8

Marriage

In which I describe a decline of marriage in white America that took different courses for Belmont and Fishtown during the 1980s, and an unprecedented increase in white nonmarital births that has been concentrated in Fishtown and scarcely touched Belmont.

IHAVE CHOSEN TO present class divergence in marriage first because it is so elemental. Over the last half century, marriage has become the fault line dividing American classes.1

What Whites Said About Marriage

In 1962, the Saturday Evening Post—the magazine with the Norman Rockwell covers—commissioned the Gallup Organization to conduct a survey of the attitudes of American women. For the ever-married sample whose opinions are reported here, Gallup interviewed 1,813 women ages 21–60.

“In general, who do you think is happier,” the Gallup interviewer asked, “the girl who is married and has a family to raise, or the unmarried career girl?” Ninety-six percent of the wives said the married girl with a family was happier. Ninety-three percent said that they did not, in retrospect, wish they had pursued a career instead of getting married.

More than half the ever-married women thought that the ideal age for a woman to be married was 20 through 23, with 21 being the most commonly named year. Only 18 percent thought a woman should wait until age 25 or older.

More than a third of the ever-married women knew a woman who had engaged in an affair after she married, but they didn’t approve. Eighty-four percent said there was never any justification for women having sexual affairs with men other than their husbands.

A Different World

To get a sense of just how different attitudes were in the early 1960s, perhaps this will do it. These ever-married women were asked, “In your opinion, do you think it is all right for a woman to have sexual relations before marriage with a man she knows she is going to marry?” Note the wording. Not sex with someone a woman is dating, nor with someone a woman loves, but with a man she knows she is going to marry. Eighty-six percent said no.2

Gallup’s survey for the Saturday Evening Post didn’t ask under what circumstances divorce was justified, but we have another poll conducted in 1960 that asked whether divorce should be made more difficult or easier to obtain. In 1960, no-fault divorce did not exist and a speedy divorce was possible only in Nevada. In many states, the only legal grounds for divorce were adultery or cruelty. Even so, 56 percent of the respondents said that divorce should be made more difficult, compared to only 9 percent who thought it should be made easier.3

The General Social Survey, abbreviated GSS and conducted since 1972 by the National Opinion Research Center at the University of Chicago, is the most widely used database for tracking American social trends. By the time the GSS began to ask questions about attitudes toward marriage, things had already changed, and then continued to change even more in the next decades. For example, the GSS began asking in 1977 whether their respondents agreed with this statement: “It is much better for everyone involved if the man is the achiever outside the home and the woman takes care of the home and family.” We cannot know exactly what the 1960s answers to that question would have been, but if 96 percent of wives in the 1962 Gallup survey thought women were generally happier with a family than with a career, we have to assume that the “agree” responses for the GSS item would have been at least somewhere above 90 in the early 1960s. Appendix D describes how I reached my estimate of 95 percent, designated by the unattached X you see in Figure 8.1. It represents an estimate for the first half of the 1960s, not the entire decade.

FIGURE 8.1. THE WOMAN’S ROLE IN MARRIAGE

Source: GSS. Sample limited to whites ages 30–49. Data smoothed using locally estimated regression (LOESS). See appendix A for a description of LOESS.

The main effect applied across all classes, and comes as no surprise: The traditional conception of marital roles took a big hit from the 1960s through the 1980s. A substantial class difference remained, however. As of the 2000s, almost 40 percent of Fishtown still took a traditional view of the woman’s role, compared to less than 20 percent of Belmont.

About the Graphs

Almost all of the graphs in part 2 go from 1960 through 2010 on the horizontal axis, even when data for the end points are not available, and show percentages on the vertical axis. This raises a perennial problem in presenting such data: How big a range should the vertical axis show? If the range goes all the way from zero to 100 percent, the shape and importance of trends can be obscured. Make the range too narrow, and a small change can be visually exaggerated. There’s no pat answer, because sometimes a small percentage change over fifty years is important and sometimes it isn’t. I have included only graphs with changes that I judge to be important, gearing the range to the minimum and maximum values of the variables being plotted, but with a minimum range of twenty percentage points.

Figure 8.1 gives us the first example of the markers for the top 20 percent and the bottom 30 percent. In this instance, there is very little difference between their values at the beginning of the time series in 1977 and the values for all of Belmont and Fishtown. The changes over time are not importantly affected by the changing proportions of people in Belmont and Fishtown from the 1970s through 2010.

On other GSS items relating to marriage, the social classes became more alike, not more different. In the 1970s, large majorities in Fishtown thought that premarital sex was wrong, that the wife should help her husband’s career first, and that young children suffer if the mother works. Among the college-educated people of Belmont, support for all these propositions was much lower. By the 2000s, support had dropped everywhere, but most of all in Fishtown, so that there was little remaining difference between Belmont and Fishtown on most of them.

In two respects, Belmont did most of the moving, approaching Fishtown’s position. The first involved attitudes toward divorce. Over the decades, growing numbers of people in Belmont agreed that divorce law should make divorce more difficult, almost erasing the gap with Fishtown that had existed in the 1970s. The second and the most striking change was that Belmont became more traditional in its attitude toward married people having sex with someone other than their spouses, as shown in Figure 8.2. I put the estimate for the first half of the 1960s at 80 percent overall, for reasons explained in appendix D.

Before getting to the convergence, take a close look at the huge class differences that had emerged on this issue by the 1970s. Based on collateral evidence such as the Gallup survey of American women, we have to assume that in the early 1960s Belmont was about as strict in its attitudes as Fishtown. Within just a few years, white college-educated men and women became enthusiastic recruits to the sexual revolution. It is one of the most dramatic and rapid examples of divergence of elite norm and mainstream norms. It is also clearly concentrated among the college educated—note the difference between acceptance of extramarital sex among the college educated people of Belmont compared to the people in the top 20 percent, who in the 1970s still included many who were not college educated.

FIGURE 8.2. IS EXTRAMARITAL SEX WRONG?

Source: GSS. Sample limited to whites ages 30–49. Data smoothed using locally estimated regression (LOESS).

During the 1980s, the percentage of Belmonters who said that extramarital sex was always wrong began to rise and continued to do so. By the 2000s, Belmont still was not quite as strict on this point as Fishtown, but college-educated professionals had clearly returned to a more traditional attitude than they had held in the 1970s. While class differences remained in attitudes toward marriage, many of these differences were smaller in 2010 than they had been in the 1970s. And yet actual behavior regarding marriage diverged sharply. It is time to tell that story.

What Whites Did About Marriage

The Decline of Marriage

Starting around 1970, marriage took a nosedive that lasted for nearly twenty years. Among all whites ages 30–49, only 13 percent were not living with spouses as of 1970. Twenty years later, that proportion had more than doubled, to 27 percent—a change in a core social institution that has few precedents for magnitude and speed. Figure 8.3 uses the 1960 decennial census and the Current Population Survey for 1968–2010 (hereafter referred to as the CPS database) to show how the prevalence of marriage changed among the people of Belmont and Fishtown.

FIGURE 8.3. MARRIAGE

Source: IPUMS. Sample limited to whites ages 30–49. “Married” refers to persons married and not separated.

In 1960, the proportions of married couples in Belmont and Fishtown were separated by about 10 percentage points, but both were high—94 percent in Belmont and 84 percent in Fishtown. Nothing much changed in the 1960s. A sexual revolution may have been under way among the twentysomethings, but the proportions of whites in their thirties and forties who were married in 1970 were within a percentage point of their 1960 levels. Then, beginning during the last half of the 1970s, the neighborhoods started to diverge. By the mid-1980s, the decline had stopped in Belmont, and the trendline remained flat thereafter. Marriage in Fishtown kept falling.

The net result: The two neighborhoods, which had been only 11 percentage points apart as late as 1978, were separated by 35 percentage points as of 2010, when only 48 percent of prime-age whites in Fishtown were married, compared to 84 percent in 1960. Furthermore, the slope of the decline in Fishtown after the early 1990s had yet to flatten.

The Rise of the Never-Marrieds

People ages 30–49 are unmarried for two main reasons: They are divorced or they never got married in the first place (widowhood at that age is rare). I begin with the never-marrieds.

The percentage of whites ages 30–49 who had not yet married started going up in the early 1970s, doubling from 1977 to 1991. Figure 8.4 shows how differently those increases played out in Belmont and Fishtown.

FIGURE 8.4. REMAINING SINGLE

Source: IPUMS. Sample limited to whites ages 30–49.

The stereotypes of the 1970s and 1980s, of yuppies and feminists remaining single into their thirties or forties, had some basis in fact—the percentage of never-married whites in Belmont doubled from 1970 to 1984. But after 1984, that percentage barely rose at all, from 9 percent to 11 percent. The big news is the relentless increase in Fishtown of people who had never married. It showed no signs of decreasing through 2010, when more than one out of four Fishtown whites ages 30–49 had not yet married. That increase was driven mostly by the retreat of men from the marriage market. As of 2010, almost one out of three Fishtown males ages 30–49 had not yet married.

The Rise of Divorce

Divorce played about an equal role with the never-marrieds in explaining the overall class divergence in marriage. The story is shown in Figure 8.5 for people who have ever married (excluding those who are widowed).

FIGURE 8.5. DIVORCE

Source: IPUMS. Sample limited to whites ages 30–49 who have married and are not widowed.

It is a predictable story, given what we have already seen about the decline in marriage—similarity between the two neighborhoods for a while, then divergence. In the case of divorce, the trends were similar into the early 1980s. The trendline in Belmont flattened in the early 1980s. In Fishtown, the trendline continued steeply upward, with the slope shallowing only a little in the 2000s. As of 2010, one-third of Fishtown whites ages 30–49 had been divorced.

Happy and Not So Happy Marriages

One other divergence among the classes with regard to marriage needs to be mentioned before moving on. Not only did marriage become much rarer in Fishtown over the half century ending in 2010, the quality of marriages that did exist apparently deteriorated. Since 1973, the GSS has asked, “Taking all things together, how would you describe your marriage?” and given the respondent the choice of answering “very happy,” “pretty happy,” or “not too happy.” The results by decade are shown in Figure 8.6. Based on the 1962 Gallup survey for the Saturday Evening Post, I put the estimate of people saying they had very happy marriages in the first half of the 1960s at 63 percent.

FIGURE 8.6. SELF-REPORTED “VERY HAPPY” MARRIAGES AMONG THOSE MARRIED AND NOT SEPARATED

Source: GSS. Sample limited to married whites ages 30–49. Data smoothed using locally estimated regression (LOESS).4

If the estimate for the first half of the 1960s is correct, the implication is that the proportion of happy marriages increased during the late 1960s and early 1970s. It may well be true—the introduction of no-fault divorce in the late 1960s and the surge in divorces that followed ended a lot of unhappy marriages. But perhaps the wording of the 1962 Gallup question, which asked if people were extremely happily married, compared to the GSS’s milder very happily married, means that I have substantially underestimated the proportion of happy marriages in the early 1960s (see appendix D for a discussion of that issue).

In any case, the story from the 1970s onward is reasonably clear. In Belmont, the percentage of people saying their marriages were very happy was on an upward trend after the 1980s. Self-reported happy marriages in Fishtown declined. For the surveys in the 2000s, the gap with Belmont had reached about 20 percentage points.

Children and Marriage

Trends in marriage are important not just with regard to the organization of communities, but because they are associated with large effects on the socialization of the next generation. No matter what the outcome being examined—the quality of the mother-infant relationship,5 externalizing behavior in childhood (aggression, delinquency, and hyperactivity),6 delinquency in adolescence,7 criminality as adults,8 illness and injury in childhood,9 early mortality,10 sexual decision making in adolescence,11 school problems and dropping out,12 emotional health,13 or any other measure of how well or poorly children do in life—the family structure that produces the best outcomes for children, on average, are two biological parents who remain married. Divorced parents produce the next-best outcomes. Whether the parents remarry or remain single while the children are growing up makes little difference. Never-married women produce the worst outcomes. All of these statements apply after controlling for the family’s socioeconomic status.14 I know of no other set of important findings that are as broadly accepted by social scientists who follow the technical literature, liberal as well as conservative, and yet are so resolutely ignored by network news programs, editorial writers for the major newspapers, and politicians of both major political parties. In any case, the change in the family structure in which the children of Fishtown grow up has been huge.

Children Living with a Single Divorced or Separated Parent

Figure 8.7 shows the trends for children living in single-parent homes consisting of a divorced or separated parent.15

The trends roughly correspond to the trends in divorce shown earlier. The divergence between Belmont and Fishtown is substantial, with 22 percent of Fishtown children living with a lone divorced or separated parent as of 2010, compared to just 3 percent of Belmont children. Divorce isn’t the biggest problem that the children of Fishtown face, however. The problem is never-married mothers.

FIGURE 8.7. CHILDREN OF BROKEN MARRIAGES LIVING WITH A SINGLE PARENT

Source: IPUMS. Sample limited to married whites ages 30–49.

Nonmarital Births

From the founding until well into the twentieth century, it was unquestioned that children should be born only within marriage and that failure to maintain that state of affairs would produce catastrophic consequences for society. That universal understanding explains why children born out of marriage were called by an invidious name, bastards; had diminished legal standing; and were so relentlessly stigmatized that even children of unmarried women who rose to eminence (Alexander Hamilton, for example) felt the sting of that stigma all their lives.

In the twentieth century, illegitimate supplanted bastard as the favored label for children born out of wedlock, helped along by the imprimatur of one of the first great anthropologists, Bronisław Malinowski. In his 1930 book, Sex, Culture, and Myth, Malinowski concluded that the “principle of legitimacy” amounted to a “universal sociological law.” Every culture, he concluded, had a norm that “no child should be brought into the world without a man—and one man at that—assuming the role of sociological father, that is, guardian and protector, the male link between the child and the rest of the community.” Without that man, Malinowski wrote, “the group consisting of a woman and her offspring is sociologically incomplete and illegitimate.”16

The last half of the twentieth century saw the creation of cultures that broke Malinowski’s universal sociological law. For the first time in human history, we now have societies in which a group consisting of a lone woman and her offspring is not considered to be sociologically incomplete—not considered to be illegitimate—and so I will adapt and call them nonmarital births.

In America, white nonmarital births have grown phenomenally over the period 1960–2010. To understand just how aberrational

FIGURE 8.8. WHITE NONMARITAL BIRTH RATIO FROM 1917 TO 2008

Source: For 1917–39, National Center for Health Statistics, 1941, table Q. For 1940–60, Grove, 1968, table 29. For 1960–2008, annual Vital Statistics reports of the National Center for Health Statistics.

1960–2010 was, you have to see it in the context of the last century, shown in Figure 8.8.

The shaded area contains the decades we are studying, 1960–2010, when the percentage of nonmarital births rose steeply throughout. But before that, hardly anything had changed since the first numbers were collected in 1917. Studies of the white family in earlier eras indicate that the line hugging the bottom of the graph from 1917 to 1960 would have been flat all the way back to the Revolution.17 White children were conceived outside marriage at varying rates in different social classes, but hardly ever born outside marriage in any class.

To see which white women were having those babies, I turn to the national birth records assembled by the National Center for Health Statistics (NCHS). Since 1970, the NCHS has included information on the mother’s years of education at the time of birth. The breakdown of nonmarital births by education is shown in Figure 8.9.

FIGURE 8.9. WHITE NONMARITAL BIRTH RATIO BY MOTHER’S EDUCATION

Source: Author’s analysis of alternate years of the Natality Public Use Files of the Centers for Disease Control, beginning with 1970. Sample limited to white women.

That information reveals an extraordinarily strong relationship between the mother’s education and the likelihood that she gives birth as an unmarried woman. If she has a college education, she almost never does. Whether she has a graduate degree makes no difference—the trendlines for women with bachelor’s degrees and graduate degrees are indistinguishable. Even in the most recent data from 2008, fewer than 5 percent of babies born to women with sixteen or more years of education were nonmarital. But anywhere below sixteen years of education, the increase in the likelihood of a nonmarital birth was substantial. For women who did not finish high school, the percentage was closing in on levels in excess of 60 percent of live births that previously have been associated with the black underclass.

There is no way to translate these data into precise breakdowns for Belmont and Fishtown, partly because we lack any occupational data and partly because of a major interpretive problem. Women with high school educations can be assigned to Belmont because they are married to men with college educations and a Belmont occupation. It seems highly unlikely that this population of women has the same probability of having experienced a nonmarital birth as women with high school educations who remain unmarried or who marry a man with a high school education and a Fishtown occupation.

Despite this interpretive problem, we know that the shape of the trends in Figure 8.9, which are based on educational attainment at the time of birth, wouldn’t look much different if they were based on the woman’s ultimate educational attainment. Occasionally women have babies and later go back to school, but not enough of them to make Figure 8.9 look much different.

Our Rosetta stone for knowing such things is the experience of the women in the 1979 cohort of the National Longitudinal Survey of Youth (NLSY-79), who were in their prime childbearing years from the late 1970s to the mid-1990s. As shown in table 8.1, their nonmarital birth ratios based on education when their children were born and based on their education when they had reached age 40 are quite close.

TABLE 8.1. NONMARITAL BIRTH RATIO BY MOTHER’S EDUCATION

Nonmarital birth ratio based on …
Highest grade completed Mother’s years of education at the child’s birth Mother’s years of education by age 40
16 years and more 2.9% 3.5%
13-15 years 5.9% 9.1%
12 years 12.1% 12.8%
Fewer than 12 years 21.2% 18.8%

Source: NLSY-79. Sample limited to children of white mothers.

So whereas I cannot calculate precise numbers for the trends in Belmont and Fishtown, you can get a good idea of what they would look like by imagining a line for Belmont that is close to the line for women with sixteen or more years of education, but slightly higher, and a line for Fishtown that is moderately higher than the line for women with twelve years of education. My best estimate is that nonmarital births in Belmont as of 2008 were around 6 to 8 percent of all births, whereas in Fishtown they were around 43 to 48 percent of all births.

Maybe It Isn’t as Bad as It Looks

There are two reasons why this portrait of the breakdown of the family in working-class white America might not be the disaster that I have inferred. One is statistical. People with lower levels of education marry at younger ages and have babies at younger ages than people who are busy with school through most of their twenties. If we control for these differences, how different would the results in this chapter look? The answer (not much) is discussed in appendix D. The other is a hot topic in today’s America: cohabitation. The old-fashioned dichotomy between married and unmarried is unrealistic in today’s world, the argument goes. People may cohabit rather than formally marry, but the children are still being raised in a two-parent family, with the advantages of a two-parent family.

The increase in cohabitation has been rapid and large. For the last two decades, a majority of people in their twenties and thirties have cohabited.18 In the 1990s, about 40 percent of all births to single women actually occurred to women who were cohabiting with the biological father of the baby, and presumably that percentage grew during the 2000s.19 Statistically, almost all of the increase in nonmarital childbearing in the last few decades is explained by an increase in children born to cohabiting parents.20

The question then becomes: How do the children of cohabiting parents fare? The answer: About the same as the children of the old-fashioned form of single parenthood, women who are unmarried and not cohabiting.

The differences begin in infancy, when most of the cohabiting couples are still living together and the child has a two-parent family. Stacey Aronson and Aletha Huston used data from a study of early child care conducted by the National Institute of Child Health and Human Development to assess the mother-infant relationship and the home environment for children at ages 6 months and 15 months.21 On both measures and at both ages, the children of married couples did significantly better than the children of cohabiting parents, who in turn had scores that were only fractionally higher than the children of single mothers. The differences could only be attenuated, not explained away, when other demographic variables were entered into the analysis.22 But the demographic variables made a difference, too. The mothers in cohabiting couples tended to have lower education, to be younger, to have poorer psychological adjustment, less social support, and less money than the married mothers. Those factors statistically explain some of the difference—but they make no difference at all to the divergence of the social classes. Cohabiting mothers come disproportionately from the lower socioeconomic classes and they tend to provide worse environments for raising children than married mothers. That’s not only the reality on the ground, shaping the environments in the neighborhoods where they live, it is a reality that is likely to accelerate the deterioration of those neighborhoods as the children reach adulthood. Examples will be found when it comes time to discuss life in the real Fishtown in chapter 12.

The disadvantages of being born to cohabiting parents extend into childhood and adolescence, even when the cohabiting couple still consists of the two biological parents. Susan Brown used the 1999 cohort for the National Survey of America’s Families to examine behavioral and emotional problems and school engagement among six- to eleven-year-olds and twelve- to seventeen-year-olds. Same story: Having two unmarried biological parents was associated with worse outcomes than having two married biological parents, and the outcomes were rarely better than those for children living with a single parent or in a “cohabiting stepparent” family.23 Once again, entering additional variables explained some but not all of the difference, but those additional variables revealed the same story that others have found—there is a strong inverse relationship between socioeconomic status and the likelihood that children are born to cohabiting women. Cohabitation with children occurs overwhelmingly in Fishtown.

Cohabitation has been a common feature in American life for more than two decades, and it may be asked whether there are signs that cohabitation itself will evolve for the better. Not so far. The two studies I cited were the most recent available as I write, but there is also a literature from studies conducted in the 1990s and 1980s.24 The story seems to be consistent. If you are interested in the welfare of children, knowing that the child was born to a cohabiting woman instead of a lone unmarried woman should have little effect on your appraisal of the child’s chances in life. That’s the common theme of the systematic studies of this issue for more than twenty years.

It’s Even Worse Than It Looks

The pessimistic title of this section springs from my belief that families with children are the core around which American communities must be organized—must, because families with children have always been, and still are, the engine that makes American communities work—and from my conclusion that the family in Fishtown is approaching a point of no return. The extent of the collapse of the Fishtown family may not be evident from the separate pieces that I have presented, so let me conclude this chapter with two summary measures.

The first, Figure 8.10, is based on whites ages 30–49 who are in happy marriages, expressed as a percentage of all whites ages 30–49 (not just married whites, as shown in Figure 8.6).

If the issue were happiness, the steep decline in Fishtown would not be as bad as it looks. Many people who are not married are happy, including people who are divorced. Instead of thinking about percentages of happy and unhappy people, think about the role of marriage as the bedrock institution around which communities are organized and, writ large, around which the nation is organized. A neighborhood in which that function is being performed will be characterized by a large core of happy marriages.

FIGURE 8.10. PROPORTION OF ALL WHITES AGES 30–49 WHO SELF-REPORT BEING IN VERY HAPPY MARRIAGES

Source: GSS. Sample limited to whites ages 30–49. Data smoothed using locally estimated regression (LOESS).

When we get our first look at the situation by social class in the 1970s, over half of both neighborhoods consisted of people in happy marriages. In the 2000s, such people still constituted almost 60 percent of the prime-age whites of Belmont. But the percentage in Fishtown had been halved, from 52 percent in the GSS surveys of the 1970s to 26 percent in the surveys of the 2000s. Twenty-six percent is arguably no longer a large enough group to set norms or to serve as a core around which the community functions. Fishtown in 2010 was rudderless in a way that it had not been in the 1970s and earlier.

My second summary measure is the percentage of all children who are raised with both biological parents. We cannot obtain that measure from CPS data because the CPS does not discriminate between families consisting of married and remarried couples. Instead, I turn to three National Longitudinal Surveys to reconstruct the trendline for children living with both biological parents: the Mature Women Survey, whose subjects turned age 40 from 1964 to 1977, the Young Women Survey, whose subjects turned 40 from 1982 to 1993, and the 1979 National Longitudinal Survey of Youth, whose subjects turned 40 from 1997 to 2004. Figure 8.11 shows the trendlines for Belmont and Fishtown.

FIGURE 8.11. PERCENTAGE OF CHILDREN LIVING WITH BOTH BIOLOGICAL PARENTS WHEN THE MOTHER WAS AGE 40

Sources: NLS Mature Women, NLS Young Women, NLSY-79. Top 20 percent and bottom 30 percent are based on women who were age 40 in 1963 and 2004.

Fishtown’s higher divorce rate and much higher nonmarital-birth ratio combined to produce wide divergence from Belmont; this divergence continued to widen at the end of these observations. For the NLSY-79 cohort, whose mothers turned age 40 between 1997 and 2004, the percentage of children living with both biological parents when the mother was 40 was sinking below the 30 percent level, compared to 90 percent of Belmont children who were still living with both biological parents. The divergence is so large that it puts the women of Belmont and Fishtown into different family cultures. The absolute level in Fishtown is so low that it calls into question the viability of white working-class communities as a place for socializing the next generation.