PROMISES THEY CAN KEEP: LOW-INCOME WOMEN’S ATTITUDES TOWARD MOTHERHOOD, MARRIAGE, AND DIVORCE

Abstract

Using survey data on low-income mothers in Boston, Chicago, and San Antonio (n = 1,722) supplemented with ethnographic data, we test 3 propositions regarding mothers’ attitudes toward childbearing, marriage, and divorce. These are drawn from Edin & Kefalas (2005) but have also arisen in other recent studies. We find strong support for the proposition that childbearing outside of marriage carries little stigma, limited support for the proposition that women prefer to have children well before marrying, and almost no support for the proposition that women hesitate to marry because they fear divorce. We suggest that mothers’ attitudes and preferences in these 3 domains do not support the long delay between childbearing and marriage that has been noted in the literature. Throughout, we are able to study attitudes among several Hispanic groups as well as among African Americans and non-Hispanic Whites.

The question of why some low-income women and men have children years before they marry, if they ever marry, has interested social scientists for a century, with most of the attention focused on African Americans. Before World War II sociologists such as DuBois (1908), Frazier (1939), and Drake and Cayton (1945) noted the effects on black family life of the heritage of slavery, urbanization, segregation, and discrimination, sometimes drawing cultural distinctions between the poor and the middle class. In mid-century, ethnographic accounts (Liebow, 1968, Hannerz, 1969, Ladner, 1971; Stack, 1974) and the controversial analysis of Moyhihan (1965) continued this tradition. The issue received further attention in the 1990s (Tucker & Mitchell-Kernan, 1995; Franklin, 1997). More recently, the question has reemerged as part of the policy debates over government programs to promote marriage among the low-income population (Amato & Maynard, 2007).

In response to this resurgence of interest, a small but influential literature has emerged that examines contemporary low-income women’s attitudes and preferences toward life-course pathways involving childbearing and marriage. The most widely-cited contribution to this literature is Edin and Kefalas (2005), a qualitative study of low-income mothers in the Philadelphia area. They argue that a radical separation of childbearing and marriage exists in the life course of the typical low-income woman in the neighborhoods they studied. Young women, they claim, are unwilling to postpone childbearing because of the high value they place on children, even though it is difficult for them to find suitable young men to marry. Many of them, consequently, begin bearing children as teenagers. Moreover, it is argued, many of them prefer to marry at a substantially older age, such as in their thirties. The authors also claim that women are hesitant to marry because, should the marriage fail, they fear the shame and stigma of divorce.

Edin and Kefalas focused on answering policy questions rather than on placing their findings in theoretical context. Yet their study is relevant to life course theory, which asserts that norms and expectations about the timing and ordering of events can serve as points of reference, guiding individuals through the life course in a socially prescribed order. Individuals are said to benefit when life events occur “on time” (e.g., within a normatively expected age-range) and in a socially-prescribed sequence (e.g., marriage before childbearing). On-time, normatively-sanctioned transitions are less stressful and usually occur in the context of reasonable social support (Hagestad & Neugarten, 1985; Elder & Shanahan, 2006). However, when an individual experiences an event or transition outside of what the social context considers the appropriate order and “off-time,” less social support may be available to them and the consequences of the transition may be negative. Applications of the life course perspective to the timing and sequencing of childbearing and marriage in low income populations are well-represented in the research on non-marital pregnancy in the 1980s and 1990s (Hogan & Kitagawa, 1985; Hamburg, 1986, Furstenberg, Brooks-Gunn, & Morgan, 1987; Burton, 1990).

Until the last third of the twentieth century, a strong norm against having a first child prior to marrying existed. Even if a woman became pregnant before marrying, she and her partner often married whether they had planned to or not – in the latter case, submitting to a so-called shotgun wedding. The norm began to weaken after mid-century: Census data show that the percentage of brides who were pregnant at the time of their wedding dropped sharply, particularly among African Americans, between the early 1960s and the mid-1970s as the acceptability of non-marital childbearing increased (U.S. Bureau of the Census, 1978). Still, the belief that it is preferable to have children within marriage seemed widespread throughout most of the twentieth century. Edin and Kefalas’s findings suggest, however, that low-income women’s attitudes and practices regarding childbearing and marriage today differ sharply from the conventional view of how the life course should proceed in adolescence and early adulthood.

Edin and Kefalas also contributed to the literature by moving the field beyond an exclusive focus on African Americans. They included low-income Puerto Ricans and non-Hispanic whites; and their findings were similar across these socially-disadvantaged groups. Puerto Ricans were the only Hispanic group represented because no other Hispanic group was present in sufficient numbers. Other studies have been similarly limited by the exclusion of couples in which both partners spoke only Spanish (Edin & Reed, 2005; Gibson-Davis, Edin, & McLanahan, 2005) or by only interviewing whites and African Americans (Waller, 2002). There is reason to expect substantial variation in attitudes among mothers from the different sub-groups that constitute the category of Hispanic. It is well-known that Americans of Mexican origin have tended to marry at earlier ages than their generally modest economic circumstances would predict (Oropesa, Lichter, & Anderson, 1994). Mexico has been viewed as a country with a high cultural value on marriage relative to other Latin American countries and to the United States (Oropesa & Landale 2004). About half the current population of Mexican-origin Americans were born in Mexico (U.S. Bureau of the Census, 2005), and they may bring these values with them to the United States, resulting in an earlier age at marriage, at least for first generation immigrants. Puerto Ricans and Dominicans, the second and fourth largest Hispanic groups (Cubans being the third), come from a Caribbean region in which consensual unions and childbearing outside of marriage are more common. Oropesa (1996) found significant differences in marital attitudes between Puerto Ricans and Mexicans, with Mexicans being more pro-nuptial but Puerto Ricans being only marginally different from whites. Consequently, one might expect a different relationship between childbearing and marriage among Mexicans, Puerto Ricans, and Dominicans. Moreover, there is little information in this literature about low-income non-Hispanics whites.

Do contemporary low-income women hold the attitudes and preferences that Edin and Kefalas and others say are helping to greatly transform their adolescent and young adult lives? To what extent do norms about the ordering and sequencing childbearing and marriage appear to have changed? In this article, we will test 3 propositions about women’s attitudes toward motherhood, marriage, and divorce that are drawn from Edin and Kefalas’s work but which are also relevant to the findings from other studies. We will use data from the Three-City Study of low-income families in Boston, Chicago, and San Antonio, which provides information not only on the frequently-studied African American population but also on several Hispanic groups and on non-Hispanic whites. It is of interest to determine whether the findings from qualitative studies involving non random samples, such as Edin and Kefalas’s, can be confirmed in a large, random sample survey of low-income families.

The first proposition is that in low-income neighborhoods little stigma is attached to bearing a child outside of marriage and that, correspondingly, women who start childbearing at early ages do not believe they will be penalized in the marriage market. The low-stigma argument can be found in classic works on African American families such as Drake & Cayton (1945) and Ladner (1971). These studies, however, maintained that bearing a child outside of marriage produced short-term shame and embarrassment that was overcome by the accommodating and tolerant values of the community. For example, Ladner (1971, pp. 217–218) writes that a teenager’s giving birth outside of marriage was seen as a mistake because it was “an act that was completely against the morals of the larger society” (emphasis in original); however, there was compassion for the mother and a sense that she had been “unlucky.” Two decades later Kaplan (1997) reported that a majority of the mothers of the 32 adolescent mothers she studied in Oakland in 1985 reacted angrily to the news that their daughters were pregnant and that all except one initially demanded that their daughters get abortions. Although the prevailing research on Latino and White adolescent mothers is less extensive, comparable findings have been reported for these populations (Howell, 1973; Becerra & de Anda, 1984; Romo & Falbo, 1996). For example, Gonzalez-Lopez (2005, p 99), reporting ethnographic findings of the sex lives of Mexican immigrants, writes that a teenage daughter’s pregnancy, “fractured family honor and caused shame, and the moral damage done to the family had to be repaired.” A growing number of studies, however, do not even suggest short-term shame and stigma. For example, Edin and Kefalas state that the mothers of adolescents in their study reacted with disappointment but rarely with anger.

The second proposition is that, given the constraints of their lives, low-income women see childbearing as an activity that occurs well before marriage – sometimes a decade or two beforehand. They do not necessarily think that having children during adolescence is optimal. “Everyone, including the poor,” write Edin and Kefalas, “acknowledges that having children while young and not yet finished with schooling is not the best way to do things” (p. 65). However, the authors state, having children young is far preferable to waiting for the uncertain prospect of marriage at a later date, given the social and economic situation young women face: “While most girls don’t plan to become mothers at fourteen, they almost all agree that no reasonable woman would postpone childbearing until her thirties” (p. 35) – her “thirties” being the stage of life when marriage might occur. Childbearing therefore starts early: “Many believe that the ideal time for childbearing is between the late teens and the mid-twenties” whereas marriages comes much later, “somewhere between the ages of 25 and 40” (p. 109). The authors imply that women’s desire to have children is strong enough that, in the absence of good marriage prospects and with little stigma to fear, they prefer to have children young and to marry at substantially older ages, if they marry at all. But the concept of an “ideal” time that starts as early as one’s late teenage years may overstate the extent to which early childbearing is a conscious choice for low-income adolescents. Rather, young women who are disengaged from school and without goals may drift into pregnancy or be coerced into sexual activity (Hill, 2005). Consistent with the notion of drifting into pregnancy, Edin and Kefalas report that nearly half the women they studied said that becoming pregnant was neither fully planned or fully unplanned but “somewhere in between – something that “just happened” as a result of sexual activity without contraception.

The third proposition is that low-income women fear divorce and that this fear is an important factor in their reluctance to marry. They fear divorce so much, it is said, that they will not marry unless they are sure the marriage will last a lifetime. They hold marriage in such high esteem that they believe that a person who marries but soon divorces would be embarrassed before friends and family. Thus, although there is little stigma in having a child without marrying, there is substantial stigma in failing at one’s marriage – such shame that some women hesitate to marry. Several scholars have recently put forth this contention. In one study based on qualitative interviews with low-income, unmarried parents in the Trenton, New Jersey, area, the author observed:

Parents typically framed the decision not to marry in terms of minimizing the high likelihood of divorce. Citing numerous stories of failed marriages, the parents I interviewed suggested the exposure to divorce in their own families and in the larger society had made them approach marriage more cautiously…Rather than embracing a casual attitude toward marriage, parents maintained that they postponed marriage precisely because they thought marriage should last ‘forever’…(Waller, 2002, p. 27).

Edin and Kefalas also advanced this position. In the metaphor that became the title of their book, Promises I Can Keep, the authors argued that women avoid marriage because they are not sure they can keep the sacred promise to remain with their husbands until death, whereas they are sure they can meet their obligations to care for their children: “Most poor women we spoke with say that it is better to have children outside of marriage than to marry foolishly and risk divorce, for divorce desecrates the institution of marriage” (p. 207). Edin and Reed (2005), after reviewing qualitative interviews with 49 unmarried couples in the Fragile Families study, reported that in 53 percent of them, “one or both partners say their fear of divorce is part of what is keeping them from getting married” (p. 125).

If fear of shame and embarrassment over being divorced is, in fact, significantly influencing women’s family patterns, it would represent a reversal of a long-term trend toward greater acceptance of divorce in the United States. Divorce has been available since the colonial era, but until the twentieth century, it was awarded on limited grounds and was a mark of failure – a last resort (Hartog, 2000). The sense of failure and shame persisted through the mid-twentieth century, but then attitudes shifted. When a sample of young women were asked in 1961 whether they agreed with the statement, “Divorce is usually the best solution when a couple can’t seem to work out their marriage problems,” 61 percent agreed. By 1977, when the same women were reinterviewed, 80 percent agreed (Thornton and Young- DeMarco, 2001). Other survey responses also suggest a movement toward more liberal attitudes toward divorce in the late 1960s and early 1970s and little change since then (Cherlin, 1992). Moreover, people with less education seem to be more likely to approve of divorce. In a 2002 national study of women of childbearing age, the respondents were asked whether they agreed with the same statement that divorce is usually the best solution. Sixty-five percent of those without a high-school degree said they agreed or strongly agreed with it, compared to 43 percent of college graduates (U.S. National Center for Health Statistics, 2006). It would seem unlikely, then, that low-income women would be avoiding marriage because of a pervasive fear of the shame of getting divorced.

DATA AND METHODS

In this paper we will report on the results of a survey of the well-being of children and their families who were residing in low-income neighborhoods in Boston, Chicago, and San Antonio, with supplementary information from an ethnographic sample of families in the same neighborhoods. The Three-City Study, as this project has come to be known, began in 1999 with a random-sample survey of 2,402 children and their caregivers. The survey was conducted as follows: In households in low-income neighborhoods (93 percent of the selected block groups had poverty rates of 20 percent or more) with a child age 0 to 4 or age 10 to 14, with a female primary caregiver, and with incomes below 200 percent of the federal poverty line, interviewers randomly selected 1 child and conducted in-person interviews with that child’s primary caregiver (a mother in over 90 percent of the cases). Families receiving benefits from Temporary Assistance for Needy Families (TANF), the main cash welfare program, were over-sampled. The response rate was 74 percent. The survey data are weighted to correct for over-sampling and to give equal weight to the experiences of families in each city (see Cherlin, Fomby, & Moffitt, 2002). Because the sample is restricted to 3 cities, it cannot provide nationally-representative data. On the other hand, it created the opportunity to conduct an ethnographic study in the same neighborhoods as the survey; it provided substantial numbers of respondents from several of the largest Hispanic groups; and it allowed for within-city comparisons of caregivers from racial and ethnic groups while controlling for welfare-reform rules (which vary by state and sometimes locality).

For the 2005 survey wave, a set of questions was designed to test the 3 propositions described above. Eighty percent of the families that had been interviewed in 1999 were successfully reinterviewed in the 2005 wave. There were no significant differences between the 1999 characteristics of the caregivers who were reinterviewed in 2005 and the 1999 characteristics of those who were not reinterviewed in 2005 with respect to age, race and Hispanic ethnicity, educational attainment, welfare usage, or marital status. To be consistent with the qualitative literature, we further restrict the 2005 survey sample to 1,722 caregivers who were the biological mothers of the target child. Table 1 displays a comparison of the characteristics of the survey sample as of 2005 and the qualitative sample studied by Edin and Kefalas. The 2 samples contain comparable percentages of African American mothers, but the survey sample contains a much higher percentage of Hispanics and, correspondingly, a much lower percentage of Whites.

Table 1.

Characteristics of the Three-City Study Survey Sample (n=1722) and the Edin and Kefalas Sample (n=167)

	Three-City Study	Edin and Kefalas
Race/ethnicity
African-American	41%	39%
Hispanic	51%	29%
Non-Hispanic White	8%	32%
Mean age	36	26
Age at first birth	20	18
First birth as a teenager	53%	72%
Mean number of children	3.3	2.2
Education
Less than high school	35%	44%
GED	13%	14%
High school diploma	20%	8%
Some college	31%	31%
Ever-married	44%	14%
In a relationship currently	32%	65%
Work Status
Not working or in school	38%	48%
Working part-time	20%	21 %
Working full-time	36%	19%
In school	11%	--
Received TANF in last 2 years	17%	49%
Ever received TANF/AFDC	80 %	--

The large Hispanic sub-sample in the survey allowed us to analyze Hispanic attitudes in greater detail than in previous studies of this topic. The respondents in the 2005 wave of the survey, which was conducted in English and Spanish, included mothers who were of Mexican (n = 445), Puerto Rican (n = 217), or Dominican (n = 117) origin, and a diverse group we labeled as “other Hispanic” (n = 90). (Thirty of the “other Hispanic” mothers said they were from Central-American countries, 8 identified as Cuban, and smaller numbers mentioned other countries or did not name a country.) The number of mothers of Mexican origin was sufficient to allow us to differentiate mothers born in the United States (n = 320) from mothers born in Mexico (n = 125). Nearly all of the Dominicans were foreign-born, and 52 percent of the Puerto Ricans were born on the island. We also included a scale of English language competence based on responses to 3 questions asked of women who said that English was not their first language: “How well do you speak English?” “How well do you read English?” and “How well do you write English.” Response categories ranged from not at all (= 1) to very well (= 4), yielding a 3-item scale with an alpha reliability of 0.90. We assigned people whose first language was English (and who therefore were not asked these 3 questions) the maximum score. We would caution, however, that the small non-Hispanic White sub-sample (n = 143) is probably less representative of all low-income mothers of the same ethnicity than is the case for the Hispanic and African American (n = 710) mothers because White poverty is less concentrated in central city neighborhoods (Massey and Eggers,1990; Krivo et al., 1998).

Table 1 also shows that the mothers in the survey sample were 10 years older, on average, than the mothers in the Edin and Kefalas sample. As befits an older sample, they had more children, on average; more of them had high school degrees (although the percent with some college was comparable); and more had married at some point. In addition, fewer were currently in a relationship. Moreover, fewer were currently receiving TANF; but we would note that the survey sample was conducted 9 years after the implementation of welfare reform, and the TANF participation rates for the survey respondents had dropped sharply since the start of the Three-City Study in 1999. In fact, 80 percent of the women in the survey sample had received TANF or its predecessor, Aid to Families with Dependent Children (AFDC), at some point in their adult lives. These responses suggest that we are studying women who were in a somewhat older life stage – perhaps half a generation – than were the women in the Edin and Kefalas study. They were also interviewed about 10 years after Edin and Kefalas started their fieldwork. It is possible that the attitudes of women in the survey sample could have shifted since they were a decade younger or that Americans’ attitudes in general could have changed since the time of the Edin and Kefalas’s interviews. Nevertheless, we would argue that the survey responses represent a test of whether the climate of opinion in low-income neighborhoods about issues such as stigma and divorce is consistent with the findings of Edin and Kefalas. Moreover, the survey respondents seem to have had similar backgrounds to the Philadelphia mothers: half of them had given birth for the first time as teenagers, and the majority had received welfare at some point in the past. In some sense, then, we are reporting on what a group of women with similar backgrounds to the Philadelphia sample think about childbearing, marriage, and divorce 10 years later.

To examine the amount of stigma associated with childbearing outside of marriage, we used four statements: “Having a child without being married is embarrassing for a woman;” “Having children when a woman is single hurts her chances of later getting married;” “A woman should have children if she wants to, even if she is not married”; and “A woman does not need to be married before having a child.” The 4 questions can be combined into a scale with an alpha reliability of .68, with higher scores indicating greater stigma. We regressed the logarithm of this scale on a set of racial-ethnic categories (native-born Mexican, foreign-born Mexican, Puerto Rican, Dominican, other Hispanic, African American, and non-Hispanic White), competence in English as described previously, age in years, religion (coded Catholic = 1, non-Catholic = 0), education (coded high school degree or more = 1), and city of residence (Boston, Chicago, or San Antonio). We used the logarithm of the scale because it was skewed: Most people, as we will show, responded that the stigma was low. All tabulations and regressions are based on weighted data, which take into account differentials in who was interviewed in 2005 according to sample-design criteria and which weight the sub-populations in each of the 3 cities equally.

In order to determine women’s preferred ages for having children and marrying, the survey included 2 questions, “What do you think is the best age to start having children?” and “What do you think is the best age to get married for the first time?” Respondents were asked to choose among 5-year age categories. To balance the influence of one response on the other, the order in which these 2 questions were presented was randomized in the survey instrument. Finally, to examine the issue of fear of divorce, we included 4 questions about marriage and divorce. The first two reflect whether or not women associate divorce with embarrassment or stigma: “Most people do not care if a woman gets a divorce;” and “Getting divorced is embarrassing for a woman.” The other two are about whether marriages are usually unsuccessful: “Marriage usually changes a relationship for the worse;” and “It’s best to avoid marriage because it usually doesn’t work out.”

We also have a limited amount of information from an ethnographic study of 256 families conducted in the same neighborhoods. They were recruited into the ethnographic sample between June, 1999, and December, 2000; and most had at least one child age 2 to 4. Recruitment sites include formal childcare settings (e.g., Head Start), the Women, Infants and Children (WIC) program, neighborhood community centers, local welfare offices, churches, and other public assistance agencies. Families were visited an average of once or twice per month for 12 to 18 months and then every 6 months thereafter through 2003. The ethnographic study, however, was completed before several of the recent studies, including Edin and Kefalas, were published; and it focused more on the consequences of welfare reform for the mothers’ and children’s day-to-day experiences of living in poverty than on the mothers’ general attitudes about childbearing and marriage, per se. We cannot, therefore, present a full ethnographic analysis of the propositions tested in the 2005 survey.

Still, there are two ways in which we can make modest use of the ethnography. First, we pretested the key 2005 survey questions in follow-up interviews with 37 families in the Boston ethnographic sample and 4 families in the Chicago ethnographic sample. We present mothers’ comments from these follow-ups to elucidate the survey responses. Second, we can examine whether the themes of the 2005 survey questions emerged without prompting during the long, extensive ethnographic field work period among the entire 256 families. If a theme such as fear of divorce is important to individuals, one would expect it to emerge without prompting from time to time. In fact, one of the advantages of ethnographic research is that it allows the subjects, rather than the investigators, to define what the important issues are.

STIGMA

We turn first to the low-stigma proposition. About 80 percent of the mothers disagree or strongly disagree that having a child without being married is embarrassing and that having children when single hurts a woman’s chances of later getting married. About 80 percent agree or strongly agree that a woman should have children if she wants to, even if she is not married; and about 70 percent agree or strongly agree that a woman does not need to be married before having a child. The responses to these 4 questions support Edin and Kefalas’s finding that childbearing outside of marriage carries little stigma among women in low-income neighborhoods.

Table 2 reports the results of a regression of the stigma scale (comprised of the four questions) on the set of variables presented earlier. The results show that older mothers were more likely than younger mothers to think that births outside of marriage were stigmatizing, native-born Mexican mothers were significantly more likely than African American mothers (the reference category) to see births outside of marriage as stigmatizing, and residents of Boston were less likely than residents of Chicago (the reference category) to see births outside of marriage as stigmatizing. Otherwise, the sizes of the coefficients for the right-hand-side variables are small compared to the constant, which suggests that the mothers almost uniformly believed there is little stigma, much as Edin and Kefalas and others have asserted. Even among native-born Mexicans, the most conventional racial-ethnic group in this regard, 80 percent disagreed that having a child outside of marriage is embarrassing and 68 percent agreed that a woman does not need to be married before having a child. (The corresponding figures for foreign-born Mexicans, 86 percent and 70 percent, respectively, were similar.)

Table 2.

Ordinary Least Squares Regression of Stigma Scale on Selected Characteristics (n=1625).

Variable	B	SE B
Age	.0033**	0.00
High school diploma	−.018	0.02
Race/ethnicity:
Native-born Mexican	.12***	0.05
Foreign-born	.020	0.07
Mexican
Puerto Rican	.016	0.05
Dominican	−.041	0.07
Other Hispanic	.032	0.05
Non-Hispanic White	−.0063	0.05
Catholic	−.053	0.03
English proficiency scale	.0052	0.02
Boston	−.082***	0.03
San Antonio	−.051	0.03
Constant	.572***	0.09
R2		0.05
F		3.17***

Note: Dependent variable is the natural logarithm of the score on the stigma scale (mean = 0.65; s.d. = 0.30). African American is the reference category for race/ethnicity. Chicago is the city reference.

^*p <.10.

^**p <.05.

^***p <.01.

The 41 women in the ethnographic follow-up sample also strongly rejected the idea that having a child outside of marriage is stigmatizing. For instance, 90 percent disagreed or strongly disagreed that having a child without being married is embarrassing. A 30-year-old White woman in Chicago commented:

This ain’t 1920, where they put you in a home because they’re, you’re a young girl and you got pregnant. Or you’re 20 years old and you got pregnant and you’re not with the father or whatever. No, I disagree with that.

Eighty-three percent agreed that a woman does not need to be married before having a child. When asked if a woman should have children whether she marries or not, a 26-year-old White woman from Chicago said, “I think basically you should be thinking about if you want a child.” Another White woman in Chicago, age 27, said, “Just because you’re my baby’s father doesn’t mean you should be my husband.” In sum, we find strong support for the low-stigma proposition.

BEST AGES TO HAVE CHILDREN AND TO MARRY

Table 3 presents a crosstabulation of the responses to the two questions on the best times to start having children and to marry. The percentage of the total n in each cell is displayed. The marginal distribution across the bottom of the table shows the responses to the question about the best age to start having children. The marginal distribution down the right side of the table shows the responses to the question about the best age to get married for the first time. Very few women only 4 percent of the survey sample – chose an age under 20 as the best time to start having children. This small percentage is noteworthy because, as we reported earlier, half of the respondents to this question had themselves given birth for the first time as adolescents. Perhaps when these women were teenagers they would have answered differently and the response pattern might have been more consistent with the findings of Edin and Kefalas and others. But looking back on their lives in their late twenties, thirties, or early forties, few endorsed the idea of giving birth as teenagers. This result suggests that there is little cultural support in low-income neighborhoods for the idea that it is best to start having children as teenagers. The most commonly-chosen response category was ages 25 to 29, and 85 percent of the women chose either that category or ages 20 to 24. Relatively few chose age 30 or older. One’s twenties are the best time to start having children, according to these women’s responses, and one’s teenage years are the worst time.

Table 3.

Crosstabulation of responses to questions, “What do you think is the best age to start having children?” and “What do you think is the best age to get married for the first time?” (Cell percentages; n = 1,658)

Best age to marry	Best age to start having children
	Under 20	20–24	25–29	30–34	35 or older	Total
Under 20	1	1	0	0	0	2%
20–24	1	21	6	0	0	28%
25–29	1	10	30	4	0	45%
30–34	0	3	8	6	0	17%
35 or older	0	2	4	1	1	8%
Total	4%	37%	48%	11%	1%	100%

The mothers also chose one’s twenties as the best time to get married. Once again, the modal category was age 25 to 29; and 73 percent chose either that category or 20 to 24. Nevertheless, the distribution was weighted toward older ages more than was the case for the responses about the best age to start having children. One-fourth chose either the category 30 to 34 or the category 35 or older as the best time to marry. Thus, the proportion who think it is best to marry in one’s thirties is greater than the proportion who think the best time to begin having children is in one’s thirties a result which suggests that women are more willing to wait to marry than to have children, as the qualitative literature asserts. Still, the bulk of the sample thinks that one’s twenties are the best time to marry, just as the bulk of the sample thinks that one’s twenties are the best time to start having children.

In fact, 59 percent of the sample chose the same 5-year age category for each of the 2 questions (i.e., their responses placed them in the diagonal cells that go from the top left to the bottom right of the table), which implies that a majority of the mothers thinks the best ages to start childbearing and to marry are within 5 years of each other. The farther one moves from the diagonal (i.e., the more dissimilar the 2 responses), the lower is the percentage of women who fall into a given cell: Thirty-one percent had responses that differed by one 5-year category; 8 percent had responses that differed by two 5-year categories, and just 2 percent had responses that differed by three 5-year categories. These results do not support the idea that low-income women commonly have a preference map in which the ideal is to begin having children a decade or more before marrying. Rather, the vast majority of women’s responses to the 2 questions varied by no more than 1 category and a simple majority of the responses were within 5 years of each other. Nevertheless, if a woman’s responses to the 2 questions differed, she tended to choose a earlier age category for starting to have children than for marrying, which provides some evidence that that unconventional views of the place of marriage in the life course are common.

We investigated racial and ethnic differences in the responses to the 2 questions in the survey using the following method: we divided the mothers into 3 groups. The first consists of those who said the best age to marry was 1 or more age categories younger than the best age to start having children (the above-diagonal cells). For convenience we will label this group as “marriage before childbearing.” The second consists of those who chose the same 5-year age category for the best ages to marry and to start to having children (the diagonal cells). And the third consists of those who said the best age to marry is 1 or more age categories older than the best age to start having children (the below-diagonal cells). For convenience we will label this third group as “childbearing before marriage.” We then estimated a multinomial logistic model with this 3-category classification as the dependent variable. We included the same set of right-hand-side variables as in the stigma scale regression.

The results are presented in Table 4, which has 2 sets of regression coefficients and their corresponding odds ratios (e^β), one for “marriage before childbearing” and one for “childbearing before marriage.” The reference (omitted) group is to marry and start to have children within the same age category. The first column displays the odds of preferring marriage before childbearing, relative to the reference group. We see that mothers who were more proficient in English were significantly more likely to choose this pattern; and mothers who lived in San Antonio were less likely to choose it than were mothers in Chicago, perhaps due to regional differences in values regarding about family life. Two racial-ethnic coefficients were marginally (p <.10) significant: Non-Hispanic White mothers were unexpectedly less likely to choose this option than were African Americans (the omitted racial-ethnic group) – perhaps a sign of the atypicality of the White respondents who live in heavily minority central-city neighborhoods – and native-born Mexicans were less likely to choose this option than were African Americans.

Table 4.

Multinomial Logistic Regression of Preferred Ordering for Best Ages to Start Having Children and to Marry (n=1656). (Reference category: Start to have children and marry within a 5 year period.)

	Marriage Before Childbearing			Childbearing Before Marriage
Predictor	B	SE B	eB	B	SE B	eB
Age	.02	.02	1.02	−.02	.01	.98
High school diploma	.23	.27	1.26	−.23	.18	.80
Race/ethnicity
Native-born Mexican	.95*	.55	2.59	−.85***	.33	.43
Foreign-born Mexican	.60	.66	1.83	−1.01*	.54	.36
Puerto Rican	.16	.53	1.18	−.45	.34	.64
Dominican	.46	.60	1.58	−.99**	.48	.37
Other Hispanic	.03	.56	1.03	−.69	.43	.50
Non-Hispanic White	−1.00*	.56	.37	−.55	.41	.58
Catholic	.06	.39	1.06	−.21	.24	.81
English Proficiency Scale	.47***	.16	1.60	.15	.17	1.11
Boston	−.25	.34	.78	−.03	.22	.97
San Antonio	−.83**	.39	.43	−.39	.25	.68
Constant	−4.10***			.25

Note: African American is the reference category for race-/ethnicity. English Scale scored from 1 for cannot speak, read or write English to 4 for speaks, reads, and writes English well. Chicago is the city reference category.

^*p <.10.

^**p <.05.

^***p <.01.

The second column compares the odds of preferring childbearing before marriage. Here we can see that both native-born and (marginally: p = .06) foreign-born Mexican-origin mothers, as well as Dominican mothers, were less likely to choose this option than were African Americans. In other words, compared to African Americans, mothers in these Hispanic groups were more likely to choose the more conventional response of starting to have children and to marry within the same age category than to choose the less conventional response of having children earlier. These results are consistent with Edin and Kefalas’s observation that African American women were more likely to endorse a life course in which having children occurred well before marrying.

The responses in the ethnographic follow-up sample followed the same pattern. The majority (22 out of 37 valid replies) chose the same age categories for having children as for marriage. Twelve chose categories that correspond to having children more than 5 years before marrying. Just 3 chose categories the correspond to marrying more than 5 years before having children. The comments of a 23-year-old African American woman from Boston who had her first child at age 19 were typical of the majority. On when to start having children, she said, “I think the best age would be, like, 25 to 29 even though I did start early.” On when to marry, she said, “I like to get to know my men, so about 25 to 29.”

FEAR OF DIVORCE

Table 5 presents the responses of the survey sample to the four questions on marriage and divorce. Almost two-thirds of the sample agree or strongly agree with the statement, “Most people do not care if a woman gets a divorce.” (See the “Total” column). Only about one-fourth agree or strongly agree with the statement, “Getting divorced is embarrassing for a woman.” (See the “Total” column.) Moreover, there were no significant differences in the response patterns of Hispanics, African Americans, and non-Hispanic Whites, even when we regressed the responses to these questions individually on the same set of right-hand-side variables as in earlier tables (not shown). Across the board, large majorities of low-income mothers responded in ways which suggest that they do not consider divorce to be stigmatizing or embarrassing. The findings still leave open the question of whether the small group of mothers who say they find divorce embarrassing or stigmatizing are less likely to marry. Table 5 also shows cross-tabulations of the responses to these 2 questions by whether or not the mother is currently married and whether or not she has ever married. The reader can see that there is virtually no relationship between a mother’s answers to these 2 questions and whether or not she is currently married or has ever married.

Table 5.

Attitudes Toward Marriage and Divorce by Marital Status

Question	Currently Married		Ever married		Total
	Yes	No	Yes	No
“Most people do not care if a woman gets a divorce.”
Strongly agree	15	16	16	16	15
Agree	50	50	51	48	50
Disagree	28	30	28	31	29
Strongly disagree	8	5	6	5	6
	101%	101%	101%	100%	100%
χ2	6.02		3.44
n	1,652		1,596		1,652
“Getting a divorce is embarrassing for a woman.”
Strongly agree	7	8	9	7	8
Agree	15	16	15	16	16
Disagree	59	57	56	59	57
Strongly disagree	19	20	20	19	19
	100%	101%	100%	101%	100%
χ2	1.36		2.80
n	1,652		1,593		1,651
“Marriage usually changes a relationship for the worse.”
Strongly agree	4	6	6	6	6
Agree	11	20	13	22	18
Disagree	60	58	61	57	59
Strongly disagree	25	14	20	15	16
	100%	98%	100%	100%	99%
χ2	48.07***		26.56 ***
n	1,647		1589		1647
“it’s best to avoid marriage because it usually doesn’t work out.”
Strongly agree	1	8	6	7	7
Agree	10	15	10	17	16
Disagree	61	60	62	59	58
Strongly disagree	27	16	22	17	19
	99%	99%	100%	100%	100%
χ2	46.92***		18.13 ***
n	1,654		1,595		1654

Note: Totals may not sum to 100% due to rounding.

^*p <.10.

^**p <.05.

^***p <.01.

The bottom half of the table presents responses to 2 questions about whether marriages are usually unsuccessful. Neither mentions divorce explicitly, but both are relevant to the issue of whether women think marriages should be avoided because they typically fail. The responses show that just one-fourth of the women in the sample agree or strongly agree with the statements “Marriage usually changes a relationship for the worse” and “It’s best to avoid marriage because it usually doesn’t work out.” For these 2 questions, there are no significant differences in the response patterns of Hispanics, African Americans, and non-Hispanic Whites (not shown). Women who are currently married or who have ever married are less likely to agree with these pessimistic statements about the success of marriage; nevertheless, even among the never-married less than 30 percent agree or strongly agree with these statements.

Taken as a whole, the responses to these 4 questions do not support the contention that most low-income women have a fear of divorce that could lead them to avoid marriage. Large majorities responded that there is little stigma to divorce, that marriage does not usually change a relationship for the worse, and that it is not best to avoid marriage because it might not work out. Less than 10 percent answered more than 2 of the 4 questions in ways consistent with the idea that divorce was shameful enough that marriage should be avoided. Thus, our survey findings provide almost no support for the claim in the qualitative literature that women’s fear of divorce is a major factor in causing them to postpone marriage

Because this conclusion differs from the literature, we examined the field notes for all 256 families in the full ethnographic sample. If fear of divorce were prevalent, we would expect it to appear at least occasionally in the field notes for the multi-year fieldwork period for this very large ethnographic sample. That is to say, we would expect it to be an emergent theme that would not need direct prompting by the ethnographer. We found very little mention of divorce at all, perhaps because marriage itself is not an issue that weighed heavily on these women’s minds (7 percent were married with a husband present in the household, and 15 percent were cohabiting, when first visited by an ethnographer); and almost no mention of anything that could be construed as fear of divorce.

The responses of the 41 families in the ethnographic follow-up sample (who were directly asked the four questions in the 2005 survey) were similar to those of the survey sample: just 22 percent agreed or strongly agreed that “it is best to avoid marriage because it usually doesn’t work out,” and 21 percent agreed or strongly agreed that “marriage usually changes a relationship for the worse.” As for stigma, 63 percent agreed or strongly agreed that getting a divorce “is not a big deal” (as the question was worded during the pilot test); and 80 percent disagreed or strongly disagreed that divorce is embarrassing. On the question of embarrassment, a White woman from Boston commented:

For me, I strongly disagree. I can see somebody maybe who has some certain high place in society, and the whole world is watching and waiting and – but not, not for regular people. What we did find in the field notes is evidence of a generalized anxiety about the reliability

of men in relationships. Many women expressed concern about getting hurt in a relationship, regardless of its legal status. As one thirty-year-old native-born Mexican woman from San Antonio said, “I try not to get my hopes up out of any relationship because it always seems to fall…. I’m scared. I’m scared to be in a relationship.” In addition, the older women in the ethnography seemed less idealistic about marriage than younger women. As one thirty-year-old White woman from Chicago said, “I learned what I know from experience.” Based on their experiences, women are reluctant to enter a marriage they view to be inadequate; a thirty-year-old Mexican woman in San Antonio says, “after going through what I’ve gone through, I’m not just settling for pennies. I want the new shiny dollar now.” These women do not feel they should have to settle for an inadequate man, yet they worry that there may not be a man who is adequate. However, whereas worries regarding a failed relationship are mentioned, divorce is not. Perhaps the fear of divorce that other researchers claim to have found reflects, to some extent, this generalized anxiety about relationships rather than a specific fear of being shamed by divorce itself.

CONCLUSION

Do low-income women hold attitudes that support an alternative life course, one in which childbearing occurs far earlier than marriage, if marriage occurs at all? Our results provide less support for this contention than do recent studies by Edin and Kefalas and others. Support from our study is strongest for the proposition that having a child outside of marriage carries very little stigma in low-income neighborhoods. The vast majority of the mothers in the study agreed that a woman should have children if she wants to, even if she is not married. Most disagreed with the statements that having a child without being married is embarrassing and that it hurts a woman’s chances of marrying. These sentiments were widespread; they held for African Americans, Puerto Ricans, Dominicans, other non-Mexican Hispanics, and non-Hispanic Whites as well as for women with and without high school degrees. Although native-born women of Mexican origin were more likely to respond that having a child outside of marriage was stigmatizing, even among these mothers a large majority responded in ways suggesting that the stigma is modest. It seems likely, then, that the social climate in which our mothers live treats births outside of marriage as events that may not be optimal but which carry little stigma.

We found limited support, however, for the proposition that women prefer to start having children well before marrying. The gap between the best ages to have a first birth and to marry seems to be smaller than the Edin and Kefalas study, in particular, suggests. Very few of the women in the survey sample said that the best age to start having children was under 20, even though about half of them had given birth as teenagers. Most mothers chose an age range in the twenties as best for both starting to have children and marrying. We did not, therefore, find evidence of a set of beliefs that favors a large gap such as having a first child in one’s late-teenage years and getting married in one’s thirties. The gap between ages at first childbirth and marriage was narrower than that. At least ideally, most women in our sample viewed the start of childbearing and marriage as both occurring during one’s twenties.

On the other hand, embedded within this narrower gap were some modest but noticeable differences: among women who did not choose the same 5-year interval for both events, more chose an earlier age for first childbirth than for marriage. There was more sentiment, in other words, in favor of births coming well before marriage than for marriage coming well before births. Moreover, the earlier-birth, later-marriage ordering was favored by African Americans mothers more than by mothers in most Hispanic groups. This distinction fits with Edin and Kefalas’s observation that the African American women in their study were more likely to follow a path in which having children occurred well before marriage. In our study, we were able to examine responses among several Hispanic groups: mothers of U.S.- and foreign-born Mexican origin, Puerto Rican mothers, Dominican mothers, and mothers in a residual, heterogeneous “other Hispanic” group. We found that Puerto Ricans and the residual group were the only Hispanic groups whose preferences for ages at childbirth and marriage did not differ significantly from African Americans.

Finally, we found almost no evidence to support the proposition that fear of divorce is an important part of the story of the postponement of marriage among the mothers in the neighborhoods we studied. For instance, only one-fourth of the mothers agreed that getting a divorce is embarrassing or that one should avoid marriage because it usually does not work out. It is probably true, as many other studies have found, that low-income women and men (like the more affluent) will not enter into a marriage until they are convinced that the material and emotional foundations are present – steady incomes, love and companionship, interpersonal trust, and so forth (Gibson-Davis, Edin, & McLanahan, 2005; Smock, Manning, & Porter, 2005). And most people only begin a marriage if they expect it to last. These ideas are supported by the women in the ethnographic sample, who expressed a generalized anxiety about the success of relationships with men. Taken together, however, the responses of the survey and ethnographic samples suggest that few women avoid marriage because they are concerned that it may end in an embarrassing divorce. If our findings are confirmed in other studies, one might conclude that fear of divorce does not stand in the way of getting married for most low-income women.

We were also able to differentiate attitudes among Hispanic sub-groups and compare them to African Americans. As the literature would lead us to expect, we found the strongest differences between mothers of Mexican origin and African American mothers. Moreover, this difference was as strong, if not stronger, when just U.S.-born Mexican mothers were compared with African Americans as when just Mexican-born mothers were compared with African Americans. This result is inconsistent with an acculturation model under which second- and subsequent-generation Mexican Americans are said to develop attitudes that are more in line with non-Hispanic Americans. Rather, our results suggest that attitudinal differences persist among the U.S.-born Mexican-origin population. Dominicans and the residual group of “other Hispanics” also expressed more conventional attitudes toward the timing of marriage and childbearing than did African Americans, who showed, as expected, the least conventional attitudes. It is possible that the more conventional attitudes among some Hispanics reflect a cultural practice of providing socially-desirable answers to sensitive questions (Sanchez Acona 1964). But it also is of interest that we did not find significant differences between Puerto Ricans and African Americans, consistent with previous evidence that Puerto Ricans’ attitudes toward marriage are less conventional than the attitudes of Mexican Americans (Oropesa & Gorman, 2000). The similarity between the responses of the Puerto Ricans and the African Americans suggests that the attitudes shown by African Americans may reflect not only their specific history and culture but also more general factors that they share with other minority groups in which single-parent families are common.

In terms of life course theory, our study implies that the pathways taken by many low-income women, particularly starting to have children at a young age and delaying marriage, do not necessarily reflect an alternative value system. To be sure, attitudes and preferences may be playing a role, but they do not appear to support a long lag of a decade or more between having a first child and marrying. Nor does fear of divorce appear to stand in the way of marrying, at least in our study. Perhaps apprehension about the difficulty of successfully managing an intimate relationship is the emotion that other observers have taken as an aversion to divorce. In any case, the long delays between having a first child and marrying which have been reported in the literature may reflect the constraints that low-income individuals face more than their preferences. These would include internally-imposed constraints such as waiting until one’s financial situation is stable and externally-imposed constraints such as a shortage of suitable marriage partners. Finding a suitable partner and judging oneself ready to marry, given these constraints, could take longer than a woman might prefer. If these structural barriers were lessened, women’s attitudes might not stand in the way of a more conventional life course in which marriage played a greater role, even if premarital childbearing remained common. Caring for children may be the promise low-income women think they can keep whether they are married or not; but, ideally at least, most approve of making the promise of marriage relatively close to the birth of their first child.