Cohabitation and Family Change

Judith A Seltzer. Handbook of Contemporary Families. Editor: Marilyn Coleman & Lawrence H Ganong. Sage Publication. 2004.

Increasingly delayed marriage (Goldstein & Kenney, 2001), greater approval of sex before marriage (Thornton & Young-DeMarco, 2001), and high rates of marital instability (Goldstein, 1999) pave the way for unmarried couples in the United States, including those who do not plan to marry, to live together. For the past three decades, cohabitation rates have risen dramatically among all demographic groups, including parents. Growth in the numbers of adults who have ever cohabited has shifted researchers’ conception of cohabitation as an appropriate topic in reviews on nontraditional family forms (Macklin, 1980; Sussman, 1972) to one that highlights the widespread acceptance of cohabitation whether or not the couple plans to marry (Thornton & Young-DeMarco, 2001).

Despite greater acceptance of cohabitation, the public and researchers still disagree about whether cohabiting unions harm individuals and threaten the larger social order. Couples who live together before marriage have higher divorce rates than those who do not live together before marriage (Bumpass & Lu, 2000; Lillard, Brien, & Waite, 1995). Children born to cohabiting couples have unstable family lives compared to children born to married parents (Manning, Smock, & Majumdar, 2002). The relative importance researchers assign to cohabitation itself and to the characteristics of those who choose to cohabit or bear children in cohabiting unions (e.g., those with more liberal attitudes or who are less religious) is at the core of the debate about whether cohabitation is harmful. Both the experience of cohabitation and selection into cohabiting unions contribute to marital instability among cohabitors who divorce (Axinn & Thornton, 1992; see Seltzer, 2000b, for a review of this evidence).

A major theme since the 1970s in research on cohabitation is whether cohabitation is a stage in courtship that leads to marriage or an alternative to marriage. This theme still motivates most contemporary research on cohabitation in the United States. Because cohabitation and marriage are similar, at least superficially, in their status as heterosexual, co-resident unions, many studies compare the two relationships. The comparisons demonstrate the tensions between treating unions as private relationships governed only by the feelings of the two couple members and treating the unions as public phenomena governed by informal social rules and administrative and legal rules about the rights and responsibilities of couple members. Marriage is a well-developed, albeit changing, social institution in the United States, whereas cohabitation, at best, is an incomplete institution (Blumstein & Schwartz, 1983; Nock, 1995). Because cohabiting unions have become more common over the past 30 years, they are a more routine component of social life in the United States in much the same way as remarriage after divorce became more routine in the 1970s (Cherlin, 1978).

Emphasis on the changing institutionalization of cohabitation points to a weakness in debates that frame cohabitation in terms of dichotomies: that is, as either a courtship stage or an alternative to marriage. This type of debate ignores the diversity of cohabiting experiences for individuals and for the U.S. population. The meaning of a cohabiting union may change over time for the cohabiting individuals. What begins as a convenient living arrangement or strategy in the dating game may become a more serious relationship that leads to marriage or plans for marriage (Sassler & Jobe, 2002). Alternatively, those who live together as a step on the way to marriage may change their minds about whether the other person would make a good spouse. Three quarters of cohabiting women expect to marry their partners (Manning & Smock, 2002), but only about a third of all cohabitors marry within 3 years (Bumpass, 1995). Of course, the decision to marry involves two people. About one fifth of cohabiting partners disagree with each other about whether they plan to marry (Bumpass, Sweet, & Cherlin, 1991). Men’s preferences may matter more than women’s (Brown, 2000), a finding to which I return later.

The meaning of cohabitation also may change over historical time at the population level as it is adopted by a wider variety of people who differ in their reasons for living together (see Manting, 1996, on the changing meaning of cohabitation in the Netherlands). Shifts in the meaning of cohabitation at the population level occur because of changes in the composition of cohabiting couples: that is, the relative mix of couples who cohabit as a step on the way to marriage and those who do not intend to marry when they begin living together. Even when one group dominates the population of cohabitors, substantial variation in types of cohabiting unions remains (Casper & Sayer, 2002). The primary challenge of describing the meaning of cohabitation in the United States over the past 30 years is characterizing a moving target. The meaning of cohabitation continues to shift over time as it becomes more acceptable in a wider range of circumstances. The changing social context further alters the meaning of cohabitation.

The quality and quantity of social science data on who cohabits and the meaning of cohabitation have improved dramatically since 1970. As cohabitation has become more common, researchers have included more questions about cohabitation on demographic surveys and in interviews about family life. The repeated finding in population surveys that cohabitation is more common among those with fewer economic and educational resources has fostered new research on couple relationships that examines the reasons for this difference instead of focusing on relationships between college students who cohabit (Macklin, 1980). Attempts to understand the meaning of cohabitation in the kinship system use both quantitative and qualitative evidence. At its best, knowledge gained from one style of research informs the collection and analysis of other types of data on cohabitation. Improvements in the quality of data about cohabitation in the United States parallel improvements in European countries (Kiernan, 2002), thereby permitting comparisons of cohabitation and its effects across social contexts that differ in their economic and social supports for cohabiting couples.

This chapter describes changes in U.S. cohabitation since 1970. It emphasizes what social scientists know, how they know it, and what they still do not know but should try to learn in the coming decades. I treat cohabitation as a dyadic union between unmarried heterosexual adults who co-reside. New information about gay and lesbian couples who live together (Black, Gates, Sanders, & Taylor, 2000) and changes in policies that define the rights of homosexual partners make this an important area for new research but one that is beyond the scope of this chapter. My chapter builds on other recent reviews and collections of papers on cohabitation (Booth & Crouter, 2002; Casper & Bianchi, 2002; Seltzer, 2000b; Smock, 2000; Smock & Manning, 2001; Waite, Bachrach, Hindin, Thomson, & Thornton, 2000).

Facts about Cohabitation

Demography of Cohabitation

In 1970, approximately 523,000 households were maintained by unmarried couples (U.S. Bureau of the Census, 1998). By 2000, the number had increased to 4.9 million (Simmons & O’Neill, 2001). I use 1970 as the comparison point because this handbook examines change over the past 30-some years. However, census data also show that the start of the dramatic rise in the rate of increase in cohabiting couples in the United States was around 1970 (Glick & Spanier, 1980). This cross-sectional evidence understates the number of adults who have ever cohabited because cohabiting unions typically do not last long. For instance, by 1995 nearly 40% of women 20 to 24 years old had ever cohabited, but only 11% were currently in cohabiting unions (Bramlett & Mosher, 2002, Tables B and C). Estimates of whether a person has ever cohabited also show that cohabitation has become increasingly common (Bumpass & Lu, 2000). By the mid-1990s, over half of all first unions (marriages and cohabitations combined) were cohabiting unions (Bumpass & Lu, 2000).

Young persons are more likely to be cohabiting than older persons (Bumpass & Sweet, 1989, Table 1), a pattern consistent with interpretations of cohabitation as a stage in the courtship process. However, older persons have also become more likely to cohabit in recent decades (Waite, 1995). The rapid rise in rates of cohabitation for newer cohorts means that age differences in whether individuals have ever cohabited are likely to diminish over time as older adults who came of age when cohabitation was less acceptable than it is today are replaced by those who came of age when cohabitation was more acceptable (Bumpass & Sweet, 1989, 1995; Oropesa, 1996). Cohabiting unions are more common among those who have been married previously than among the never married (Bumpass & Lu, 2000, Table 1). The prevalence of cohabiting unions among the previously married is especially important because much early research on why unmarried couples live together focuses on those who have never been married.

Patterns of marriage and divorce differ substantially by race and ethnicity (Cherlin, 1992), so it is not surprising that there are race-ethnic differences in cohabitation as well. In 1995, just over half of non-Hispanic black women had married by age 30 compared to more than three quarters of Hispanic women and non-Hispanic whites and Asians (Bramlett & Mosher, 2002, Figure 2, Table 1). In addition, rates of marital separation and divorce are much higher for black than white women, a continuation of a long-term trend in the United States (Bramlett & Mosher, 2002, Table D; Sweeney & Phillips, 2002). Compared to whites, black women are more likely to cohabit than to marry as their first union (Raley, 2000). In fact, when both cohabiting unions and marriages are taken into account, the race gap in union formation is reduced by about half (Raley, 1996). Black women are less likely to formalize their first cohabiting unions by marriage than are white women. Rates of marriage for Hispanic women in their first cohabiting union are between those of black and white women (Bramlett & Mosher, 2002). Cohabitation rates have increased in recent decades for blacks, Hispanics, and non-Hispanic whites, but the pace of increase has been greater among whites than for the other groups (Bumpass & Lu, 2000, Table 2).

Persons with less education and more uncertain economic prospects are more likely to cohabit than to marry, perhaps because the institution of marriage is defined by long-term economic responsibilities (Bumpass & Lu, 2000; Clarkberg, 1999; Willis & Michael, 1994; Xie, Raymo, Goyette, & Thornton, 2003). Economic resources also affect whether cohabitors plan to marry their partners (Bumpass et al., 1991) and whether they are able to fulfill these plans (Smock & Manning, 1997). Although much of the academic literature in the 1980s and 1990s argued that women’s economic opportunities were a primary source of historical change in couple relationships (see Oppenheimer, 1997, for a review), men’s economic resources and career prospects are more important determinants of whether a cohabiting couple will marry than women’s (Smock & Manning, 1997). The importance of men’s economic prospects for marriage is consistent with historical and more contemporary evidence about the economic requirements for marriage in Western kinship systems (Hajnal, 1965; Oppenheimer, Kalmijn, & Lim, 1997). Although the educational differential in cohabitation has persisted since the 1970s, cohabitation rates have increased for all education and income groups (Bumpass & Lu, 2000; Glick & Norton, 1979). Economic characteristics explain some but not all of the race-ethnic differences in cohabitation and marriage (Raley, 1996). Cultural differences also play an important role in accounting for race-ethnic variation in union formation. For instance, Puerto Rican women’s attitudes and behavior demonstrate greater acceptance of cohabitation and of childbearing in cohabiting unions compared to those of non-Hispanic whites (Manning, 2001; Oropesa, 1996).

Cohabiting partners are very similar to each other on race-ethnicity and education but not as similar as spouses are to each other (Blackwell & Lichter, 2000; Casper & Bianchi, 2002). Compared to spouses, cohabiting partners are also more likely to differ on religion (Schoen & Weinick, 1993). Women in cohabiting unions are more likely to be a couple of years older than their partners than wives are to be older than their husbands (Casper & Bianchi, 2002). The greater heterogamy or differences between partners in cohabiting unions may occur because those who cohabit have more liberal attitudes about a variety of aspects of family life, including who is an appropriate partner. Alternatively, cohabiting unions require less commitment than marriages, so partners may be willing to experiment more with relationships they might be unwilling to formalize.

Relationship Stability, Marriage, and Childbearing: Facts about Change in what Cohabitation Means

Three demographic trends suggest that the meaning of cohabitation is shifting. First, although more than half of marriages today are preceded by cohabitation (Bumpass & Lu, 2000), cohabiting unions are less likely to be a prelude to marriage now than in the recent past (Bumpass, 1995, 1998). Second, cohabiting couples are more likely to be parents either because they have had a child together or because they live with children from previous relationships (Casper & Bianchi, 2002; Raley, 2001). Finally, single women who become pregnant are almost as likely to cohabit as to marry the child’s father by the time the child is born (Raley, 2001).

Cohabitation and Marriage

Cohabiting unions do not last long before the couple marries or ends the relationship. About half of cohabiting unions end in a year or less, and only 1 in 10 lasts at least 5 years. Cohabiting relationships became less stable between the mid-1970s and the early 1990s, mainly because of a decline in the percentage of couples who married their cohabiting partner. Among first cohabiting unions formed between 1975 and 1984, 60% of the couples eventually married compared to 53% of those in first cohabiting unions formed in 1990-94 (Bumpass & Lu, 2000;

Bumpass & Sweet, 1989). Recent cohorts are also more likely to live with at least one partner other than their eventual spouse compared to those of the early 1980s (Bumpass & Lu, 2000, Table 3). Cohabiting is one way to learn more about a potential spouse’s characteristics. As it becomes more difficult to identify an appropriate potential spouse, individuals may have several cohabiting unions with different partners as a way to sort partners. Oppenheimer’s (1988) theory of marital search suggests a lengthening of the search process because it takes longer for both men and women to complete their schooling and establish themselves in careers than it did 30 or 40 years ago. Even if the increase in the number of partners is part of the search process, the wider experience of cohabiting unions that do not end in marriage reduces the stigma associated with cohabitation as a union that might not lead to marriage. Reduction in stigma, in turn, fosters a rise in cohabiting unions in which partners do not anticipate marriage. Decline in the percentage of cohabiting couples who eventually marry is consistent with the interpretation of change in the meaning of cohabitation as a stage on the way to first marriage.

Cohabitation and Childbearing

Cohabitation is also becoming a more common setting for rearing and even bearing children. In 1978, almost 28% of cohabiting couples lived with one or both partners’ children, but by 1998 this percentage had increased to 37% (Casper & Bianchi, 2002, Table 2.1). Another way to consider the extent to which cohabiting unions are a setting for child rearing is by examining whether children have ever lived with their mother and her cohabiting partner. Two out of every five children will live in a cohabiting family at some point during childhood on the basis of estimates for the 1990s (Bumpass & Lu, 2000). In addition to raising children from previous relationships, a small but growing percentage of cohabiting couples are having biological children together. Between the early 1980s and 1990s, the percentage of cohabiting couples who had a child together increased from 12% to 15% (Bumpass & Lu, 2000). Even without this increase in fertility rates among cohabitors, the increasing number of cohabiting couples has contributed to the rise in the number of children born in cohabiting unions (Raley, 2001).

Childbearing outside marriage increased dramatically in the second part of the 20th century. Just in the past 30 years, the birthrate for unmarried women increased by 70%, from 26.4 to 45.0 (Martin et al., 2002). Most unmarried women who bear a child are single (Raley, 2001); however, the nature of non-marital childbearing has changed over time so that increasing percentages of children born outside marriage are born to cohabiting parents. In fact, most of the recent increase in childbearing outside marriage is due to the increase in births to cohabitors (Bumpass & Lu, 2000). That the rates have increased as well as the number of cohabitors having children suggests a change in cohabiting behavior. Having a child reflects substantial commitment to the relationship and to joint responsibilities, even in an era when many adults become single parents. The increase in child rearing in cohabitation points to cohabitation as something other than a precursor to first marriage for a growing percentage of the population. That is, cohabitation may be an alternative to marriage for some subgroups, or at least a suitable context for raising children whether or not marriage will follow. Edin’s (2002) finding that low-income parents think marriage is unnecessary for couples raising children together is consistent with this interpretation. Future research should also examine attitudes about marriage and parenthood for those without children and in different socioeconomic groups.

Another sign that cohabiting relationships are a more acceptable setting for child rearing is that by the mid-1990s single women who became pregnant were more likely to move in with the child’s father than had been true a decade earlier (Raley, 2001, Table 3). In earlier cohorts, pregnant single women were likely to marry to legitimate their child’s birth, but by the early 1990s about the same percentage began to cohabit as marry by the time their child was born. Because childhood experiences affect children’s attitudes and behavior in adulthood, children raised by cohabiting parents will be even more likely to cohabit than children who never live with cohabiting parents. This implies further increases in U.S. cohabitation rates.

Change in the meaning of cohabitation is interdependent with other changes in U.S. families. Men’s and women’s obligations in marriage are shifting to an arrangement in which both spouses help fulfill their family’s economic needs. More young adults think that both mothers and fathers should do housework and look after children. Divorce is more acceptable, and childbearing outside marriage no longer carries the severe stigma for mother or child that it once did (Thornton & Young-DeMarco, 2001). Both economic and cultural factors contribute to these broad changes (Bumpass, 1990; Lesthaeghe, 1995; Smock & Gupta, 2002). Whatever the initial sources of change, since cohabitation became widespread, the change in behavior has had a momentum of its own.

How Cohabiting Couples Live

Maintaining a household requires that someone accomplish basic tasks. Money must come into the household, usually through earnings from employment. Someone must spend the money to pay rent and provide food, heat, and other basic necessities. Someone must cook and clean. If children live there, someone must supervise and care for them. Who does these different tasks depends on social rules about the appropriate work for men and women, adults and children. The division of labor also depends on personal negotiations between household members. The demographic facts about cohabitation are better known than the facts about how cohabitors manage their lives. This section provides a broad interpretation of existing evidence, but knowledge of the facts about cohabitors’ lives would benefit from more research.

Young women and men with liberal attitudes about gender roles are more likely to cohabit than to marry (Clarkberg, Stolzenberg, & Waite, 1995). Their behavior is consistent with these attitudes. Both partners in cohabiting couples are more likely to work for pay than partners in married couples (Casper & Bianchi, 2002, Table 2.2). Cohabiting couples also divide housework somewhat more equally than do married couples, with cohabiting women doing less housework than married women (Brines & Joyner, 1999; Nock, 1995; Shelton & John, 1993; South & Spitze, 1994). Women, however, do much more housework than men do in both types of couples. When men begin cohabiting, they reduce the time they spend on housework compared to when they were single. The same is true for men who marry (Gupta, 1999). Differences in women’s and men’s preferences, their earning potential, and expectations about gender differences in household responsibilities contribute to the greater amount of housework women do.

Unfortunately, there are no data on trends in men’s and women’s relative contributions to housework in cohabiting relationships. Compared to married couples, cohabiting couples in recent cohorts are likely to divide work more evenly than in the past, at least among those without children. This is because fewer cohabiting unions today lead to marriage, where behavior is guided by the implicit long-term contract about the exchange of women’s unpaid housework for their male partner’s earnings. In addition, increases in cohabitation among more highly educated women may be a bargaining strategy that women use to test their partners’ willingness to share housework before marriage. Wage rates for highly educated women have increased relative to men’s in recent cohorts, providing women with somewhat more bargaining power than in earlier cohorts (Cherlin, 2000).

Married couples are much more likely to pool their incomes than cohabiting couples (Bauman, 1999; Blumstein & Schwartz, 1983; Kenney, 2002; Landale, 2002). Income pooling may signal commitment to the relationship, but it may also enhance union stability by reducing conflict about how money is spent and other transaction costs (Seltzer, 2002; Treas, 1993). Cohabitors with children are more likely to pool their incomes, perhaps as a way to recognize the shared responsibilities for children’s needs (Winkler, 1997). Compared to married parents, cohabiting parents spend less money and a smaller share of their incomes on health care and educational expenses (DeLeire & Kalil, 2002). However, compared to those who live with a single parent, children whose parents cohabit may still benefit from the cohabiting partner’s income through public goods, such as improved housing and other shared resources (Citro & Michael, 1995; Moffitt, Revelle, & Winkler, 1998). Taking account of cohabiting parents’ incomes would have reduced by nearly 30% the proportion of children in cohabiting-cou-ple families who were in poverty in 1990 (Manning & Lichter, 1996). Children whose parent or parents are cohabiting suffer economic disadvantages compared to those whose parents marry, largely because those who cohabit have less schooling and lower earnings than those who marry (see the above discussion in “Demography of Cohabitation”; also Manning & Lichter, 1996; Morrison & Ritualo, 2000).

How much cohabiting partners contribute to children’s material needs may depend on whether the children are the biological children of both partners or from one partner’s previous relationship. For instance, Case, Lin, and McLanahan (1999) showed that households in which men are helping to raise their partners’ children from a previous union spend less on food than households in which married parents and children live. Income pooling and money management among cohabitors is an important topic for future research (Bumpass & Sweet, 2001; Seltzer, 2002). New studies should compare cohabiting parents to couples without children because the former already demonstrate a higher level of commitment to the union by having a child together.

Cohabitation and Children’s Welfare

Until recently, children who lived with cohabiting parents were treated by researchers as though they were living with a single parent. This assumption understates children’s access to economic resources. It probably also understates the time and attention children receive from adults, especially when the cohabiting partners are the children’s biological parents. A recent study of new parents found that mothers and their infants born outside marriage get more support from the child’s biological father when the parents are living together than when the parents are romantically involved with each other but live in separate households (Carlson & McLanahan, 2002). The same study showed that compared to married biological fathers, cohabiting fathers interacted with their young children in similar ways (Carlson & McLanahan, 2001). Time-use data also show that biological fathers spend about the same amount of time engaged with their children whether they are cohabiting or married to the children’s mother. However, once other differences between cohabiting and married families are taken into account, cohabiting biological fathers spend less time in activities with their children than married biological fathers do (Hofferth & Anderson, 2003). Social scientists are just beginning to study how cohabiting fathers participate in child rearing when their relationship with the children’s mother ends (Cooksey & Craig, 1998; Landale & Oropesa, 2001; Manning, 2002; Seltzer, 2000a).

Marriage has little effect on the types of child-rearing activities men pursue with their partner’s children (i.e., children who are not the men’s biological children). Married stepfathers, however, spend more time in children’s organized activities, such as school, religious, and community activities, than do male cohabiting partners (Brown, 2002; Thomson, McLanahan, & Curtin, 1992). Marriage to the child’s mother may legitimate the stepfather’s participation in the child’s public life because schools and other public settings have administrative structures that recognize the social authority over children that comes with marriage to the child’s mother. Rules about adults’ responsibilities for their cohabiting partner’s children are less clearly defined. This may create institutional barriers to cohabiting fathers’ participation as well as increase their reluctance to participate.

Cohabitation, like remarriage, also may affect how mothers interact with their children. A second adult can reinforce the mother’s authority, provide her with emotional support, and help supervise the children. Single mothers who remarry or begin a cohabiting relationship do not discipline their children as harshly as mothers who do not acquire a new partner, but there are no differences between mothers who cohabit and those who remarry (Thomson, Mosley, Hanson, & McLanahan, 2001). Evidence on whether cohabitation improves supervision is mixed (Thomson et al., 2001).

Evaluating the effects of cohabitation on children’s well-being requires distinguishing between children who live with their unmarried biological parents and those who live with biological parents and cohabiting partners (informal stepfamilies). Biological parents usually invest more in their children than do stepparents (for a review, see Anderson, Kaplan, & Lancaster, 2001). Many children who live with a parent and a cohabiting partner have experienced the disruption of their biological parents’ union, an experience that also affects children’s welfare. Because the rise in the numbers of cohabitors who bear children is relatively recent, few studies have a large enough sample to support a comparison between the well-being of children who live with cohabiting biological parents and that of children whose biological parents are married. These comparisons are more feasible in studies of younger children than in studies of teenagers because so few cohabitors stay together for more than a few years.

The lack of precision due to small sample sizes can be addressed, in part, by looking for consistent findings across studies. It is sometimes difficult, however, to reconcile inconsistent findings across studies because the studies differ in whether they control for other factors, such as income, that affect both cohabitation and child well-being. For instance, data from the 1997 National Survey of America’s Families (NSAF) show that among teenagers living with both biological parents, teens whose parents are married are much less likely to have behavioral or emotional problems than teens whose parents are cohabiting (Brown, 2002). In contrast, data from the National Survey of Families and Households (NSFH) conducted 10 years earlier suggest that children who live with both biological parents look very similar on a range of child well-being measures, regardless of whether their parents are married to each other (Hanson, McLanahan, & Thomson, 1997). The greater difference between teenagers in families with married and unmarried biological parents in Brown’s study might be due to the absence of statistical controls for family characteristics that explain both cohabitation and teens’ well-being.

Whether cohabitation is good or bad for children also depends on what alternative the children have to living in a cohabiting-couple family. For many children, the alternative is living with a single mother. For others, it is living with a biological parent and the parent’s new spouse (formal stepfamily). Compared to children living with a single mother, those who live with their mother and her cohabiting partner (not the child’s father) have more behavioral problems, worse school performance, and higher rates of delinquency (Manning & Lamb, 2002; Nelson, Clark, & Acs, 2001; Thomson, Hanson, & McLanahan, 1994). Whether these differences between children in single-mother and cohabiting families persist once differences by family type in economic resources and child-rearing practices are taken into account varies across studies. Nelson et al. (2001) and Thomson et al. (1994) found disadvantages for children in cohabiting families when they controlled for other family characteristics in the NSAF and NSFH, respectively. Manning and Lamb (2002), using the National Longitudinal Study of Adolescent Health (Bearman, Jones, & Udry, 1997), found that teenagers in single-mother and cohabiting families were very similar on a wide range of outcomes once other family characteristics were taken into account. Finally, there is some evidence that teenagers who live with a mother and her cohabiting partner do not fare as well as teens living with a mother and stepfather on behavioral problems, including delinquency and academic achievement (Manning & Lamb, 2002; Nelson et al., 2001). Thomson et al. (1994) reported a similar finding for an earlier cohort.

Research on the effects of cohabitation on children’s well-being is still quite sparse. Much of it relies on cross-sectional evidence in which causal associations are ambiguous. In particular, many studies do not take account of preexisting differences between those who cohabit and those who do not that might explain an association between cohabitation and children’s welfare. Longitudinal studies, including those that use retrospective reports of childhood living arrangements, are hampered by small numbers of cases. Small samples and incomplete histories of parents’ unions limit the dimensions of cohabitation that can be taken into accountfor instance, whether the cohabiting union is between the children’s biological parents or a parent and nonparent, the duration of the union, and the number of formal and informal unions the child has experienced. Duration and stability of parents’ unions are especially important for assessing the effects of cohabitation on children because children benefit from stable living arrangements (Hao & Xie, 2002).

Restricting cohabitation histories to information obtained from household composition observed annually or to only marriagelike relationships also severely hampers research on the effects of cohabitation on children. For instance, the relationship history in the National Longitudinal Study of Adolescent Health (Add Health) questionnaire asks parents to report about current and previous relationships that were the same as marriage even if the partners were not formally married. This restriction to marriagelike relationships probably reduces reports about short-term relationships in which the partners were not very committed to each other. These relationships may increase disruption in children’s lives at the same time that they provide a model of casual unions. Annual observations also miss short-term unions, a serious problem given that about half of cohabiting unions end in a year or less (Bumpass & Lu, 2000). As increasing numbers of cohabiting couples live together without planning marriage or behaving as if their unions are long-term alliances, collecting data on only marriagelike cohabiting unions will underestimate children’s exposure to cohabitation and its effects on their welfare.

Many of the weaknesses in past research arise because of the low incidence of cohabitation involving children for many of the cohorts represented in commonly used data sources, such as the 1987-88 NSFH panel and the 1979 National Longitudinal Survey of Youth (Zagorsky & White, 1999). The cohorts represented by these studies may also differ in the effects of cohabitation on children because the meaning of cohabitation has changed over time as it has become more common and as childbearing in cohabiting unions has been less stigmatized. Other studies, such as McLanahan, Garfinkel, Brooks-Gunn, and Tienda’s Fragile Families and Child Wellbeing Study and Add Health, will become more useful as additional waves of data are released. Because race-ethnic groups may differ in the acceptability of having children in cohabiting unions and because the effects of cohabitation on some child outcomes vary by race-ethnicity (Dunifon & Kowaleski-Jones, 2002), it is important for new research to examine the effects of cohabitation on children within race-ethnic groups whenever possible (see Manning, 2002, for a more complete discussion of this and related issues).

Cohabitation and Family Ties

Whether a couple decides to cohabit is a personal decision, but the partners make this decision in the context of the broader social environment, including their network of family and friends. Couples who are considering living together take into account whether their parents and friends will think they are making a good decision (Liefbroer & de Jong Gierveld, 1993). Young women whose mothers approve of cohabitation are more likely to cohabit than marry (Axinn & Thornton, 1993, Table 1). If parents disapprove, cohabitation may strain their relationships with adult children. Parents whose adult children are married say they are closer to their children than parents whose children are cohabiting (Aquilino, 1997). Cohabitors also report less positive relationships with their parents than married persons (Nock, 1995). However, cohabiting couples are more integrated into their parents’ social and leisure activities than are unmarried adult children (Aquilino, 1997). An indication of similarity between cohabiting and married couples’ integration in their kin networks is that partners in each type of couple are about equally likely to have been introduced to each other by family members (Laumann, Gagnon, Michael, & Michaels, 1994). Moreover, the cohabiting experiences of young adult children increase mothers’ acceptance of cohabiting unions (Axinn & Thornton, 1993). These findings suggest the importance of examining the causes and effects of cohabitation on couples’ broader social networks instead of focusing solely on the couple or the couple’s children.

Improvements in Data

Counting Cohabitors

Two innovations have revolutionized knowledge about cohabitation: (a) the creation of demographic time series that chart the growth in cohabitation and its spread across socioeconomic and demographic groups and (b) the inclusion of cohabiting couples in studies of family life and household organization. The U.S. Bureau of the Census did not identify unmarried partners with a relationship code on the household roster until the 1990 census. However, since the 1970s, researchers have used the creative strategy of identifying likely cohabiting partners in decennial and Current Population Survey data, a strategy first developed by Paul Glick and his colleagues (e.g., Glick & Norton, 1979). Cohabitors were identified as persons of the opposite sex sharing living quarters (POSSLQ). Other researchers have modified the procedures for identifying cohabitors to improve counts of couples who live in households with more than two adults (Sweet, 1979), including a recent study that includes a comparison between a revised POSSLQ measure and explicit survey reports about who is cohabiting (Casper & Cohen, 2000). The early POSSLQ measures appear to underestimate the numbers and rate of growth in cohabiting couples in the United States compared to more recent adjustments to the POSSLQ measure to include households in which couples live with children older than 15 years. The adjusted measure captures a higher percentage of cohabitors who identify themselves as partners and misses fewer self-identified partners than the unadjusted POSSLQ measure (Casper & Cohen, 2000).

The development of retrospective cohabitation histories in surveys about family-related topics complemented the data on cross-sectional trends available from census sources. The 1987-88 NSFH identified cohabitors on the household roster and asked all respondents, male and female, if they had ever lived with someone, the dates each cohabitation began and ended, and a full marriage history, including whether the cohabiting partners married each other. Asking questions on the start and end dates of cohabiting unions continued the principle in several decades of demographic research that treated the end of marriage as when the couple stopped living together rather than the date spouses were legally divorced (e.g., see Bumpass, 1984). The detailed retrospective reports from the NSFH are the data used in much of my summary of the facts of cohabitation. Several other surveys include cohabitation histories, such as the 1995 National Survey of Family Growth (NSFG) (Kelly, Mosher, Duffer, & Kinsey, 1997). The NSFG is particularly valuable for studying change over time in cohabitation because since 1982 it has included some cohabitation information and because the NSFG is repeated periodically, although not at regular intervals. The 2002 survey data will provide important new information about the cohabitation experiences of recent cohorts by including cohabitation histories for men as well as women. Including men in the study is a valuable addition in light of the importance of men’s economic characteristics and attitudes for cohabitation and marriage decisions.

It is now much more routine for surveys, including those about topics other than family relationships, to ask if respondents are cohabiting. Without the innovative use of census and survey data to describe the changes in cohabitation over the past several decades, social scientists would know very little about who cohabits and the association between cohabitation and important social and economic characteristics. Future researchers will not be able to answer the most basic factual questions about cohabitation or address theoretical debates on the meaning of cohabitation unless they make systematic efforts to collect data at routine, periodic intervals on individuals’ history of cohabiting relationships. Large samples are necessary to describe reliably differences among subgroups. The need to collect data routinely is particularly important at this historical juncture because many standard sources of data on marriage, divorce, and other family relationships have deteriorated during the 1990s. For instance, vital statistics on marriage and divorce were not compiled at the national level after 1995. The age-at-first-marriage question was deleted from the decennial census in 1990, and the U.S. Bureau of the Census’s Current Population Survey omitted the marriage history from the June 2000 survey, ending a time series that has provided invaluable data on trends in children’s experiences in single-parent and stepfamily households. (See the U.S. Forum on Child and Family Statistics, 1996, for a recent report about measures of marriage and cohabitation in federal data sources.)

Gathering the Facts about Cohabitors’ Lives

Perhaps because demographic data showed that commonly accepted myths about cohabitation were false (e.g., the myth that cohabitation was more common among college students than among those with less education), researchers began to broaden their investigations to examine other aspects of cohabitors’ lives beyond the demographic contours. Among the richest accounts of how cohabiting couples manage their lives is Blumstein and Schwartz’s (1983) American Couples study, which compares cohabiting couples to three other types of couples: married heterosexual couples, gay couples, and lesbian couples. The comparison provides variation on the degree of institutionalization and gender, both of which provide social guidelines for how partners divide work. By including information about the quality of couples’ relationships, the study also speaks to how individual couples negotiate their roles within the relationship in the context of institutional constraints. Despite these strengths, the study has some weaknesses, including the volunteer sample and largely cross-sectional design. Studies using national surveys and probability samples complement the American Couples study, but because most of the surveys address multiple topics, they do not provide the same rich account of couples’ lives. The NSFH comes closest to the American Couples study because it has broad coverage of household life, asks the same questions of cohabiting and married couples, and includes reports about family relationships from multiple perspectives, including partners and spouses as well as children (Sweet & Bumpass, 1996; Sweet, Bumpass, & Call, 1988).

The NSFH and several other commonly used surveys for studying cohabitation, such as the National Longitudinal Survey of the High School Class of 1972 (U.S. Dept. of Education, 1999)7 and Thornton, Freedman, and Axinn’s Intergenerational Panel Study of Parents and Children, begun with children born in the early 1960s, provide valuable insight into the meaning of cohabitation for the cohorts represented by the samples. Data from these studies are much less appropriate for questions about what cohabitation means today, now that it is a considerably more common experience and is acceptable under a wider range of circumstances than when respondents in these earlier studies were making decisions about cohabitation. To describe family change requires repeated surveys. These are expensive projects to mount, but it may be possible to combine small extensions to surveys already planned as periodic surveysfor instance, the Survey of Income and Program Participation (SIPP) and the NSFGwith new, more complete surveys at much longer periodic intervals. Efforts to evaluate how well cohabitors are identified in the SIPP lay the groundwork for more researchers to use these data to study cohabitation (Baughman, Dickert-Conlin, & Houser, 2002).

Whether members of a couple think of themselves as living together may differ from whether researchers or even friends define them as a cohabiting couple. For instance, a couple may spend every night together and have most of their possessions in one place but not describe themselves as cohabiting (Sassler & Jobe, 2002). How questions about cohabiting relationships are worded and the context in which questions are posed affect estimates of levels and trends (Casper & Cohen, 2000). Change over historical time or differences across social settings in the stigma associated with cohabitation may bias estimates of cohabitation rates derived from explicit reports about cohabitation. Social scientists know little about how respondents understand questions about cohabitation. Comparisons among estimates from different surveys provide some insight into whether estimates of incidence and duration are robust to changes in question wording and sequence, but new studies of cohabitation would benefit from careful instrument development.

A challenge to developing good measures of cohabitation is that each partner may view the relationship differently. Just as marriage has a “his and hers” dimension, men and women who cohabit may differ in whether and when they define themselves as living together (e.g., when they moved their clothes to the same place; when they stopped paying rents in two places; when they spent most, but not all, nights together). Both partners’ reports are important because the decision to cohabit, how the couple manages daily life together, and the stability of the relationship depend on both partners’ preferences. Unfortunately, the quality of men’s reports about family events and the timing of these events is not as good as that of women’s reports (Auriat, 1993; Rendall, Clarke, Peters, Ranjit, & Verropoulou, 1999). Because cohabiting unions are typically short, much of what we know about their timing comes from retrospective reports. It is crucial for new studies to investigate how to improve the quality of reports from both women and men, but especially from men.

Couple data are very valuable for examining the competing interests of cohabiting partners and the quality of their relationship with each other. Although it is difficult to obtain high response rates in surveys of individuals, response rates for couple samples are even lower. Couples in which both partners participate in the study differ from those in which only one participates (Sassler & McNally, 2002). New work should explore how to improve survey participation of both partners and the conditions under which reports from one partner are sufficient.

Cohabiting relationships are important from an academic and social policy perspective because cohabitors share some resources and obligations. Yet it is not necessary to live together full time to share resources and obligations. Couples in serious dating relationships are also likely to share resources. Some couples who are even more committed to each other than many cohabitors may live apart, a relationship sometimes called “living apart together” (Leridon & Villeneuve-Gokalp, 1989; Liefbroer & de Jong Gierveld, 1993). As cohabiting relationships become more varied in the degree of commitment between partners, this increases the conceptual ambiguity of examining only couples who live together instead of broadening the focus of research to include those who are dating and other couples who live apart.

Methods of Studying Cohabitation

In addition to the advances in the quality of cohabitation data available, there have been several methodological advances. Statistical techniques for estimating rates of cohabitation and the duration of relationships from partial or censored histories are more widely used now than 30 years ago. These event history techniques allow researchers to examine the associations between background characteristics and rates of cohabitation, taking account of other individual or family characteristics that may vary over an individual’s life (e.g., educational attainment, full-time employment). Over the past three decades, estimation procedures also have been developed to take into account that individuals decide about their relationships in a context that includes choices and expectations about what the future might bring (Lillard, 1993; Lillard et al., 1995). These methods recognize that whether a person cohabits or has a child outside marriage may depend on factors that also predict whether the person will marry (Brien, Lillard, & Waite, 1999).

Recently investigators have begun to work harder to combine qualitative and quantitative data to learn more about cohabitation. Researchers use several strategies for combining these styles of data. Some use open-ended, in-depth interviews to examine the attitudes, perceptions, and plans of recent cohabitors to explore questions unanswered so far by quantitative studies (Smock & Manning, 2001; see also the Cohabitation and Marriage Study that these authors are conducting, Manning and Smock, 2003). Others collect ethnographic and in-depth interview data as components of multi-method projects, including formal surveys. Examples of such studies include McLanahan et al.’s Fragile Families and Child Wellbeing Study and Cherlin et al.’s Welfare, Children and Families Study, conducted in three cities. A benefit of combining strategies is that investigators can modify survey questions or use open-ended interviews to explore puzzling findings and develop a more complete understanding of the actors’ own explanations for their behavior.

It is tempting to pursue new research that focuses exclusively on actors’ feelings and interpretations of their behavior, but this ignores the importance of the social environment in which adults decide about cohabitation and other couple or family relationships. Whether others approve of cohabitation and the conditions under which it is appropriate also affect adults’ decisions about cohabitation. Individuals can report their perceptions of others’ attitudes, but this cannot substitute for public opinion data from representative, probability samples. Public opinion data from repeated cross-sectional studies have helped describe the changing normative environment that contributed to the rise in cohabitation (Thornton & Young-DeMarco, 2001). These data are still valuable for describing the trajectory of future changes. Researchers should also collect data about attitudes using new questions about cohabitation and obligations between members of a couple and how these relate to other obligationsfor instance, between spouses and between parents and children.

Cohabitation and the Future

It is always dangerous to predict the future, particularly when the prediction is about a phenomenon as rapidly changing as cohabitation is in the United States today. Since 1970, cohabitation has become much more prevalent among diverse demographic and social groups. Recent data suggest that these trends will continue. Cohabitation is here to stay. The meaning of cohabitation and its place in the U.S. kinship system continues to change, making it difficult to characterize. Debates that describe cohabitation dichoto-mouslyas either a courtship stage or an alternative to marriagemiss the heterogeneity of cohabitation and ignore the changing composition of cohabitors over time.

Differences in the meaning of cohabitation for those who are educated and have sufficient, secure incomes and those who are less well educated and more economically vulnerable will be important for future studies of couple and family relationships. This is especially true because recent trends suggest an increasing bifurcation in family experience.

Cohabitation is sometimes called the poor man’s marriage. When men have limited economic resources, they and their potential wives postpone marriage until they can afford the material goods and economic stability they believe that marriage requires. Couples may live together while they wait to attain these economic goals. Some equate marriage with success (Edin, 2002). Many never achieve their goals because of the declining wages and job opportunities for men who lack college educations. Some who are waiting to marry have children while they are cohabiting. Unmarried parents discuss the decision to marry and the economic requirements for marriage separately from their reasons for having children (Edin, 2002). They do not believe that they must marry to raise children. Cohort comparisons show that women who bear a child outside marriage are increasingly likely to bear all of their children outside marriage (Hoffman & Foster, 1997). Other evidence of an increasing division between the advantaged and less advantaged comes from marriage studies that show that college-educated women are more likely to marry than those with less schooling (Goldstein & Kenney, 2001), that husbands’ education reduces the chance of divorce (Teachman, 2002), and that there is growing similarity between husbands’ and wives’ educations (Mare, 1991).

Children depend on their parents’ resources. Increasing disparity between the family circumstances of children whose parents are highly educated and have reasonably secure economic futures and those whose parents are less secure exacerbates the disadvantages children experience when they are born outside marriage. Cohabitation may alleviate some of these disadvantages but expose children to other threats, such as multiple transitions in family structure as the biological parents end their relationship and start new ones. How these children grow up and the relationships they form or consider appropriate to form will determine the future of cohabitation and its place in the kinship system.

Finally, debates about cohabitation emphasize the experience of young adults and implications of cohabitation for children. This ignores that cohort replacement will increase cohabitation among older persons. Older persons who cohabit may be less able or willing to share their household with an unmarried daughter and her child or to transfer money to adult children. Older cohabitors may also need less help from adult children. As the U.S. population ages, it will become even more important to examine the meaning of cohabitation during old age and its implications for the distribution of resources within and between generations.