Bound by Children: Intermittent Cohabitation and Living Together Apart

Abstract

In this article, we examine variations in low-income mothers' patterns of intermittent cohabitation and the voluntary and involuntary nature of these unions. Intermittent cohabitation involves couples living together and separating in repeating cycles. Using Three-City Study ethnographic data, we identified 45 low-income mothers involved in these arrangements, 18 of whom resided with their children's fathers occasionally while saying that they were not in a cohabiting relationship. We term such relationships living together apart (LTA). Data analysis revealed that distinct patterns of voluntary and involuntary separations and reunifications characterized intermittent cohabitation and LTA and that these relationships were shaped by the bonds that shared parenting created and the economic needs of both parents. We argue that these dimensions may explain some disparate accounts of cohabitation status in low-income populations. They also demonstrate previously unexplored diversity in cohabiting relationships and suggest further questioning contemporary definitions of families.

Women living in poverty have frequently exhibited intimate union patterns that fall outside of mainstream norms. In the United States, childbearing without marriage has historically been most common among lower income families (Cherlin, 2005), and cohabitation, though now common among all income groups (Sassler, 2010), was previously practiced primarily among the poor (Cherlin, 2008). Among middle-class families, cohabitation is usually part of a marriage trajectory, with a cohabiting relationship serving as a trial for marriage or as a precursor to the actual wedding (e.g., Lichter, Qian, & Mellott, 2006). Among lower income and minority women, however, cohabitation has sometimes served as a marriage alternative (Phillips & Sweeney, 2005) and has increasingly become an alternative to being single (Manning & Smock, 2005; Sassler & Miller, 2009). During the past few decades, marriage rates have declined substantially among lower income women, urban women, and African American women in particular, with the number of women in these groups who marry before age 30 falling far below national averages (Bramlett & Mosher, 2002; Gibson-Davis, 2011).

Research now indicates that there are not only higher rates of cohabitation among the poor, but also increasingly frequent variations in cohabitation patterns. Rates of nonmarital births continue to be highest among those with lower income and who are less educated (Cherlin, 2005), but women in their 20s, rather than their teens, now account for the majority of all nonmarital births (Ventura, 2009), and a majority of nonmarital births may be to cohabiting mothers (Kennedy & Bumpass, 2008). Multiple partner fertility has become more frequent (Carlson & Furstenburg, 2006). Lower income women are most likely to engage in serial cohabitations, in which they cohabit with more than one man in sequence without ever marrying (Cohen & Manning, 2010; Lichter, Turner, & Sassler, 2010). Unmarried parents are frequently cohabiting at the time of their child's birth, but the relationship often does not last beyond the child's first years (Parents' relationship status, 2007). Scholars, however, have noted that having a child together often creates lasting ties between parents whether or not they remain romantically involved (Edin & Kefalas, 2005; Roy, Buckmiller, & McDowell, 2008).

To a certain extent, these ongoing ties are legally mandated. Changes in the law that began in the later 1960s recognized parental rights and obligations outside of marriage. Before this era, men had no legal rights to their children unless they were married to the child's mother. Subsequently, men not only earned rights to visitation and custody (Mason, Fine, & Carnochan, 2001), but were also charged with the obligation to financially support their children (Aberg, Small, & Watson, 1977). Child support is often vigorously enforced for lower income families in which the mother and children are eligible for public benefits (Koball & Douglas-Hall, 2004; Roy, 1999). Evidence also indicates, however, that fathers have a great social interest in their children (Edin, Lein, & Nelson, 2002; Roy et al., 2008) and that low-income fathers often remain substantially involved in caregiving even when they are not coresident, though these contributions are often not acknowledged or valued by state institutions (Roy, 1999). For fathers in these circumstances, ongoing contact with their children generally involves ongoing contact with the children's mother.

Because of the continued linking of marriage and cohabitation in the literature, researchers may miss patterns among cohabiting couples who have children but have no intentions of creating a sustainable marriage. Cohabiting relationships are known for their instability. Among couples in the Fragile Families Study who were cohabiting at the time of their child's birth, nearly half were no longer romantically involved 5 years later (Parents' relationship status, 2007). Some in cohabiting relationships say the ease of exit from cohabitation is a reason not to marry (Reed, 2006). Among the poor, cohabitations may not have the same voluntary nature that they do among the more affluent and may be initiated with no thoughts of marriage whatsoever (Sassler & Miller, 2009). Circumstances may dictate that cohabitation is necessary rather than desired, such as when one partner has no other place to stay (Goffman, 2009) or when the couple must pool resources to make ends meet (Edin, 2000). When couples wish to cohabit, this too may be constrained by outside forces, for instance if one partner receives public housing assistance that restricts who may stay in the home on a regular basis (Curtis, 2007).

This article seeks to examine unstable cohabitations among low-income parents and the voluntary and involuntary nature of these unions. The broad practice is intermittent cohabitation: couples who live together, then separate, then live together again, sometimes repeating this pattern multiple times. Within the practice of intermittent cohabitation, we seek to explore involuntary cohabitations, those in which the parents live together, but do not see themselves as cohabiting. This practice is not only interesting in itself, but may also be indicative of other variations of cohabitation and family structure that have not been widely considered. Although the number of these ambiguous intermittent cohabiting relationships is modest, they serve as an indicator of how non-marriage-based cohabitations may facilitate the economic, parenting, and social needs of low-income mothers and also speak to the ongoing family involvement of low-income fathers. These distinctive cases challenge us to consider the commonly understood boundaries of interpersonal relationships and family life.

BACKGROUND

Life course theory posits that there is a contextual order for life events, with the establishment of a couple's relationship and marriage occurring before childbearing (Elder, Johnson, & Crosnoe, 2003). The assumption is that the couple's relationship with each other should precede, and indeed supersede, the relationship either partner has with their children. Most assessments of cohabitation have also assumed the primacy of the relationship between the members of the couple, with children often not considered. Although many cohabitations are based on the adults' interest in forming an intimate union, low-income women may face constraints that lead them to form cohabiting relationships with other motivations.

In her theories regarding women's life choices, Gerson (1985) pointed out that choices are all made in the context of women's social conditions and that social class powerfully influences these choices, especially regarding work and motherhood. Wilson (1990, 1996) has emphasized the practical inability of low-income men to earn sufficient wages to support a household, making such men unmarriageable, and men themselves may view a “good job” as necessary for maintaining stable family coresidence even when possibilities for consistent employment are minimal (e.g., Roy, 2005). Looking beyond earning potential alone, Edin and Kefalas (2005) found that the low-income Philadelphia-area women they interviewed had little trust that the men available to them could earn a living and also remain faithful, refrain from illegal activity, and treat them as equals, conditions the women felt were necessary to make a relationship permanent. Many of these women also saw themselves as unprepared for the commitment of marriage, as they believed themselves to be too immature and financially unstable.

The constraints of finances and available partners may lead lower income adults to marry later in the life course or not at all (Edin & Reed, 2005), and as a result, the importance of romance to family relationships in general may diminish. In the Families and Communities Study, a sample of 302 low-income African American families in Chicago with adolescent daughters, mothers reported rarely discussing relationships within the realm of marriage, even though many of them had married or cohabited themselves. Only 9% mentioned either marriage or a long-term, stable relationship as a goal they had for their daughters' futures (Coley, 2002). These mothers instead emphasized their daughters' financial and emotional independence. As marriage rates are lower for women who have lower incomes or live in a central city as well as those who are African American (Bramlett & Mosher, 2002), there may be similar relationship trajectories across racial and ethnic lines among the urban poor.

For poor couples in particular, childbearing may not be planned (Finer & Henshaw, 2006), and cohabitation may be a result of childbearing rather than its precursor. Raley (2001) found that young, single pregnant women were more likely to enter cohabitations than marriages as a response to pregnancy and that cohabitors who became pregnant were unlikely to marry formally. Although some parents may ultimately commit to one another, research has indicated that even when low-income, cohabiting parents express an interest in marrying, few actually do (Parents' relationship status, 2007; Parke, 2004). Both low-income mothers and fathers, however, often emphasize their desire for children and the importance of the parent role in their own lives (Edin et al., 2002; Edin & Kefalas, 2005).

In cases in which cohabitation forms in response to pregnancy or childbearing, relationships may converge around the production and needs of children rather than the adult couple. Parents may decide to cohabit for the benefit of their children or because cohabitation allows both parents direct daily contact with their off-spring. In her analysis of the Time, Love and Cash in Couples with Children (TLC3) study, Reed (2006) found that nearly three quarters of the respondents began their cohabitations in response to a pregnancy and talked about their decisions to cohabit in terms of sharing expenses and parenting. Regarding his baby daughter, one father said it was important to “be a part of her everyday living … at least for the first year or whatever” (Reed, 2006, p. 1123). His phrasing “for the first year or whatever” indicates that once the need for daily intensive parenting decreases, the cohabitation will end. Similarly, one mother said her primary reason for cohabiting was so that her partner “can help me with the baby” (Reed, 2006, p. 1123). Rather than viewing cohabiting relationships as marriage-like, these couples specifically valued the ease with which they could leave the relationship should they want or need to while still being able to share in parenting their child. As a result, though couples may initially intend for their relationship to be fleeting, they instead may maintain long, ambiguous, uncommitted, “suspended relationships” (Roy et al., 2008).

Nock (1995) argued that cohabitation, like remarriage, is an “incomplete institution.” Roy et al. (2008) suggested that among unmarried parents, relationships are better viewed as trajectories rather than statuses. Like remarriage, cohabitation is rife with boundary ambiguity (Stewart, 2005), and the status of the relationship at any particular point may not be determinable. There is no institutional ceremony and no law that enforces or recognizes the inception or demise of a cohabiting relationship, and analyses have found discrepancies in direct versus inferred reports of cohabitation. Manning and Smock (2005) observed that couples often have trouble citing exactly when a cohabitation began, as they “slide into cohabitation,” blurring the lines between cohabitation and singlehood (p. 995). In their analysis of the Fragile Families study, Teitler, Reichman, and Koball (2006) found that cohabitation reports are influenced by current relationship quality. Those currently cohabiting tended to revise past cohabitation status upward, but those who were no longer cohabiting revised it downward. In their analysis of cohabiting stepfamilies in the Add Health data, Brown and Manning (2009) identified boundary ambiguity among teens and their mothers in which adolescents and mothers living in the same household reported different family structures. In some cases in which there may have been coresidence, the teen and the parent could not agree that a cohabiting relationship existed.

Alternative Relationship Forms

Alternative relationship forms are increasingly addressed in the literature. A number of scholars have explored what they call “living apart together,” or LAT (Haskey, 2005; Levin, 2004; Milan & Peters, 2003; Strohm, Seltzer, Cochran, & Mays, 2009), a relationship in which couples "agree they are a couple; others have to see them as such; and they must live in separate homes” (Levin, 2004, p. 227). Scholars have acknowledged that couples may coreside sporadically (e.g., Roy et al., 2008), but there has been little exploration of intermittent cohabitation and its processes. LAT couples often have high levels of commitment, yet they are not living together. We would suggest that, in contrast, some couples who coreside may not agree that they are a couple, and others may not see them as such, even though they are living together. Rather than living apart together, these couples are living together apart, or LTA. Both kinds of relationships can be part of intermittent cohabitation patterns. In general, however, we know very little about the reasons behind or dynamics involved in LTA cohabitations among the poor and how they fit into patterns of intermittent cohabitation, suspended relationships, and coparenting. This paper seeks to investigate the patterns inherent in these unstable cohabitations.

METHOD

To investigate the nature of low-income mothers' cohabitation behaviors, we used an archive of ethnographic data on economically disadvantaged families who participated in the Three-City Study. This study was a longitudinal, multisite, multimethod project designed to examine the impact of welfare reform on the lives of low-income African American, Latino, Hispanic, and non-Hispanic White families and their young children (see Winston et al., 1999). Study participants resided in poor neighborhoods in Boston, Chicago, and San Antonio. In addition to longitudinal surveys and an embedded developmental study, the Three-City Study included an ethnography of 256 families and their children. These families were not in the survey sample but resided in the same neighborhoods as survey respondents.

Sample Description

Families were recruited into the ethnography between June 1999 and December 2000. Recruitment sites included formal child-care settings (e.g., Head Start), the Women, Infants, and Children (WIC) program, neighborhood community centers, local welfare offices, churches, and other public assistance agencies. Of the 256 families who participated in the ethnography, 212 families were selected if they included a child age 2 to 4 to ensure comparability across study components (the remaining 44 families had a child with a moderate to severe disability). At the time of enrollment, all families had household incomes at or below 200% of the federal poverty line.

The majority of mothers (42%) were of Latino or Hispanic ethnicity, with the largest groups being Mexican Americans, Puerto Ricans, and Dominicans, in that order. Over half of the mothers (58%) were age 29 or younger when they enrolled in the study, and a majority (57%) had a high school diploma or GED or had attended trade school or college. Forty-nine percent of the mothers were receiving Temporary Assistance to Needy Families (TANF) when they entered the study; one third of these were also working. The majority of mothers (56%) also indicated that they were neither married nor cohabiting at the outset of the study, but longitudinal interviews and observations of the sample revealed that more respondents were in marital or cohabiting relationships than said they were, and some were also serially moving from one intimate union to another.

Given our interests in the features of intermittent cohabitors, the present study focuses on a subsample of the ethnographic participants, namely, those mothers we identified as demonstrating intermittent cohabiting behaviors. The use of secondary data allowed us to identify intermittent cohabitations as an emergent pattern among participants. Among the 212 families with a child aged 2 to 4, we identified 45 mothers who evidenced various forms of intermittent cohabitation. These mothers were identified through modified grounded theory analysis of the ethnographic data (Charmaz, 2006). This process of subsample identification as it relates to an emergent typology of intermittent cohabitators is presented in the Results section of this article.

Ethnographic Methodology

To gather and to analyze ethnographic data on the mothers and their families, a method of structured discovery was devised to systematize and to coordinate the efforts of the Three-City Study ethnography team (for a detailed description of the research design of the ethnography, see Winston et al., 1999). An integrated and transparent process was developed for collecting, handling, and analyzing data that involved consistent input from over 215 ethnographers, qualitative data analysts, and research scientists who worked on the project over the course of 6 years. Interviews with and observations of the respondents focused on specific topics but allowed flexibility to capture unexpected findings and relationships among variables. The interviews covered a wide variety of topics, including living arrangements, intimate relationships, family economies, support networks, and neighborhood environments. Ethnographers also engaged in participant observation with respondents that involved attending family functions and outings; witnessing relationship milestones (e.g., a couple's decision to cohabit); accompanying mothers and their children to the welfare office, hospital, day care, or workplace; and noting both context and interactions in each situation. Ethnographers met with each family once or twice per month for 12 – 18 months, then every 6 months thereafter through 2003. Respondents were compensated with grocery or department store vouchers for each interview or participant observation.

Data Sources

The ethnography generated multiple sources of data that we used to examine mothers' cohabiting patterns. The primary data sources were ethnographers' field notes about their interviews and participant observations with families and transcripts of all of their tape-recorded interviews. In addition, we consulted transcripts of principal investigators' group and individual discussions with ethnographers and qualitative data analysts about the families. All sources of data were coded collaboratively (according to a general thematic coding scheme developed by the principal investigators) by ethnographers and qualitative data analysts for entry into a qualitative data management software application (nVivo) and then summarized into detailed family case profiles.

During the course of the ethnography, we followed criteria that have been established as appropriate to qualitative and ethnographic methods to assure the reliability and validity of the data collection, coding, and analysis (Atkinson, 2001; Denzin & Lincoln, 2005). Having two analysts devise the coding scheme, continually testing interpretations against the actual data (i.e., analytic induction), carefully coding, and establishing an acceptable level of interrater agreement were used to increase reliability and validity of content analysis. In addition, validity was increased by the prolonged engagement of the ethnographers with families; the ability to cross-check data across sources derived from multiple methods; monthly meetings between ethnographers and principal investigators; and the use of member checks, in which researchers asked participants for corrections or elaborations of researchers' interpretations.

Coding and Analysis

Three phases of data coding were conducted in this analysis. The first phase involved open coding of family profiles (in which all families were assigned the pseudonyms used here) to identify the subsample of intermittent cohabitors. To conduct our analysis, we reviewed the family profiles of the 212 families with a 2- to 4-year-old. Reading through each profile at least twice, we noted the types of relationships in which mothers engaged, which included stable marriages or cohabitations, serial cohabitations, intermittent cohabitations, LTAs, and abated unions (in which the mother does not engage in any romantic, sexual, or cohabiting relationship; see Burton, Cherlin, Winn, Estacion, & Holder-Taylor, 2009; Cherlin, Burton, Hurt, & Purvin, 2004).

Many scholars have noted the limits of retrospective data, as “memory belongs to the imagination” (Pearson, Ross, & Dawes, 1992), and research indicates that recall of cohabitation status is influenced by the current state of the relationship (Teitler et al., 2006). Some mothers reported intermittent cohabitations that occurred prior to the study, but to increase the accuracy of our analysis and conclusions, we limited our sample to mothers who had at least one intermittent cohabitation spell that was observed during the course of the study.

Continuing with the grounded theory coding sequence outlined by Strauss and Corbin (1998), after identifying women who engaged in intermittent cohabitations, we used axial coding of the transcripts and field notes to identify particular factors that emerged in these types of relationships. Closely examining the circumstances surrounding intermittent cohabitations and LTAs in particular, we observed the language women used to refer to partners, such as “boyfriend,” “friend,” “roommate,” or “baby daddy,” and the way they described their living situations, such as viewing themselves as cohabiting even when separated from a partner or insisting their child's father was not a member of the household even when he obviously lived in the home. We also noted the primary reasons women gave for reinitiating cohabitations, especially in LTA situations. Reasons generally centered around financial concerns, housing needs, and parenting.

After themes of finances, housing, or parenting emerged in all cases in the axial phase, we used selective coding (Strauss & Corbin, 1998) to identify the narrative story underlying intermittent cohabitations and LTAs in particular. Here we found that within people's practical needs, obligation accompanied the parenting ties of couples. Some couples reunited out of a mutual desire for cohabitation in addition to other needs. For these couples, separations had not always been voluntary in the first place. Among other couples, need led one parent to seek kin ties of last resort—the other parent. These parents did not view coresidence as a rekindling of their relationship, and help was often extended reluctantly. Women in LTA relationships in particular usually had little trust in men. If they trusted their partners at all, they often exhibited “compartmentalized trust” (see Burton et al., 2009), in which they trusted their partners to fulfill certain roles, such as meeting financial obligations or parenting responsibilities, but did not trust them overall as husbands or partners, and they did not call them such. Obligations to children mandated the assistance between parents.

Ultimately, couples were classified as being in an intermittent cohabitation if they met all of the following criteria:

• The couple had been in a romantic relationship in the past and cohabited for at least 1 month

• The couple had separated at some point for at least 1 month

• The couple was subsequently coresiding for at least 1 month.

The couple was classified as LTA if, in addition,

• The pair shared at least one biological child

• The coresidence was based on a sense of need or obligation on the part of one or both parents, rather than on romantic interest

• The woman said that she and the father were not in a relationship.

RESULTS

Intermittent Cohabitor Sample

Our analysis identified 45 women who engaged in an intermittent cohabitation during the study period. Of the 45, 8 were White, 20 were Latina, and 17 were African American, proportions similar to the composition of the ethnography as a whole. Within the Latina category, 9 women were Mexican Americans born in the United States, 8 were Puerto Rican, and 3 were of mixed heritage. Though there were a number of Mexican immigrant mothers in the study, none exhibited intermittent cohabitation patterns (this group was disproportionately represented among those in stable marriages); Dominican mothers, all in Boston, had some variation in relationship patterns, but none intermittently cohabited during the study. The women in our sample ranged in age from 16 to 42 and had from one to nine children. Of the 45, 18 met our criteria for being involved in an LTA relationship. Shared children were present in all LTA relationships and all but two of the other intermittent cohabitations. More than half of the mothers had children with more than one man. Only 6 women had ever been married, and women included were intermittently cohabiting with a man other than a legal husband. In each LTA case, the woman identified the relationship as LTA, usually by specifically saying that she and the man she lived with were “not together” or “separated” and referring to him as a “roommate,” a “tenant,” or her children's father (it was generally unclear what the man thought, as only the women were interviewed). Women engaged in intermittent cohabitations with both practical and emotional motives. Both separations and reunifications were sometimes voluntary and sometimes involuntary. In many cases, there were elements of choice that were constrained by outside circumstances, such as when a mother could apply for TANF or save her allotted months by cohabiting with a man who was able to make financial contributions.

Because our data focused on mothers, the voluntary or involuntary nature of separations and reunifications refer to the women's perspectives. Women who had multiple intermittent cohabitation spells often exhibited more than one pattern. Involuntary separations often occurred as a result of incarceration or because the man decided to leave (often to take up residence with another woman). When reunifications were voluntary, mothers rekindled romantic relationships with their partners, though the cohabitations generally had financial or other motivations as well. Involuntary reunifications usually resulted in an LTA relationship spell, but then occasionally drifted back to romantic attachment. There was a long middle ground between voluntary and involuntary, but most separations and reunifications tilted strongly toward one end or the other of the continuum. The terms voluntary and involuntary are used to indicate the mother's expressed preference in the context of the structural constraints she faced.

Voluntary Separations

Mothers in the Three-City Study ethnography cohabited for many reasons, only some of which involved romantic interest. Many cohabitations had primarily utilitarian purposes, and if a man no longer contributed adequately to the household, the mother might ask him to leave regardless of her romantic feelings. Edin (2000) termed such relationships “pay and stay,” as the man was only welcome in the home as long as he was financially useful. Diane, an African American mother in San Antonio, intermittently cohabited with her younger son's father. By the end of the study, the two had another son together, but Diane was unsentimental about their relationship and considered it mostly in terms of its financial advantages or liabilities. After a longer cohabitation spell, Diane found herself unemployed and told the ethnographer, “We couldn't get welfare [if he was in the house]. So I put him out.” His financial contributions could not make up for a lack of TANF payments, and Diane felt giving him household permanence was too financially risky. When the topic of marriage arose, she said, “What if … [I] get married to this person, then that person loses their job? Then what? Then it's another mouth to feed, another person in the house.” Diane viewed male parenting contributions as expendable, preferring to have primary or sole say in the standards of care for her sons. When her son's father tried to participate in potty training, Diane's response was, “you don't need to be disciplining them, I got to take care of them.”

Voluntary separations also occurred when the mother felt other costs of the relationship were too high, particularly in cases of domestic violence. Male violence significantly increases the chance of relationship disruption among cohabiting couples (DeMaris, 2000). Deanna, an African American mother in Boston, intermittently cohabited with George, the father of her youngest child. George was jailed multiple times for his violence against her and their son, but they resumed cohabiting when he was released. Deanna said she needed his financial support and also wanted George to be there as a father, but after their final breakup during the study, Deanna told the ethnographer “domestic violence was one reason why it didn't work out.” Though George's contributions were significant to Deanna, who had never held a job, his repeated violence was ultimately too much to bear.

Voluntary separations occurred in the context of relationships that mothers did not want to make permanent. Relationships were often rife with problems, such as the man's infidelity, violence, propensity for illegal activity, or excessive alcohol or drug use. The mothers often did not like their partner's friends or the way he spent his free time, and they often felt the man was generally unreliable. His opportunity cost at any given time, however, could determine whether he was allowed to remain in the household or not.

Involuntary Separations

Sometimes when a woman preferred for the man to stay in the home, he chose to leave, but most often involuntary separations resulted because of housing policies or incarceration. Prison has become an ordinary life course stage for many low-income urban men, particularly African Americans (Pettit & Western, 2004), and prison was a normal aspect of life for most families in this sample, regardless of race or ethnicity. Many mothers had had relationships with several men with records, and some women had never had a relationship with a man who did not have a criminal record. Several mothers reported having two or more partners incarcerated simultaneously. Most of these men were fathers of one or more of the mother's children. Their time ranged from a few days in jail to prison sentences of a decade or more. Incarcerations weighed heavily on stability and duration in mother's relationships. The ethnographer reported how Dana, an African American mother in Chicago, described her sons' father's efforts to remain part of the family while imprisoned:

Dana said that if he was not in prison he would be contributing financially and emotionally. She says that she writes him and tells him about the problems that she is having with the kids and then he writes them and disciplines them for their behavior. She said that his letters have an effect on her kids' behavior.

When a cohabitation was interrupted by jail time, sometimes a mother put the relationship on hold with plans to reunify, whereas at other times she found another partner. Dana's partner's parenting contributions from prison reduced Dana's need to seek out another father figure to help her with her eight sons.

The length of a prison sentence often factored into mothers' decisions regarding their relationship status. Yuri, a Mexican American mother in San Antonio who intermittently cohabited with several men, found her cohabitation with Mark cut short when he went to prison. She debated whether to reunite with him when his 12-year sentence was up, especially as she had begun a new cohabitation with Jed. She said, "Jed knows about Mark, and I never took his pictures down, but I thought maybe it might be time for me to move on." Mark's long sentence made waiting impractical, regardless of Yuri's preferences.

When mothers received subsidized housing benefits, housing policy could limit their ability to cohabit, and mothers sometimes had to undergo involuntary separations when they felt they were under surveillance. Public housing vouchers usually included only the mother and her children on the lease, with policies regulating who else was allowed to stay in the home. Though TANF did not count the resources of a cohabiting male in assessing eligibility, public housing and other subsidized housing, such as Section 8, counted the income of all resident adults in determining rent payments. Cohabiting was a financial liability if a man's income reduced a mother's rent subsidy, and among poor mothers, living in subsidized housing is associated with living without a partner (Curtis, 2007). Eliza, a Mexican mother in San Antonio, discussed with the interviewer the reason for her intermittent cohabitation with her children's father:

E: And they'd be checking up on the houses and stuff. See, and that's why, you see, cause so like, he's not supposed to be here. You know, but it's like, how I tell him, you know, “You're the father of my kids, you wanna come see your kids, you can come see them whenever you want, you know, I'm not gonna take that away from you.” And if the office has something to tell me, then I'm just gonna tell them, “You know what? This is the father of my kids, he does come and see them, you know, what, now you're gonna tell me he can't?”

I: But he, but he can visit?

E: Yeah, he can visit, just that he can't live here, yeah. So, that's, that's how, how they go about if someone's living with you, if they, if someone reports it, or if like, if you get into problems. Yeah, that's how they find out.

Some mothers allowed fathers to cohabit until they were caught, but that sometimes meant that the man was banned from the housing complex altogether, forcing a longer term separation.

Voluntary Reunification

When a mother reunited with a partner, acknowledged both the cohabitation and the romantic nature of the relationship, and did not say that she wished to be living independent of her partner, we classified the reunification as voluntary. Even under voluntary circumstances, however, a mother's desires for reunification usually had utilitarian foundations. Marjorie, an African American mother of nine in Chicago, lived intermittently with Jamal, the father of her younger children. She discussed with the interviewer the difference in her partner Jamal's contributions when he lived with Marjorie versus when he was living with another woman with whom he had a concurrent relationship:

I: So financially you said that, does he help out in a good way sometimes?

M: I guess when he gets a job. Well he bought two boxes of cereal and a gallon of milk one day. [laughing]

I: … We've talked about the financial things, the kids. You said he helps out with watching the kids?

M: Yeah, he'll help watch.

I: Like when you were at work?

M: Right. He'll help watch them. He'll pick them up from school.

I: No problems with that?

M: When he moved in with [his other girlfriend] he stopped picking them up.

Even though Jamal periodically left Marjorie for his other partner, the pleasure of his company and his contributions to the household were valuable enough to make it worth Marjorie's while to reunite with him when he was available.

Involuntary Reunification and Living Together Apart

Because cohabitations are often perceived as easy to dissolve, involuntary reunification may seem like a contradiction in terms. The nature of these relationships speaks to the bond shared children create (Edin & Kefalas, 2005) and to the tenaciousness of suspended relationships among coparents (Roy et al., 2008). When Stack (1997) studied a low-income community in the late 1960s, she found that extended kin networks kept people afloat through hard times. More contemporary scholars have indicated that these extended kin networks have broken down and for many people may be virtually nonexistent (Dominguez & Watson, 2003; McDonald & Armstrong, 2001). Among our participants, when a former partner was in need, the bonds of shared parenthood were sometimes the last vestige of extended kin that could be tapped for resources, leading to an LTA relationship. In most cases, mothers did not feel that involuntary reunifications were cohabiting relationships and often went out of their way to clarify that they were not in a relationship, even though they were living with their children's father. This was generally true even if the couple was sexually intimate. These relationships are what we call “living together apart.”

Parenting Based LTAs

An LTA couple often had had little or no interest in being with one another but maintained a commitment to coparenting. Parenting could benefit the children, the mother, and the father on a practical level—the parents could pool finances, the children lived in a two-parent home, the mother did not have to parent alone, and the father was able to be directly involved with his children on a day-to-day basis. For instance, Sonny, a Mexican American man, and Joanne, a White woman, lived in San Antonio and had an 11-year relationship history riddled with domestic violence. Sonny had regularly moved in and out of the household, sometimes because he was in jail, sometimes because of housing policy, and at other times, because Joanne evicted him. When an interviewer mistakenly assumed Sonny and Joanne were married, she said, “We're not married. We're bound by children.” Joanne no longer felt they were a couple, but she still viewed Sonny's role as a father to their six children as very important. A field note explained the current state of Sonny's coresidence:

Sonny has been helping Joanne with the day-to-day care of the children. Importantly, she notes that she and this man work together to be mother and father to their children, even though they cannot be together as husband and wife. Joanne says she has had only one boyfriend since she and her children's father separated, and she tried to keep it clear that he was not taking over their father's role with them. She says she would never bring a guy into the house because she thinks it would be disrespectful and confusing to the children.

In addition to being a father figure, Sonny's presence allowed Joanne, who had no car, to work or run errands without having to arrange child care or bring her six young children with her. Sonny slept at Joanne's, but Joanne was clear that they were not intimate, as that would give Sonny the kind of control over her that she felt a husband or cohabiting partner expected to have. She said, “Because once we sleep together, it's all over but the complaining and `you can't go here, you can't go there.'”

Similarly, Yasmin, a 23-year-old Puerto Rican mother of a preschooler in Chicago, allowed Joel to live with her in part because she saw him as the “perfect father.” When asked about their relationship status, however, Yasmin said, “We live together but we are not really together together,” and she identified him as “my baby's father.” She did not trust him as a partner because of his previous infidelities and had no interest in rekindling their relationship, and she outlined specific conditions of his living in her home: “I was like, the only way you gonna stay here is if you do what I want you to do, if you do what I tell you to do. That's pay all the bills … and do everything.” The field notes stated that “Yasmin doesn't want him there but he's helping out a lot and [Yasmin says] `That's why I haven't kicked him out yet.”' Joel slept in the same bed with Yasmin and their son Jake because the apartment was so small, but Yasmin said, “We need more space; I need more space.” Joel left for work very early and Yasmin worked a later shift, so though they both spent substantial time with Jake, they rarely saw one another awake. Yasmin dated other men and planned to terminate the relationship as soon as she no longer needed Joel's assistance, at which point she assumed he would move back to his mother's home.

Parenting-based LTAs could also be motivated by a mother's strong feelings about presenting as a conventional family and preserving her own sense of respectability (see Roy & Burton, 2007). Marka, a Puerto Rican mother in her mid-20s, began dating John in her teens in an on-and-off relationship involving frequent domestic violence. Marka moved to try to escape from John and dated others, but when Marka was 22, she and John had a daughter together. While John was serving a three-year prison sentence, he called Marka constantly and was anxious to resume their relationship. Conventional parenting ties were especially important to Marka, for her own sake as well as her daughter Tia's. Marka was reluctant to resume cohabiting, but said she believed John would change in part because “he is Tia's father.” Though she said, “I don't feel anything for him anymore” and she dated another man, she allowed John to move back in with her when he was released from prison. When he moved in with her, she watched him carefully for many months, until finally she told the ethnographer

He was working making over 30 per hour through a union job. For Thanksgiving, he baked the whole meal, turkey, lasagna, rice and beans and I started to get confused. I decided that it was worth trying it out as a family. But I am not sure about him yet.

Later, she conceded that they had resumed their relationship and she had conceived another child with him. Still, she said that “I am not in love, you know?” and that she was ready to terminate the relationship. On the other hand, she did not want to “be having kids in my 30s” and she was opposed to having children with several different men. In many ways, Marka was more invested in the social appearances of the relationship than in the relationship itself.

Housing-based LTAs

Even if their personal ties were tenuous, mothers and fathers were reluctant to let their coparent become homeless. LTA coresidences based on accommodating housing needs, however, were often devoid of any trust or affection. For instance, Tonya, an African American mother in Boston, allowed Curtis, the father of her youngest child, to live in her home. The ethnographer referred to Curtis as Tonya's “recliner tenant” because he slept on a recliner in her living room. Tonya did not find the relationship itself worthwhile. She was upset that Curtis did not contribute food to the household and was angered when he tried to get in her bed. After he had lived in the apartment for about 1 year, Tonya had not given him a key.

Tonya suffered from depression, had been welfare reliant her entire adult life, and had unreliable alternative sources for cash income. Tonya's mother, sisters, and adult daughter also lived in poverty. Family members traded in child care and surplus WIC food allotments, but no one had cash to share. Though Tonya may have initially hoped that Curtis would contribute socially to the household, her decision to allow Curtis to live with her was indicative of the “pay and stay” rule (Edin, 2000): Curtis offered token rent money. Although Curtis' household role was marginal—he did not play any kind of parental role, contributed no tangible necessities, and was rarely in the apartment except to sleep—his $50-a-week contribution for room and board did provide a small financial cushion. In addition, allowing him to stay meant that Tonya did not have to see her child's father living in the street.

Tonya may also have had hopes that if she gave it time, the relationship would change for the better. In a follow-up interview conducted months after she found a new apartment, Tonya said that she no longer lived with or saw Curtis because “he just wasn't for me. Just too boyish, wasn't grown up yet.” Her statement that she finally let him go because he wouldn't “grow up” indicates that she may have kept the relationship in suspension in hopes of a transformation. Like Marka, who believed John would change, “because he is Tia's father,” Tonya may have hoped that Curtis would care enough for her or his daughter to “grow up.” Without the connection of their shared child, however, Tonya probably would not have allowed Curtis live with her or have maintained any hope for their relationship.

Though public assistance provides some low-income women with housing resources that men may try to access, other women lack social, familial, and financial resources and risk homelessness. These women may call on the obligations to their children to obtain housing and other financial support from their children's fathers. When men controlled the residence, they had the power to set certain parameters regarding issues such as housework or sex. In describing the LTA situation of Margerita, a Mexican American mother in Chicago, the field notes reported that

Margerita has been at the mercy of her callous, selfish [partner] since the birth of their first child 15 years ago. She feels dependent on the little financial contribution he makes. Jesus pays only the mortgage and Margerita covers everything else. In the past, Margerita was homeless without Jesus. She has not been able to rely on assistance nor her parents for support.

As is true for many low-income women, what kin networks existed were limited in the support they could provide. Margerita made too much to qualify for many forms of public assistance while not earning enough to fully support herself and her three children.

Jesus, who was born in Mexico, took advantage of Margerita's dependence. She did all of the housework, and Jesus did almost nothing—not even repair a broken window in the winter. Avoiding homelessness drove Margerita to stay; even when she landed a better job, she did not believe she could make it on her own. Even though the coresidence lasted for several years, Margerita never conceded that their relationship had resumed. In describing Margerita's perspective, the research notes reported

Margerita says that [Jesus] makes twice as much as she does and that it's his money, “That's why I say we are only roommates.” I ask her if they are only roommates, we both laugh, and she says, “Well once in a little while.” Margerita blushes, tilts her head a bit to the right and shrugs her shoulders. [The ethnographer] blushes too because they both understand that she is admitting to having intimate relations with him. “I see it as roommates, because we don't get along, we just try to live day by day calm because of our kids.” He has told her that if she leaves that he will take her kids away. She tells me that that is part of the reason that she stays with him because she does not have the money to hire a lawyer.

Margerita had no trust in Jesus other than her belief that he would continue to provide housing for her and their children. She placed little value on his minimal contributions to parenting, but rather believed staying preserved her own role as a mother and gave her children stable housing. She agreed to have sexual relations on occasion because “it is the only time when [we] don't argue.” In contrast to Yasmin, who controlled Joel because she controlled their housing and access to their son, Margerita saw Jesus as the person with all the control, thus obligating her to stay in their LTA relationship on his terms.

DISCUSSION

Couples cohabiting on the edge, like most cohabiting couples of yesteryear, tend to be found among low-income populations. The lack of material resources available to low-income women hinders their ability to form more conventional intimate partnerships, though they may desperately need the resources those partnerships provide. Developing a permanent attachment to one man limits a woman's network and impedes her ability to draw on all resources potentially available to her. As such, even when a couple would prefer to cohabit, they may be forced to separate in order to meet their family's needs. On the other hand, they may feel forced into cohabitations they would prefer not to have. In the case of the LTA relationships observed here, women had practical and emotional connections with men whom they did not view as even casual boyfriends, much less committed partners. Even though cohabitations are thought to be easy to dissolve, in these cases, shared parenting tied people together long after their interest in their relationship as a couple had ended. Ongoing concomitant obligations led to coresidence that does not fit any current notions of cohabitation.

Until recently, marriage was the only socially acceptable way to maintain a cohabiting relationship. As a result, as rates of cohabitation climbed over the past two decades, observers focused on cohabitation's connections to marriage. As the dominance of marriage diminishes in American family life, it is clearer that some cohabiting relationships do not fit into a marriage-oriented perspective. Intermittent cohabitations demonstrate the difficulty of shoehorning cohabitation into the marriage paradigm, even when a couple maintains a long-term relationship.

In some sense, the reasons women gave for maintaining intermittent and LTA relationships resemble the practical considerations that were central to marriage until the 20th century, such as shared parenthood and the pooling of resources. Occasionally marriages still mirror these values, as among unhappy middle-class married couples who remain together for the sake of the children or because they cannot afford to support two households (Martin, 2007; Martin, Cherlin, & Cross-Barnet, in press). Until the 20th century, sexual relations and companionship were less important for marriage (Coontz, 2005). In this way, for mothers in the Three-City Study ethnography, intermittent cohabitations, and LTA relationships in particular, resemble the marriage-based families of the past more than the present.

Yet sharp differences with married families of the past also exist. Notably, in our sample, intermittent cohabitors frequently had intimate relations with other partners during their separations, and LTA partners frequently had sexual partnerships with people outside of the household even when they were coresiding. Though some intermittent cohabitation spells endured for a year or more, others lasted little more than a month. Even among longer term couples, there was generally no sense of permanence. Within LTA coresidences, one or both people usually wished to cease living together as soon as possible.

In addition, we must consider the changing role of fathers in these configurations. From the Industrial Revolution until relatively recently, heterosexual relationships assumed gendered divisions of labor that emphasized male bread-winning and female caregiving, roles that were considered essential to functional marriages (Coontz, 2005). In contemporary low-income families, mothers often have access to financial resources outside of employment that are not perceived as available to fathers, and it is posited that low-income fathers often abandon their families because their financial support is not essential (Roy, 1999; Willis, 1999). Although there is a continued emphasis on fathers' role as providers (e.g., Roy, 2005; Wilson, 1996), there is also an increased expectation that fathers should participate in relational aspects of family life. Many lower income fathers would like to provide financially, but have also come to value their relationships with their children and their role as caregivers (e.g., Edin et al., 2002; Roy, 1999; Roy et al., 2008). Maintaining connections with their children's mothers may be a survival strategy for obtaining resources such as housing, but it is also an essential component of maintaining involvement in their children's lives.

As researchers and practitioners, we must consider relationships that stretch conventional boundaries of family life, relationships that are evolving, ambiguous, and sometimes short term. It may be difficult to determine the exact nature of many cohabiting relationships, as even the people involved may have different interpretations. Constant shifts in cohabitation status between the same two people can further blur these lines. Household composition may have frequent shifts that are not even acknowledged by household members. Intermittently cohabiting couples are, nevertheless, a part of the reality of intimate partnerships. Relationships may not have the shared emotional bonds that contemporary Americans consider as a necessary component of intimate partnerships nor the sexual exclusivity and stability that were emphasized in the past. What remains are needs-based relationships centered on the ties of shared parenthood: These partners are, as one woman said, “bound by children.”

Although LTA relationships are not the norm, they provide the potential for developing or extending our understanding of the changing nature of personal life in the same way that other ethnographic accounts of unique cases often enrich our understanding of the social world (Small, 2009). The broader significance of intermittent cohabitations and LTA relationships may be to reorient the viewpoint of researchers, practitioners, and policymakers regarding the way they approach the concept of relationships, families, and household composition. Just as cohabitation itself became mainstream, as more families face economic hard times, we may see widespread adaptations within cohabiting relationships. Children may be more central to family ties than romantic relationships between their parents. By not assuming that all relationship patterns fall into existing categories, we may be able to systematically observe families in as yet undefined configurations. In this way, research on the edges of personal life may help us to better understand the diverse nature of today's families and their impact on children and society.