by Susan Faludi
As the backlash consensus solidified, statistics on women stopped functioning as social barometers. The data instead became society’s checkpoints, positioned at key intervals in the life course of women, dispatching advisories on the perils of straying from the appointed path. This prescriptive agenda governed the life span of virtually every statistic on women in the ’80s, from initial gathering to final dissemination. In the Reagan administration, U.S. Census Bureau demographers found themselves under increasing pressure to generate data for the government’s war against women’s independence, to produce statistics “proving” the rising threat of infertility, the physical and psychic risks lurking in abortion, the dark side of single parenthood, the ill effects of day care. “People I’ve dealt with in the [Reagan] government seem to want to recreate the fantasy of their own childhood,” Martin O’Connell, chief of the Census Bureau’s fertility statistics branch, says. And results that didn’t fit that fantasy were discarded, like a government study finding that federal affirmative action policies have a positive effect on corporate hiring rates of women and minorities. The Public Health Service censored information on the beneficial health effects of abortion and demoted and fired federal scientists whose findings conflicted with the administration’s so-called pro-family policy.
“Most social research into the family has had an immediate moral purpose—to eliminate deviations like divorce, desertion, illegitimacy, and adultery—rather than a desire to understand the fundamental nature of social institutions,” social scientist Kingsley Davis wrote in his 1948 classic Human Society. More than forty years later, it is one of the few statements by a demographer that has held up.
THE MAN SHORTAGE: A TALE OF TWO MARRIAGE STUDIES
Valentine’s Day 1986 was coming up, and at the Stamford Advocate, it was reporter Lisa Marie Petersen’s turn to produce that year’s story on Cupid’s slings and arrows. Her “angle,” as she recalls later, would be “Romance: Is It In or Out?” She went down to the Stamford Town Center mall and interviewed a few men shopping for flowers and chocolates. Then she put in a call to the Yale sociology department, “just to get some kind of foundation,” she says. “You know, something to put in the third paragraph.”
She got Neil Bennett on the phone—a thirty-one-year-old unmarried sociologist who had recently completed, with two colleagues, an unpublished study on women’s marriage patterns. Bennett warned her the study wasn’t really finished, but when she pressed him, he told her what he had found: college-educated women who put schooling and careers before their wedding date were going to have a harder time getting married. “The marriage market unfortunately may be falling out from under them,” he told her.
Bennett brought out the numbers: never married college-educated women at thirty had a 20 percent chance of being wed; by thirty-five their odds were down to 5 percent; by forty, to 1.3 percent. And black women had even lower odds. “My jaw just dropped,” recalls Petersen, who was twenty-seven and single at the time. Petersen never thought to question the figures. “We usually just take anything from good schools. If it’s a study from Yale, we just put it in the paper.”
The Advocate ran the news on the front page. The Associated Press immediately picked up the story and carried it across the nation and eventually around the world. In no time, Bennett was fielding calls from Australia.
In the United States, the marriage news was absorbed by every outlet of mass culture. The statistics received front-page treatment in virtually every major newspaper and top billing on network news programs and talk shows. They wound up in sitcoms from “Designing Women” to “Kate and Allie;” in movies from Crossing Delancey to When Harry Met Sally to Fatal Attraction; in women’s magazines from Mademoiselle to Cosmopolitan; in dozens of self-help manuals, dating-service mailings, night-class courses on relationships, and greeting cards. Even a transit advertising service, “The Street Fare Journal,” plastered the study’s findings on display racks in city buses around the nation, so single straphangers on their way to work could gaze upon a poster of a bereft lass in a bridal veil, posed next to a scorecard listing her miserable nuptial odds.
Bennett and his colleagues, Harvard economist David Bloom and Yale graduate student Patricia Craig, predicted a “marriage crunch” for baby-boom college-educated women for primarily one reason: women marry men an average of between two and three years older. So, they reasoned, women born in the first half of the baby boom between 1946 and 1957, when the birthrate was increasing each year, would have to scrounge for men in the less populated older age brackets. And those education-minded women who decided to get their diplomas before their marriage licenses would wind up worst off, the researchers postulated—on the theory that the early bird gets the worm.
At the very time the study was released, however, the assumption that women marry older men was rapidly becoming outmoded; federal statistics now showed first-time brides marrying grooms an average of only 1.8 years older. But it was impossible to revise the Harvard-Yale figures in light of these changes, or even to examine them—since the study wasn’t published. This evidently did not bother the press, which chose to ignore a published study on the same subject—released only a few months earlier—that came to the opposite conclusion. That study, an October 1985 report by researchers at the University of Illinois, concluded that the marriage crunch in the United States was minimal. Their data, the researchers wrote, “did not support theories which see the marriage squeeze as playing a major role in recent changes in marriage behavior.” (In fact, in their historical and geographic review of marital data, they could find “marriage crunches” only in a few European nations back in the 1900s and in some Third World countries in more modern times.)
In March 1986, Bennett and his co-researchers released an informal “discussion paper” that revealed they had used a “parametric model” to compute women’s marital odds—an unorthodox and untried method for predicting behavior. Princeton professors Ansley Coale and Donald McNeil had originally constructed the parametric model to analyze marital patterns of elderly women who had already completed their marriage cycle. Bennett and Bloom, who had been graduate students under Coale, thought they could use the same method to predict marriage patterns. Coale, asked about it later, was doubtful. “In principle, the model may be applicable to women who haven’t completed their marital history,” he says, “but it is risky to apply it.”
To make matters worse, Bennett, Bloom, and Craig took their sample of women from the 1982 Current Population Survey, an off year in census-data collection that taps a much smaller number of households than the decennial census study. The researchers then broke that sample down into ever smaller subgroups—by age, race, and education—until they were making generalizations based on small unrepresentative samples of women.
As news of the “man shortage” study raced through the media, Jeanne Moorman, a demographer in the U.S. Census Bureau’s marriage and family statistics branch, kept getting calls from reporters seeking comment. She decided to take a closer look at the researchers’ paper. A college-educated woman with a doctoral degree in marital demography, Moorman was herself an example of how individual lives defy demographic pigeonholes: she had married at thirty-two, to a man nearly four years younger.
Moorman sat down at her computer and conducted her own marriage study, using conventional standard-life tables instead of the parametric model, and drawing on the 1980 Population Census, which includes 13.4 million households, instead of the 1982 survey that Bennett used, which includes only 60,000 households. The results: At thirty, never-married college-educated women have a 58 to 66 percent chance at marriage—three times the Harvard-Yale study’s predictions. At thirty-five, the odds were 32 to 41 percent, seven times higher than the Harvard-Yale figure. At forty, the odds were 17 to 23 percent, twenty-three times higher. And she found that a college-educated single woman at thirty would be more likely to marry than her counterpart with only a high school diploma.
In June 1986, Mo
orman wrote to Bennett with her findings. She pointed out that more recent data also ran counter to his predictions about college-educated women. While the marriage rate has been declining in the general population, the rate has actually risen for women with four or more years of college who marry between ages twenty-five and forty-five. “This seems to indicate delaying rather than forgoing marriage,” she noted.
Moorman’s letter was polite, almost deferential. As a professional colleague, she wrote, she felt obligated to pass along these comments, “which I hope will be well received.” They were received with silence. Two months passed. Then, in August, writer Ben Wattenberg mentioned Moorman’s study in his syndicated newspaper column and noted that it would be presented at the Population Association of America Conference, an important professional gathering for demographers. Moorman’s findings could prove embarrassing to Bennett and Bloom before their colleagues. Suddenly, a letter arrived in Moorman’s mailbox. “I understand from Ben Wattenberg that you will be presenting these results at PAA in the spring,” Bennett wrote; would she send him a copy “as soon as it’s available”? When she didn’t send it off at once, he called and, Moorman recalls, “He was very demanding. It was, ‘You have to do this, you have to do that.’” This was to become a pattern in her dealings with Bennett, she says. “I always got the feeling from him that he was saying, ‘Go away, little girl, I’m a college professor; I’m right and you have no right to question me.’” (Bennett refuses to discuss his dealings with Moorman or any other aspect of the marriage study’s history, asserting that he has been a victim of the over-eager media, which “misinterpreted [the study] more than I had ever anticipated.”)
Meanwhile at the Census Bureau, Moorman recalls, she was running into interference from Reagan administration officials. The head office handed down a directive, ordering her to quit speaking to the press about the marriage study because such critiques were “too controversial.” When a few TV news shows actually invited her to tell the other side of the man-shortage story, she had to turn them down. She was told to concentrate instead on a study that the White House wanted—about how poor unwed mothers abuse the welfare system.
By the winter of 1986, Moorman had put the finishing touches on her marriage report with the more optimistic findings and released it to the press. The media relegated it to the inside pages, when they reported it at all. At the same time, in an op-ed piece printed in the New York Times, the Boston Globe, and Advertising Age, Bennett and Bloom roundly attacked Moorman for issuing her study, which only “further muddled the discussion,” they complained. Moorman and two other Census Bureau statisticians wrote a response to Bennett and Bloom’s op-ed article. But the Census Bureau held up its release for months. “By the time they finished blue-lining it,” Moorman recalls, “it said nothing. We sent it to the New York Times, but by then it was practically the next December and they wouldn’t print it.”
Bennett and Bloom’s essay had criticized Moorman for using the standard-life tables, which they labeled a “questionable technique.” So Moorman decided to repeat her study using the Harvard-Yale men’s own parametric model. She took the data down the hall to Robert Fay, a statistician whose specialty is mathematical models. Fay looked over Bennett and Bloom’s computations and immediately spotted a major error. They had forgotten to factor in the different patterns in college-and high school-educated women’s marital histories. (High school-educated women tend to marry in a tight cluster right after graduation, making for a steep and narrow bell curve skewed to the left. College-educated women tend to spread out the age of marriage over a longer and later period of time, making for a longer and lower curve skewed to the right.) Fay made the adjustments and ran the data again, using Bennett and Bloom’s mathematical model. The results this time were nearly identical to Moorman’s.
So Robert Fay wrote a letter to Bennett. He pointed out the error and its significance. “I believe this reanalysis points up not only the incorrectness of your results,” he wrote, “but also a necessity to return to the rest of the data to examine your assumptions more closely.” Bennett wrote back the next day. “Things have gotten grossly out of hand,” he said. “I think it’s high time that we get together and regain at least some control of the situation.” He blamed the press for their differences and pointedly noted that “David [Bloom] and I decided to stop entirely our dealings with all media,” a hint perhaps that the Census researchers should do the same. But Bennett needn’t have worried about his major error making headlines: Moorman had, in fact, already mentioned it to several reporters, but none were interested.
Still, Bennett and Bloom faced the discomforting possibility that the Census researchers might point out their mistake at the upcoming PAA conference. In what Moorman suspects was an effort to avert this awkward event, Bennett and Bloom suddenly proposed to Moorman that they all “collaborate” on a new study they could submit jointly to the PAA conference—in lieu of Moorman’s. When Bennett and Bloom discovered they had missed the conference deadline for filing such a new paper, Moorman notes, they just as suddenly dropped the collaboration idea.
In the spring of 1987, the demographers flew to Chicago for the PAA conference. The day before the session, Moorman recalls, she got a call from Bloom. He and Bennett were going to try to withdraw their marriage study anyway, he told her—and substitute a paper on fertility instead. But the conference chairman refused to allow the eleventh-hour switch.
When it was time to present the notorious marriage study before their colleagues, Bloom told the assembly that their findings were “preliminary,” gave a few brief remarks and quickly yielded the floor. Moorman was up next. But, thanks to still more interference from her superiors in Washington, there was little she could say. The director of the Census Bureau, looking to avoid further controversy, had ordered her to remove all references to the Harvard-Yale marriage study from her conference speech.
Three and a half years after the Harvard-Yale study made nationwide headlines, the actual study was finally published—without the marriage statistics. Bennett told the New York Times: “We’re not shying away because we have anything to hide.” And the reporter took him at his word. The famous statistics were deleted, the news story concluded, only because the researchers found them “a distraction from their central findings.”
• • •
IN ALL the reportorial enterprise expended on the Harvard-Yale study, the press managed to overlook a basic point: there was no man shortage. As a simple check of the latest Census population charts would have revealed, there were about 1.9 million more bachelors than unwed women between the ages of twenty-five and thirty-four and about a half million more between the ages of thirty-five and fifty-four. If anyone faced a shortage of potential spouses, it was men in the prime marrying years: between the ages of twenty-four and thirty-four, there were 119 single men for every hundred single women.
A glance at past Census charts would also have dispelled the notion that the country was awash in a record glut of single women. The proportion of never-married women, about one in five, was lower than it had been at any time in the 20th century except the ’50s, and even lower than the mid to late 19th century, when one in three women were unwed. If one looks at never-married women aged forty-five to fifty-four (a better indicator of lifelong single status than women in their twenties and thirties, who may simply be postponing marriage), the proportion of unwed women in 1985 was, in fact, smaller than it had ever been—smaller even than in the marriage-crazed ’50s. (Eight percent of these women were single in 1950, compared with 5 percent in 1985.) In fact, the only place where a “surplus” of unattached women could be said to exist in the ’80s was in retirement communities. What was the median age of women who were living alone in 1986? Sixty-six years old. (The median age of single men, by contrast, was forty-two.)
Conventional press wisdom held that single women of the ’80s were desperate for marriage—a desperation that mounted with every passing unwed year. But s
urveys of real-life women told a different story. A massive study of women’s attitudes by Battelle Memorial Institute in 1986, which examined fifteen years of national surveys of ten thousand women, found that marriage was no longer the centerpiece of women’s lives and that women in their thirties were not only delaying but actually dodging the wedding bands. The 1985 Virginia Slims poll reported that 70 percent of women believed they could have a “happy and complete” life without a wedding ring. In the 1989 “New Diversity” poll by Langer Associates and Significance Inc., that proportion had jumped to 90 percent. The 1990 Virginia Slims poll found that nearly 60 percent of single women believed they were a lot happier than their married friends and that their lives were “a lot easier.” A 1986 national survey commissioned by Glamour magazine found a rising preference for the single life among women in their twenties and thirties: 90 percent of the never-married women said “the reason they haven’t [married] is that they haven’t wanted to yet.” And a 1989 Louis Harris poll of still older single women—between forty-five and sixty—found that the majority of them said they didn’t want to get married. A review of fourteen years of U.S. National Survey data charted an 11 percent jump in happiness among 1980s-era single women in their twenties and thirties—and a 6.3 percent decline in happiness among married women of the same age. If marriage had ever served to boost personal female happiness, the researchers concluded, then “those effects apparently have waned considerably in the last few years.” A 1985 Woman’s Day survey of sixty thousand women found that only half would marry their husbands again if they had it to do over.
In lieu of marriage, women were choosing to live with their loved ones. The cohabitation rate quadrupled between 1970 and 1985. When the federal government finally commissioned a study on single women’s sexual habits in 1986, the first time ever, the researchers found that one-third of them had cohabited at some time in their lives. Other demographic studies calculated that at least one-fourth of the decline in married women could be attributed to couples cohabiting.