midwest of its size, apart from New Albany, Indiana.” To further their scientific quest for pristine homogeneity, the Lynds decided to include no answers from African Americans in their tabulations—though Muncie’s black population was proportionally larger than those in Detroit and Chicago. They were trying to make themselves scientists, but they ended up endorsing a mythology: that the typical American was native-born, midwestern, and white—when a truer social science would have shown that that was no longer true.

The kind of place that social critics like the Lynds, Gallup, and the others took to be “typical” resembled the towns depicted by those with no such social scientific agenda: the novelists Sinclair Lewis (Main Street; Babbit) and Sherwood Anderson (Winesburg, Ohio). It was also the same kind of town with which the Good Housekeepings of the world saturated their pages—only to idealize it, rather than to criticize it. There’s a lesson here: whether you’re Gallup, eager to enshrine the average American, or the Lynds, vaguely distrustful of him, when you search for the “average,” you tend to find it exactly where you started looking—and, lo and behold, whether reader or author, what you find there will match your preexisting convictions.

The third set of surveyors igo examines found the average American in a very different place. The biologist Alfred Kinsey, a professor at the University of Indiana (naturally), set out to discover how the average American has sex. What he discovered was that the average American was a pervert. That’s not how Dr. Kinsey would have put it, of course. From an unprecedented course of 5,300 interviews, Sexual Behavior in the Human Male, Kinsey’s research team specified how many times their subjects obtained “sexual outlet”: masturbation, nocturnal emission, heterosexual petting, heterosexual intercourse, homosexual outlet, and, yes, “animal contacts.” What they found was that under the existing sex laws, 95 percent of American men were criminals, and that 37 percent of men had had homosexual contact. And that was a result the doctor preferred to find. He was a closet sexual liberationist who believed that enlightened rationality could strike from the earth what he saw as the tyranny of sex laws and norms, rooted in shame, hypocrisy, and ignorance.

But were his findings right? The debate has raged ever since. His underlying methodological flaw was that a person had to be willing to submit to the most probing sexual interview imaginable to show up in the data. As one angry critic charged, “no normal moral man or woman” would ever do that. The critic had a point. People who were comfortable with sex did seem to be over-represented in his sample; as some of Igo’s examples make clear, for many the thought of being interviewed by Kinsey was in itself a sexual thrill. The problem was compounded when Kinsey turned his attention to Sexual Behavior in the Human Female. “Given reigning conventions about female modesty, the idea of women submitting to interviews with the all-male research team was especially unsettling,” Igo notes. “Many concluded that Kinsey’s female interviewees must have been prostitutes, or at the very least seriously maladjusted.”

The American inclination to believe that the American who is typical is for that reason good was about to receive a workout as never before. These people were not “good”—the very fact that they answered Kinsey’s questions was evidence of that. Therefore they could not be typical. How did critics of the study know? They were typical Americans, and they didn’t do any of these things. “I have lived with one woman for 46 years,” wrote one, “and I do not agree with your findings . . . when you show as one magazine reports that 62 percent of adult women practice masturbation.”

Such letters weren’t that different from the letters Gallup got. The vast majority of letter writers, Igo reports, “were troubled that a science claiming to speak for them had never bothered to ask their opinion in the first place.” Kinsey and Gallup, after their fashion, each argued that the average American had a special moral standing. So if you disagreed, you probably weren’t average. Worse, they may have been suggesting that something was wrong with you. Such were the wages of averaging Americans. (No one ever wrote Gallup to complain, “How dare you claim my opinion is in the majority?”) It is a uniquely American problem. Alexis de Tocqueville, after all, was fascinated by the paradox that in this most free of all nations, the one in which self-invention was supposed to be an accepted part of life, demands for conformity were more pervasive than in the unfree nations of Europe. E pluribus unum sometimes means denying that “the many” exist in the first place.

We are better able to accept that America is a diverse nation now, Igo concludes. “The particular mass public of fifty years ago—characterized by an iconic Americanism, a majoritarian emphasis, and a fixation on the normal,” she claims, “does not exist in quite the same way.” I found that too sanguine. While I was reading Igo’s study, New York’s Senator Charles Schumer came out with a book entitled Positively American: Winning Back the Middle-Class Majority One Family at a Time. He offers up a panoply of policy suggestions, some any given reader might find inspired, others objectionable, others inadequate to the challenges America faces, others unworkable. Any reader will be ready to argue with Schumer; that’s how political books are supposed to work.

Only Schumer cleverly short-circuits his interlocutors. He has invented a “typical” American family, whom he calls the Baileys—white, two parents, two kids, living in a Brooklyn neighborhood that never would have been allowed to be called typical in the Lynds’ day. He has derived all his positions, he insists, via scrupulous introspection about what these “average Americans” need and want. Lo and behold, they think just like Charles Schumer. Disagree? Then you must not be an average American.
America is a patchwork. Pretending otherwise—to claim access to a notional average American—is to always make a specific moral argument. And you tend to find it exactly where you started looking—in the mirror. 

Rick Perlstein is the author of Before the Storm: Barry Goldwater and the Unmaking of the American Consensus. His new book, Nixonland: The Politics and Culture of the American Berserk,1965-1972, is forthcoming.