Marriage and Child Well-Being: Research and Policy Perspectives

Abstract

Over the past decade, the linkages between marriage and child well-being have attracted the attention of researchers and policy makers alike. Children's living arrangements have become increasingly diverse and unstable, which raises important questions about how and why family structure and stability are related to child outcomes. This article reviews new research on this topic, emphasizing how it can inform policy debates about the role of marriage in reducing poverty and improving child outcomes. It also pays special attention to new scholarship on unmarried, primarily low-income families, the target of recent federal marriage initiatives, to appraise the potential contributions of family research to ongoing policy discussions.

Historically, most children were born to and raised by married parents. In recent decades, the living arrangements of children have become increasingly varied and unstable (Kreider, 2007). The rise in unmarried families, which has coincided with an increase in child poverty, has generated considerable scholarly and policy debate (Amato & Maynard, 2007; McLanahan, Donahue, & Haskins, 2005; Moynihan, Smeeding, & Rainwater, 2004). Is marriage the ideal living arrangement for children, and if so, should government encourage marriage?

This article is designed to appraise the research published since 2000 that has addressed the relationship between marriage and child well-being, as well as its ramifications for policy and program formulations. I begin with a description of the recent trends in children's living arrangements, paying special attention to racial-ethnic and socioeconomic variation. These patterns guide the contemporary debate about marriage and serve as the backdrop for the federal marriage initiatives introduced in the 2000s, which I describe in the next section. I consider why these programs tend to be directed toward the low-income population and assess some of the controversies surrounding the initiatives.

Policy prescriptions about marriage draw heavily on research about the association between family structure and child outcomes. Consequently, I review what has been learned over the past decade about how and why family structure and stability are related to child well-being, with particular emphasis on the measurement and conceptual issues that will likely shape subsequent research. I then provide an assessment of the literature to date on marriage and well-being among low-income or at-risk families, relying primarily on research that uses the Fragile Families and Child Well-Being Study, a survey whose sample reflects the at-risk populations that contemporary marriage policies target. I consider how the patterns characterizing at-risk families differ from those of the general U.S. population and highlight the distinctive features of this line of research, such as the theoretical and empirical advances in modeling marriage behavior among low-income mothers. Results from these studies arguably represent the best answers social science can provide to inform contemporary policy debates. Throughout the review, I aim to synthesize recent literature, specifying areas in which there is consensus versus preliminary or inconsistent evidence, in which case I also assess the rigor and sophistication of studies. In concluding, I suggest future directions for research and data collection and discuss how family research can inform policy making.

Childhood Living Arrangements

Although most children still live in married families of two biological parents, the proportion has declined over time. In 2009, 67% of children resided in this family form, whereas in 1970, this figure was 84% (Kreider, 2007; U.S. Census Bureau, 2010). About 23% of children resided in single-mother families in 2009 versus 11% in 1970. Roughly one half of children can expect to spend part of their childhood outside of a married-parent family (Bumpass & Lu, 2000).

These patterns are more pronounced among minority and low-income children. In 2009, 35% of Black children resided in two-biological-parent married families, and 50% lived in single-mother families. Among Hispanic children, 64% lived in two-biological-parent married families and 25% lived in single-mother families in 2009. The comparable figures for White children were 76% and 18%, respectively (Kreider, 2007; U.S. Census Bureau, 2010). More than one half of poor children reside in a single-parent family (Fields, 2003).

This dramatic shift in children's family experiences reflects both the high levels of divorce and repartnership and the rising share of nonmarital births. Although divorce rates have stabilized since the mid-1980s, they appear to be climbing among less educated couples (Cherlin, 2005). Divorced parents are likely to form new partnerships, whether through unmarried cohabitation or remarriage, thus exposing children to family instability (Cherlin, 2009).

The proportion of births to unmarried mothers rose from 18% in 1980 to a record high of 39% in 2006 (Martin et al., 2009). In 2006, 27% of births to Whites, 50% of births to Hispanics, and 71% of births to Blacks occurred outside of marriage (Martin et al., 2009). The rise in unmarried births is largely due to a corresponding increase in cohabitation; more than one half of all unmarried births are to cohabiting mothers. In fact, the fertility rates of cohabiting women (73 per 1,000) are nearly as high as married women (77 per 1,000) (Dye, 2008). Birth context is closely linked to subsequent transitions in family living arrangements. Children born to married parents typically experience relatively few family transitions during childhood, whereas those born to single and cohabiting mothers experience substantially more transitions, on average. Minority children tend to experience more transitions during childhood than White children, largely because they are more likely to be born outside of marriage (Osborne & McLanahan, 2007; Raley & Wildsmith, 2004).

The Marriage Debate

Contemporary policy discussions about marriage and family structure have a long history (Coontz, 2004; Cott, 2000). Indeed, the publication of The Negro Family, also termed the Moynihan Report, in 1965, continues to inform today's family policy discussions and academic research. Moynihan pointed to African American family structure, namely levels of unmarried childbearing and single-mother families that were higher than those of Whites, as a key contributor to African American socioeconomic disadvantage and welfare dependence that spanned generations. The “tangle of pathology” that Moynihan identified largely ignored the critical role of structural factors, such as discrimination and barriers to employment and was predicated on the assumption that unmarried childbearing led to poverty, as opposed to the reverse. The controversy surrounding this causal relationship has changed little in the intervening years, according to one recent assessment (McLanahan, 2009), and continues to capture the interest of leading scholars. In fact, the Annals of the American Academy of Political and Social Science published an issue in 2009 devoted to the Moynihan Report.

More broadly, countless academic articles on family structure, poverty, welfare reform, and child well-being address the ongoing marriage debate. Scholars often use the debate to motivate their research questions and then draw general conclusions about the viability of various policy initiatives, especially marriage promotion, on the basis of their findings. The topic was the subject of the fall 2005 issue of Future of Children as well as the November 2004 issue of Journal of Marriage and Family. The latter featured a series of pieces on the future of marriage (e.g., Amato, 2004; Cherlin, 2004; Smock, 2004), the meaning and experiences of marriage for various subgroups (e.g., Edin, Kefalas, & Reed, 2004; Kurdek, 2004; Oropesa & Landale, 2004), and policy ramifications (Huston & Melz, 2004). At the same time, policy makers rely on academic research to bolster their claims about the linkages between family structure and well-being (Amato & Maynard, 2007; McLanahan et al., 2005).

The current marriage debate centers on questions concerning the role of government in encouraging marriage and discouraging unmarried childbearing. The policies are primarily directed toward at-risk groups, namely the low-income and poor populations who rely on government assistance (Nock, 2005). Whether marriage ameliorates some of the disadvantages linked to poverty and single-mother families is a topic that policy makers and researchers hotly debate (Amato & Maynard, 2007).

Proponents of the marriage movement insist that marriage confers a host of benefits to children, adults, and communities, and thus it is in society's interest to promote marriage (Nock, 2005; Waite & Gallagher, 2000). Poverty and unmarried childbearing are costly to society. More widespread marriage, particularly among the low income, might help reduce the welfare rolls and alleviate poverty. Marriage also may reduce nonmarital childbearing and enhance child well-being (Amato & Maynard, 2007).

Opponents are more tentative about the benefits of marriage, arguing that much of its apparent advantages are due to selection factors rather than marriage itself. That is, marriage does not really make people happier, healthier, and more financially secure. Instead, happy, healthy, secure individuals are more likely to marry in the first place (Acs, 2007; cf. Waite & Gallagher, 2000). The benefits of marriage are unevenly distributed. For instance, remarriages or marriages involving children from prior unions are less stable and more conflicted than are first marriages and marriages that include only shared children, respectively. Sometimes, marital breakup can be beneficial for children and adults, particularly in cases of high marital conflict or abuse (Amato, 2004). Another concern opponents raise is whether welfare funds should be diverted to support marriage activities as opposed to being spent directly on easing the financial hardship that low-income mothers and their children experience (Huston & Melz, 2004).

Federal Healthy Marriage Initiatives

The signal role of marriage in alleviating poverty (and it associated societal ills) and fostering child well-being is at the core of ongoing federal efforts to encourage “healthy” marriages and responsible fatherhood. A central component of the Personal Responsibility and Work Opportunity Reconciliation Act (PRWORA) of 1996, or welfare reform, was the promotion and maintenance of two-parent (married) families. Since then, the federal government has allocated funds to support research and programs to facilitate marriage, especially among the low-income population, in which marriage rates are the lowest. The U.S. Congress found that “(1) marriage is the foundation of a successful society and (2) marriage is an essential institution of a successful society which promotes interests of children” (Administration for Children and Families [ACF], n.d.). The Deficit Reduction Act of 2005 allocated funding of $150 million per year to support healthy marriage promotion and fatherhood activities by the states for the 2006–2010 fiscal years (ACF, n.d.). The ACF (n.d.) in the U.S. Department of Health and Human Services spearheads the federal Healthy Marriage Initiative (HMI), which is guided by the following mission: “to help couples, who have chosen marriage for themselves, gain greater access to marriage education services, on a voluntary basis, where they can acquire the skills and knowledge necessary to form and sustain a healthy marriage.”

Healthy Marriage Programs

The federal government supports three large demonstration projects to evaluate the extent to which various relationship skills programs promote stable relationships and marriage, as well as child well-being. Until recently, marriage education and relationship skills programs had been formulated for a primarily White, middle-class audience who were engaged or already married (Dion, 2005). There is some evidence that such programs are beneficial to more diverse populations (Stanley, Amato, Johnson, & Markman, 2006). The federally supported demonstration projects have included tested programs that have been successful among middle-class populations and have been modified to reach a disadvantaged audience by shifting from an orientation toward reading and homework to a more interactive, conversational, and role-playing style that is accessible to those with lower levels of education. The programs now address the barriers to relationship stability and marriage that are common among disadvantaged populations, including multiple-partner fertility, gender mistrust, financial planning, prior and ongoing abuse, and the lack of married role models (Dion, 2005).

For instance, Building Strong Families (BSF), an ACF-funded, 9-year, multisite demonstration program conducted by Mathematica Policy Research, is structured to promote strong relationships between couples with a new baby or who are about to have a child together and to help couples who desire marriage to achieve it (Dion et al., 2008). Using programs such as that designed by John and Julie Gottman, Loving Couples Loving Children, formulated specifically for low-income populations and derived from the successful Bringing Baby Home program (Shapiro & Gottman, 2005), BSF aims to provide group training as well as individual and family support to increase healthy marriages and to improve couple relationship quality and child outcomes. An experimental design permits a rigorous evaluation of the BSF's effectiveness with couples followed up three times over a 5-year period. Other ACF-supported programs include the Supporting Healthy Marriage initiative, which addresses instability among low-income married couples with children, a population composed primarily of Whites and Latinos, and the Community Healthy Marriage Initiative, which features a range of activities to promote marriage and responsible parenting using a community saturation model that ideally will reduce divorce and nonmarital childbearing as well as improve paternity establishment and child support compliance (Dion, 2005; Knox & Fein, 2009). Empirical studies of the effectiveness of the demonstration programs are not yet available.

Defining Healthy Marriages and Relationships

Of particular importance from both a research and policy perspective is the definition of a “healthy” marriage. According to the HMI, healthy marriages are “mutually enriching” and “both spouses have a deep respect for each other” (ACF, n.d.). But there are considerable challenges involved in trying to define and then encourage healthy marriages. Child Trends, in work supported by the ACF, developed a framework for conceptualizing and defining healthy marriages, in which it stressed the importance of distinguishing between the antecedents or consequences of a good marriage on the one hand and the characteristics that make it a good marriage on the other hand. Factors such as commitment, satisfaction, communication, conflict resolution skills, and other indicators were identified as integral components of a healthy marriage (Moore et al., 2004). The meaning of a healthy marriage merits additional conceptual and theoretical exploration, which in turn can inform research as studies of family structure and child well-being typically have not considered variation among married families (e.g., in terms of marital quality). Indeed, despite the new focus on healthy marriages, current policies on marriage derive from the growing empirical literature on family structure and child well-being, which is reviewed below.

Family Structure and Child Well-being

Over the past decade, evidence on the benefits of marriage for the well-being of children has continued to mount. Children residing in two-biological-parent married families tend to enjoy better outcomes than do their counterparts raised in other family forms. The differential is modest but consistent and persists across several domains of well-being. Children living with two biological married parents experience better educational, social, cognitive, and behavioral outcomes than do other children, on average (e.g., Artis, 2007; Broman, Li, & Reckase, 2008; Brown, 2004; Carlson & Corcoran, 2001; Manning & Lamb, 2003; Teachman, 2008; Videon, 2002). Variation in well-being among children living outside of two-biological-parent married families (e.g., married step, cohabiting, and single-parent families) is comparatively low and often negligible (Artis, 2007; Brown, 2004; Manning & Lamb, 2003). The benefits associated with marriage not only are evident in the short-term but also endure through adulthood (Amato, 2005; Biblarz & Raftery, 1999; Hill, Yeung, & Duncan, 2001; Teachman, 2002, 2004).

The relationship between family structure and child well-being appears to be more pronounced for White than for either Black or Hispanic children (Dunifon & Kowaleski-Jones, 2002; Manning & Brown, 2006). For example, one study found that time spent in a single-parent family was negatively related to White children's math scores and delinquency but unrelated to the same outcomes for Black children (Dunifon & Kowaleski-Jones, 2002). This race difference may reflect variation in the levels of social support provided to single parents in Black versus White communities, which would contribute to parental well-being and in turn parental effectiveness. Or, to the extent that diverse family forms are more normative among minorities, living outside of a two-biological-parent married family may be less stigmatizing or stressful for minority children than for White children (Heard, 2007a). Moreover, Black and Hispanic children typically experience fewer economic benefits from parental marriage than do their White counterparts (Manning & Brown, 2006). Additional research is needed to establish the extent to which race differences in the relationship between family structure and child well-being persist across various outcomes and to expand existing theoretical explanations to accommodate the significance of race-ethnicity.

Theoretical Explanations

Economic resources, parental socialization, and family stress or turbulence largely mediate the relationship between family structure and child well-being (e.g., Amato, 2005; Carlson & Corcoran, 2001; Demo & Fine, 2010; Wu & Thomson, 2001). Selection also may play a role, a point that skeptics of marriage promotion raise (Huston & Melz, 2004). Identifying the mechanisms underlying the association between family structure and child outcomes is important because it can guide policy initiatives aimed at promoting child well-being.

Economic resources

Economic resources, including parental education and income, not only have a direct influence on child well-being but also are crucial for the well-being of parents and facilitate effective family functioning (Carlson & Corcoran, 2001). Indeed, child poverty is highly differentiated by family structure, ranging from a low of about 7% for children in two-biological-parent married families to a high of nearly 44% in single-mother families (Manning & Brown, 2006). Economic deprivation impedes effective parenting by making it harder for parents to provide all of the material goods and services that are linked to child development (Amato, 2005) and by contributing to parental stress (Demo & Fine, 2010).

Parental socialization

Effective parents provide children with warmth and affection as well as acceptance and support. They are responsive to children's needs and regularly engage and spend time with children. They also exhibit consistency in rules and their enforcement through moderate control or discipline. (Thornton, 2001). Parents who are preoccupied by marital conflict may be less able to engage in competent or consistent parenting (Sun, 2001). And solo parents (typically mothers) who lack a partner to cooperate and consult with about parenting decisions and stressors tend to exert less control and spend less time with their children (Dunifon & Kowaleski-Jones, 2002; Sandberg & Hofferth, 2001), although those associations are confounded with socioeconomic disadvantage (Kendig & Bianchi, 2008).

Both mothering and fathering independently mediate the association between family structure and child well-being (Carlson, 2006; Hofferth, 2006). Family structure variation in mothering is comparatively modest, whereas that of fathering is more variable because it depends in part on residential status. Nonresident fathers are less involved with children, on average, especially if their parenting responsibilities are spread across multiple households (Manning, Stewart, & Smock, 2003). Father involvement is particularly beneficial for resident children, although nonresident fathering remains associated with well-being. Marriage is also related to paternal investment. Hofferth and Anderson (2003) found that the positive effect of marriage on child well-being persisted net of fathers’ own characteristics, but this was not true for biological status, which suggests that marriage itself may foster paternal engagement (alternatively, there may be other unmeasured characteristics that select good fathers into marriage).

Family turbulence

Children fare best in stable family environments in which well-adjusted parents have established consistent routines. Family turbulence, such as multiple school or parental employment changes, or a dramatic decline in parent health, are linked to lower levels of child well-being (Teachman, 2008). More broadly, family conflict or disruption, which often coincides with economic instability and inconsistent parenting, is stressful for children and therefore can compromise their well-being (Carlson & Corcoran, 2001; Demo & Fine, 2010).

Selection

Central to the ongoing debate about marriage and well-being is the role of selection versus causation (Hofferth, 2005). The selection perspective holds that it is not family structure per se that influences child well-being but the characteristics of parents that are related to both family structure and child outcomes. That is, children fare best in two-biological-parent married families, on average, because adults who form and maintain such families are the most stable, well-adjusted, resource-rich individuals. This proposition is difficult to rigorously test because children's family structure cannot be randomly assigned. Equally difficult to establish, though, is the extent to which the apparent benefits of marriage are causal.

In an effort to account for selection, researchers typically control for factors associated with family structure and child well-being, such as parental characteristics, parental resources, and family environment (Hofferth, 2005). Consistent with the selection perspective, these factors usually reduce the effect of family structure, often to nonsignificance (Amato, 2005). A failure to account for selection overstates the causal effect of family structure on child well-being (McLanahan & Percheski, 2008). But this approach does not rule out the possibility of a causal effect of family structure on child well-being (Amato, 2005). Indeed, on the basis of their review of the extant literature, Waite and Gallagher (2000) concluded that marriage may change people in ways that can enhance well-being.

More sophisticated approaches to dealing with selection include fixed-effects models, sibling studies, and natural experiments. Fixed-effects models of the linkages between family structure and child well-being typically have yielded few, if any, significant differences, perhaps because these models examined variation within children over time rather than across children in various family structures (e.g., Aughinbaugh, Pierret, & Rothstein, 2005; Kowaleski-Jones & Dunifon, 2004). Sibling studies enable researchers to compare children who experienced different family structure histories (e.g., timing and duration of exposure) or who resided in the same family but have unique relationships to the parents (e.g., stepfamilies with shared and blended children). A recent review of these studies revealed mixed findings, which reflects in part the unique populations analyzed. Indeed, the generalizability from this approach is limited in part because, by design, these studies are constrained to children in specific family forms, such as stepfamilies (McLanahan & Percheski, 2008). Similarly, natural experiments involving comparisons of children who experienced parental divorce versus death have generated equivocal findings. Evidence for selection comes from a study that showed outcomes for children in widowed single-mother families were more similar to children in two-biological-parent married families than to divorced single-mother families (Biblarz & Gottainer, 2000). Yet parental absence by death versus divorce was similarly related to economic well-being during adulthood, which supports the conclusion that the long-term benefits of parental marriage are causal (Lang & Zagorsky, 2001). Future research must continue to pursue innovative approaches to deciphering the relative contributions of selection versus causation in the association between family structure and child outcomes. A promising avenue to gain leverage on this issue has received considerable attention over the past decade: variation among two-parent families.

Heterogeneity in Two Parent Families

Children in two-biological-parent married families appear to enjoy the best outcomes, on average. Is this because their parents are married, or is it because they are residing with both biological parents? Marriage per se does not seem to confer advantages for children, as children in married stepfamilies fare worse than do those in two-biological-parent married families, instead appearing similar to those in single-mother families (Artis, 2007; Brown, 2004; Carlson & Corcoran, 2001; Manning & Lamb, 2003).

Similarly, biological parentage per se does not account for the advantages that children enjoy in two-biological-parent married families; children in two-biological-parent cohabiting families have worse outcomes, on average (Artis, 2007; Brown, 2004; Manning & Brown, 2006). This result obtained for kindergartners (Artis, 2007) as well as school-aged children and adolescents (Brown, 2004) across a host of domains of well-being, although for some dimensions, economic resources and parenting practices largely accounted for the differential. Children in two-biological-parent cohabiting families were more likely to be poor and to experience material hardship than were those in two-biological-parent married families, and the differential was most sizable for Whites (Manning & Brown, 2006). This pattern is consistent with the notion that two-biological-parent cohabiting families are a select group, as they are couples who did not formalize their relationships in response to pregnancy (Manning, 2004). Further evidence for selection is that child well-being is similar in two-biological-parent cohabiting and cohabiting stepfamilies (Artis, 2007; Brown, 2004; Manning & Brown, 2006).

Thus, both marital status and biological parentage are integral to children's well-being. Moreover, stability alone is not sufficient to maximize children's outcomes. Two-biological-parent cohabiting and married families are comparable on the basis of not only biological parentage but also stability, in that children in both family forms presumably have remained in them since birth and thus have experienced no disruptions or transitions in family structure. Granted, the risk of dissolution is greater in cohabiting than in married families (Manning, Smock, & Majumdar, 2004; Raley & Wildsmith, 2004), but both are intact family forms that comprise two biological parents. Future research should investigate why these two types of families differ for child development and whether marriage among cohabitors is linked to gains in well-being.

Typically, measures of family structure capture a child's relationship to resident biological parents or parent figures and ignore the presence of other family members. Many children residing with two biological married parents are actually part of a stepfamily in which half siblings are present. Shared children in blended families, as Halpern-Meekin and Tach (2008) termed them, comprise between 6% and 11% of all children living with two biological married parents (Gennetian, 2005; Ginther & Pollak, 2004). These children fared worse, on average, in terms of delinquency, depressive symptoms, academic achievement, and school detachment, than did their counterparts living in simple two-biological-parent married families (i.e., with no half siblings), even net of controls for family environment, family instability, and parental selection (Halpern-Meekin & Tach, 2008). Notably, these controls reduced to nonsignificance the differentials between adolescents in simple two-biological-parent married families and those in two types of stepfamilies (stepchildren in families with or without a shared child), which led Halpern-Meekin and Tach to conclude that current theorizing on how family structure operates on child outcomes is inadequate for explaining why teens in complex versus simple two-biological-parent married families differ across a range of outcomes. This attention to heterogeneity in two-parent families is a promising new direction, particularly as multiple partnerships and multiple partner fertility are increasingly common (Carlson & Furstenberg, 2006; Cherlin, 2009), but family structure measures have not kept pace with these developments.

Another understudied two parent family form is same-sex families. Mounting evidence indicates that children raised by lesbian parents fare as well as their counterparts raised by heterosexual married parents (for reviews, see Biblarz & Stacey, 2010; Stacey & Biblarz, 2001). Less is known about the outcomes of children raised by gay men. On several dimensions, lesbian couples are more effective parents than are opposite-sex couples, which reflects both selection factors and women's tendency to be more adept at and invested in parenting (Biblarz & Stacey, 2010). The political debate about same-sex marriage faces a curious intersection with the marriage promotion debate: if parental marriage is good for children, then why not allow same-sex parents the right to marry (Amato, 2004)? Marriage offers enforceable trust, status, and institutional support that will arguably stabilize same-sex relationships (Amato, 2004; Kurdek, 2004).

Conceptual and Measurement Issues

The heterogeneity of two-parent families illustrates existing gaps in the development of both measurement and theory. For instance, marital status (e.g., union type) and biological parentage are tied to child well-being. But future research must extend beyond the child and parents (or parent figures) to incorporate broader indicators of family membership, including sibling composition (Gennetian, 2005; Ginther & Pollak, 2004; Halpern-Meekin & Tach, 2008) and multigenerational families (DeLeire & Kalil, 2002; Foster & Kalil, 2007). And our family types must be more inclusive to encompass same-sex families (Biblarz & Stacey, 2010).

Research on measurement over the past decade has raised questions about how, when, and who we ask about family structure. For instance, the language used to describe emerging family forms, such as cohabitation, affected whether respondents identified themselves as cohabiting or single (Manning & Smock, 2005). Reports of family structure varied depending on whether they were measured prospectively or retrospectively (Teitler, Reichman, & Koball, 2006). And the more complex the family form, the less likely family members were to report the same family structure, which reflects the boundary ambiguity of arrangements such as married and cohabiting stepfamilies (Brown & Manning, 2009; Stewart, 2005). These findings, which have emerged from both qualitative and quantitative research, are of critical importance. They inform theory development and testing as well as the substantive conclusions drawn about the linkages between family structure and child outcomes. An important task for future data collection efforts is to accommodate these newly identified complexities by emphasizing longitudinal designs that incorporate multiple family members as well as more nuanced measures of family configuration.

New measurement approaches are stretching the current theories guiding research on family structure and child outcomes (Halpern-Meekin & Tach, 2008; Sun, 2003). Thus, as researchers refine how family structure is operationalized, additional theories will have to be developed or existing ones revised to inform nascent findings that do not fit the parameters of existing theory. In the future, theory and research should move beyond static comparisons of child outcomes across family structure to fuse together the concepts of family structure and stability.

Family Stability and Child Well-being

Family structure provides a snapshot of children's living arrangements but reveals little about family experiences over the course of childhood. As children are increasingly dispersed across a variety of family structures, some of which are more stable than others, it is important that researchers explicitly take account of the dynamics of family living arrangements across childhood. Family stability is as important for child well-being as family structure and has both immediate and long-term benefits for children (Cavanaugh & Huston, 2006, 2008; Cavanaugh, Schiller, & Riegle-Crumb, 2006; Heard, 2007b; Wolfinger, 2000). For instance, family stability during high school has been linked to young adult outcomes, including high school graduation, college enrollment, smoking and drinking, and sexual initiation (DeLeire & Kalil, 2002). Similarly, Wu and Thomson (2001) showed that family transitions, rather than prolonged exposure to a single mother or the extended absence of a biological father, were positively associated with early sexual initiation. Albrecht and Teachman's (2003) research confirmed this finding; the number of transitions was related to the risk of first premarital intercourse. For other domains of well-being, though (e.g., externalizing behaviors, cognitive outcomes), family structure was related to children's outcomes net of family instability (Fomby & Cherlin, 2007), which illustrates the utility of considering both family structure and stability.

Transitions: Types and Timing

Children and adolescents who experience parental divorce tend to exhibit poorer outcomes than their counterparts who remain in stable, two-biological-parent married families. This pattern of findings has been evident across multiple domains of well-being (Sun & Li, 2002; Videon, 2002) and in part reflects a decline in family income that usually accompanies divorce (Pong & Ju, 2000). However, children whose parents divorce typically exhibit lower levels of predisruption well-being, whether because of predisruption parental conflict or because of children's own problems contributing to marital instability (Sun, 2001; Sun & Li, 2001, 2002), although studies using fixed-effects models to examine intraindividual changes in child well-being prior to and following divorce have found no significant variation in cognitive ability, behavioral problems, or home environment (Aughinbaugh et al., 2005; Kowaleski-Jones & Dunifon, 2004).

If parental marriage is good for children, then it might be expected that remarriage is relatively benign in terms of child well-being. Even though remarriage often promotes family economic well-being, it nonetheless generates significant relationship and emotional stressors, as stepfamilies face considerable challenges to effective functioning, specifically, the renegotiation of family roles (Stewart, 2007). The negative effects on children's outcomes seemingly accumulate with each transition into or out of marriage (Cavanaugh & Huston, 2008; Fomby & Cherlin, 2007; Wu & Thomson, 2001), although research that explicitly tested for a nonlinear effect of transitions found children who experienced two or more transitions appeared to be distinct from those who experienced just one or none (Cavanaugh & Huston, 2006).

The impact of nonmarital transitions, or moving into and out of cohabiting families, has received comparatively little attention but is salient to the marriage promotion debate, as socioeconomically disadvantaged children are especially likely to reside in cohabiting families (Manning & Brown, 2006). More than 55% of Black and 30% of White children ever live in a cohabiting family, which means that an exclusive focus on marital transitions (i.e., divorce or remarriage) does not provide an accurate portrait of children's living arrangement dynamics (Raley & Wildsmith, 2004).

Transitions into and out of cohabitation are also notable because they appear to operate differently than marital transitions on child well-being. Early conclusions about the importance of family stability for child well-being and the negative implications of family transitions stemmed from research that ignored cohabitation transitions and focused exclusively on marital transitions. Once cohabitation transitions are considered, the conclusions about family stability and change become less clear. For instance, although stable married stepfamilies and stable single-mother families were negatively associated with children's misbehavior, the association was positive for stable cohabiting stepfamilies (Hao & Xie, 2002). Similarly, adolescents fared worse, on average, in stable cohabiting than in married stepfamilies (Brown, 2006). The more time spent in cohabiting stepfamilies, the greater was the likelihood of marijuana use (Cavanaugh, 2008). Transitioning out of a cohabiting family and into a single-mother family was related to gains in school engagement relative to remaining in a stable cohabiting family (Brown, 2006). Thus, although stable living arrangements tend to promote child well-being, not all stable family forms are beneficial for children. Moreover, instability has not always been associated with poorer outcomes. The negative outcomes faced by children who spend time in stable cohabiting families merit closer inspection both from a theoretical standpoint and for different age groups of children. The role of selection in cohabitation transitions also must be addressed.

In addition to the type of transition, the timing of family stability may be linked to child outcomes. Early childhood instability has the potential to have the most enduring influence on child development. Indeed, early family instability was related to social development at the end of elementary school (Cavanaugh & Huston, 2008) and to adolescent educational outcomes (Heard, 2007b). The apparent cumulative effects of family instability on child outcomes were largely a function of early childhood family instability (Cavanaugh & Huston, 2008). The salience of early childhood family instability for subsequent well-being deserves further investigation, with special attention to the age at which child well-being is ascertained. Few studies have considered family instability during middle childhood (for exceptions, see Fomby & Cherlin, 2007; Magnuson & Berger, 2009); most research has examined adolescents (e.g., Brown, 2006; Bulanda & Manning, 2008; Cavanaugh, 2008; Heard 2007a, 2007b), perhaps reflecting the relative paucity of longitudinal data on early childhood with sufficient sample sizes to allow for analyses of family transitions across diverse living arrangements.

Research on racial and ethnic variation in the effects of family instability on child well-being has shown patterns analogous to those documented for family structure and child well-being. Wu and Thomson's (2001) research revealed that the number of family structure transitions was related to White but not Black women's sexual debut. Among Black women, family structure as an adolescent was related to the timing of first sex. Heard (2007a) found that the associations between time spent with a single mother or nonparents on adolescent educational outcomes were smaller for Blacks and Hispanics than Whites, yet a family transition was linked to a greater decline in grade point average for Blacks. Racial differences in stress and social support tended to account for the race-specific effects of family instability on adolescent school performance, which illustrates how structural changes unfold in a cultural context (Heard, 2007a). Fomby and Cherlin (2007) documented a negative association between family transitions and children's cognitive outcomes among Whites that was largely a function of the mother's own characteristics. For Black children, there was no evidence that family transitions were related to behavioral and cognitive outcomes. How Hispanic children compare to Black and White children is largely unknown. One study indicated that neither family instability nor family structure was associated with Mexican American children's behavioral and cognitive outcomes, but this finding should be replicated with other data and for additional Hispanic groups (Fomby & Cherlin, 2007).

Conceptual and Measurement Issues

Prior research has employed varied conceptualizations of family instability, and consequently, there is room for further development and refinement to identify the salient aspects of instability (cf. Magnuson & Berger, 2009). Some studies examined the number of family transitions, under the assumption that each additional transition creates stress that is detrimental to child well-being (e.g., Bulanda & Manning, 2008; Cavanaugh, 2008; Cavanaugh & Huston, 2008; Hao & Xie, 2002). Others focused on the duration of exposure to various family types (e.g., Bulanda & Manning, 2008; Cavanaugh, 2008; Dunifon & Kowaleski-Jones, 2002; Magnuson & Berger, 2009). This approach implies that the more time (or greater share of childhood) spent outside of two-biological-parent married families, the worse child outcomes are likely to be. Still other research investigated whether a child ever resided in a particular family form and ignored the duration of time spent in the family type (e.g., Bulanda & Manning, 2008; Cavanaugh et al., 2006). Finally, some research has examined specific types of family transitions (e.g., Brown, 2006; Magnuson & Berger, 2009). Several studies included two or more of these indicators of family instability (e.g., Bulanda & Manning, 2008; Cavanaugh, 2008; Heard, 2007b). And, research on family stability sometimes included controls for current family structure (Fomby & Cherlin, 2007; Wu & Thomson, 2001). These disparate approaches make it difficult to assess precisely what aspects of family instability may impede child development. Ideally, family research should move toward developing consensus about how family stability should be conceptualized and measured.

New theoretical developments concerning both when and why family instability is consequential for child well-being could facilitate this difficult task. Early childhood instability has long-term effects, but the mechanisms underlying this pattern are unclear (Cavanaugh & Huston, 2008; Heard, 2007b). How family instability undermines economic resources or parental socialization has received relatively little research attention (but see Thomson, Mosley, Hanson, & McLanahan, 2001). Prospective, longitudinal data can be used to estimate growth curve models that can elucidate how changes in family structure are linked to changes in family processes.

Marriage and Child Well-Being Among At-Risk Populations

Until recently, much of the research on family structure and child well-being has emphasized comparisons between children in two-biological-parent married families and children whose parents had either divorced or divorced and remarried (Demo & Cox, 2000). This approach excludes a growing share of children: those born to unmarried mothers who are unlikely to ever spend time in a two-biological-parent married family or, indeed, any married family. Children born to unmarried parents are a core part of the group that recent federal marriage initiatives target. But relying on the nationally representative samples of the U.S. population of children typically used in research on family structure and child well-being may limit the ability to identify unique patterns or associations for particular subgroups, such as children born to unmarried parents or who are low income (Huston & Melz, 2004). Certainly, the research showing weaker effects of family structure and stability for Black versus White children portends differentials for other groups in which marriage is less common (e.g., children born to unmarried parents and low-income families), and some small-scale (nonrepresentative) studies that have focused specifically on economically disadvantaged children have confirmed this (Ackerman, Brown, D'Eramo, & Izard, 2002; Ackerman, D'Eramo, Umylny, Schultz, & Izard, 2001; Foster & Kalil, 2007).

A recent longitudinal survey, the Fragile Families and Child Well-Being Study, provided important new data on children and parents in urban families, which tend to be disadvantaged economically and represent a large share of children born to unmarried mothers. The study follows nearly 5,000 children born from 1998 to 2000 in 20 cities with population sizes of more than 200,000. Roughly three quarters were born to unmarried parents. Mothers were interviewed in hospitals soon after giving birth. Fathers also were interviewed around the time of the birth. The study included reinterviews of mothers and fathers when children were 1, 3, and 5 years old. Fragile Families also has a qualitative component, the Time, Love, Couples, Cash, and Caring (TLC3) study, which comprises multiple in-depth interviews over time with a subset of Fragile Families respondents (England & Edin, 2007). These data, which are publicly available, have yielded a tremendous amount of scholarly research that not only complements the body of literature generated from nationally representative data sources but also directly informs contemporary debates about the role of government in marriage.

Family Structure, Stability, and Child Outcomes

Consistent with research on the overall population of U.S. children, several Fragile Families studies have found that urban children's outcomes tend to be better in married than in unmarried families. For instance, parental marriage was positively associated with children's health outcomes (Gibson-Davis & Brooks-Gunn, 2007; Guzzo & Lee, 2008; Harknett, 2009). Among children born outside of marriage, those born into two-biological-parent cohabiting families exhibited better health and behavioral outcomes than did those born to non-coresidential parents (Heiland & Liu, 2006). Whether this differential persists as children age will be important to investigate, because research on school-aged children and adolescents in the general population has shown few differences in well-being among those residing outside of two-biological-parent married families (Artis, 2007; Brown, 2004; Manning & Lamb, 2003).

Family stability is linked to family structure at birth among at-risk children in ways that are similar to those observed for all children (Manning, Smock et al., 2004; Osborne, Manning, & Smock, 2007). Osborne et al. (2007) found that urban children experienced significantly greater family stability during the first 3 years of life if their parents were married at the time of birth. Cohabitation did not provide a comparable level of stability; parental separation was 5 times greater for children born to cohabiting than married parents. This differential was even greater among White children (which is consistent with Manning, Smock et al.'s [2004] findings for all children) but reflected sociodemographic variation between White married and cohabiting parents. Among Blacks and Hispanics, the differential persisted net of controls (Osborne & McLanahan, 2007).

Children born to unmarried parents are unlikely to experience parental marriage, regardless of the couple's relationship status (Carlson, McLanahan, & England, 2004), a pattern in line with prior research on all U.S. births (Manning, 2004; Manning, Smock et al., 2004). Those presumably in the most committed relationships, cohabiting parents, typically remained in unmarried coresidential partnerships (60%). Only about 15% married (Carlson et al., 2004; Osborne, 2005). About 5% of visiting, or non-coresidential but romantically involved, parents tied the knot, but they were likely to move in together, with estimates ranging from 21% (Osborne) to 32% (Carlson et al., 2004). Few parents married who were either friends (i.e., not romantically involved) or had no relationship (less than 2%), but 9% and 6%, respectively, began cohabiting (Carlson et al., 2004). The propensity to cohabit rather than marry following conception and birth of a child reflects the shift from shotgun marriages to shotgun cohabitations, evidence that cohabitation is increasingly operating as a substitute for marriage (Raley, 2001).

The unstable living arrangements that many children born outside of marriage experience raise questions about the consequences of this instability for well-being. Perhaps it is not surprising that the number of family transitions among young urban children was negatively associated with child well-being, regardless of family structure at birth (Osborne & McLanahan, 2007). Increased maternal stress and poorer mothering over time appeared to mediate this association. Notably, changes in income or residence were not related to children's behavioral problems, which is contrary to conclusions drawn from studies of the general U.S. population of children. These findings led Osborne and McLanahan (2007) to conclude that marriage programs designed to promote family stability may be beneficial for children, provided that they do not encourage poor-quality relationships that could increase maternal stress and lead to poor parenting. Unmarried parents are not a monolithic group.

Indeed, it does not appear that low-income children born to unmarried parents benefit substantially from parental marriage, at least in the short term. During the first year of life, children whose cohabiting parents married fared no better than those who remained cohabiting (Heiland & Liu, 2006), which is in line with research on a national sample of adolescents that showed no benefits of union formalization for child well-being (Brown, 2006). This result suggests not that marriage per se is not driving the benefits children enjoy, but that selection factors are pivotal. Otherwise, we might have expected to see gains in child well-being following parental marriage. This important result, also found in unpublished analyses by others using Fragile Families data (Osborne, McLanahan, & Brooks-Gunn, 2004; Osborne & Palmo, 2009) needs to be replicated using other data sources.

In addition to providing new information on how family structure and stability are related to child well-being among at-risk populations, Fragile Families research has expanded current conceptualizations of family structure and stability. By moving beyond coresidence to examine the relationship between the child's biological parents, researchers have illustrated the importance of distinguishing among various types of non-coresidential relationships: visiting parents, parents who are not romantically involved but friendly, and those with no relationship (Carlson et al., 2004). This research also has pushed the boundaries of family transitions beyond residential union type (i.e., cohabitation versus marriage) to include dating relationships among unmarried mothers. Osborne and McLanahan (2007) demonstrated that transitions into and out of romantic, non-coresidential relationships, which they termed visiting relationships, were linked to family structure trajectories experienced during early childhood. For example, 10%, 20%, and 30% of children in cohabiting, visiting, and single-mother families, respectively, experienced 3 or more relationship transitions (for children born to married parents, it was about 2%) between birth and age 3. Although non-coresidential relationships may have weaker effects on child well-being, they can undermine maternal well-being, thus resulting in inconsistent parenting. Accounting for visiting relationship transitions may be especially important among Black mothers, who were unlikely to form either married or cohabiting unions (Osborne & McLanahan, 2007). Still, the significance of multiple partnerships—whether or not coresidential—for child well-being is not just an issue for at-risk populations but is also relevant for the larger U.S. population, which is quick to form and then often dissolve intimate partnerships (Cherlin, 2009).

Marriage Among Unmarried Parents

Across the socioeconomic spectrum, marriage is highly revered (Edin & Reed, 2005). Yet paradoxically, these lofty aspirations and the attendant high expectations about the prerequisites for marriage ultimately deter many low-income couples from marrying, including those who have a child together. Most low-income couples want to marry, but few achieve their goal (Lichter, Batson, & Brown, 2004; Waller & McLanahan, 2005). Marriage is now a status symbol that many low-income couples lack the resources to achieve (Edin & Reed, 2005). Central to the effectiveness of initiatives designed to encourage healthy marriages and relationships is a clear understanding of the barriers to marriage that at-risk populations face.

Several studies have elucidated the hurdles low-income couples believe they must surmount to be ready for marriage (Gibson-Davis, Edin, & McLanahan, 2005; Smock, Manning, & Porter, 2005). Many of those hurdles may not be unique to low-income couples, but recent research on barriers to marriage has focused exclusively on this particular population, perhaps because relatively low levels of marriage characterize it.

A central prerequisite for marriage is financial. Couples pointed to the importance of financial stability, responsibility, accumulation of assets (e.g., a house, a car), and savings to pay for a wedding celebration (Gibson-Davis et al., 2005). Couples preferred to delay marriage rather than get married at the courthouse (Smock et al., 2005). Partners were reluctant to assume each other's debts and wanted to be sure they could pay bills and support a certain lifestyle. Ultimately, for low-income couples, marriage was a signal to family and friends that they had “‘arrived’ in a financial sense” (Gibson-Davis et al., 2005, p. 1308).

Another prerequisite for marriage is an expectation that the relationship will endure. Couples must resolve doubts about their partner or ongoing sources of relationship conflict before tying the knot. Some believed that married couples must always get along and never experience disagreements (Gibson-Davis et al., 2005). This is a very high bar, particularly among a population in which relationship stressors such as unemployment, criminal activity or time spent in prison, alcohol or drug abuse, violence, and children from prior relationships are common (Edin & Reed, 2005). Indeed, many unwed couples faced substantial gender mistrust in their relationships, which had been marred by infidelity. More generalized gender mistrust was not a deterrent to marriage (Burton, Cherlin, Winn, Estacion, & Holder-Taylor, 2009).

Some cohabitors maintained that marriage was an option only when the male partner could fulfill the traditional role of the breadwinner (Smock et al., 2005). Men's unemployment (or underemployment) is a key barrier to marriage (Edin & Kefalas, 2005). A shortage of employed Black men explained much of the lower marriage rate among Black versus other mothers in the Fragile Families data (Harknett & McLanahan, 2004).

Relationship stability is integral to marriage readiness because of paramount importance is avoidance of divorce. For many low-income couples, it would be more shameful to marry and then divorce as opposed to having never married at all. Most expressed highly unfavorable attitudes toward divorce and would not marry unless they were certain they would not get divorced (Edin & Kefalas, 2005; Gibson-Davis et al., 2005; Waller & Peters, 2008). Others however (Cherlin, Cross-Barnet, Burton, & Garrett-Peters, 2008), found little support for the idea that low-income women are reluctant to marry because of a fear of divorce. Perhaps low-income women are accepting of divorce in the abstract but less willing to agree that divorce is a viable option for them.

Modeling Marriage: Theoretical and Empirical Advances

With increasing proportions of children born to never-married mothers and the attendant family instability such children experience, new approaches to thinking about how low-income women make decisions about marriage and childbearing are essential for the field to move forward. Gibson-Davis's (2009) family expectations and family formation theory is a notable start. Arguing that traditional models of marriage formation are not applicable to women who already have a child, Gibson-Davis proposed an explanation for why many low-income parents report substantial economic barriers to marriage but not childbearing: marriage signifies the achievement of economic and relationship stability, whereas childbearing is merely a normative feature of the life course. Consequently, a rise in income should increase the likelihood of marriage but be unrelated to childbearing. Her empirical results supported those hypotheses. Both changes in couple earnings (driven mostly by paternal earnings) and the income-to-needs ratio were related to marriage in the expected direction. Similarly, the purchase of a home was positively related to the likelihood of marriage (and the loss of a home was negatively associated with marriage). This theory elucidates potential policy levers to increase marriage among low-income persons, namely to facilitate financial and residential stability to overcome the substantial barrier of economic disadvantage. It also implies that low-income couples view financial and relationship stability as antecedents to marriage rather than as consequences.

If economic stability is a prerequisite for marriage, then it stands to reason that low-income mothers who marry experience minimal economic benefits. Empirical research has tended to support this conclusion (Lichter, Qian, & Mellott, 2006). Low-income women who had a child outside of marriage—the target group for recent marriage initiatives—are highly unlikely to marry (Lichter, Graefe, & Brown, 2003). Those who married fared better economically than did their unmarried counterparts, but only if they stay married. Unwed mothers who married and then divorced were ultimately worse off economically than were those who never married (Lichter, Qian et al., 2006), which lends credence to low-income women's adamancy against marriage unless they are certain they will not get divorced. Marriage appears to minimize poverty and welfare dependence among disadvantaged women, but it does not necessarily provide economic stability (Lichter, Qian et al., 2006).

Conclusions

Research on family structure and child outcomes has taken on a new urgency with the ongoing federal initiatives to encourage healthy marriages, especially among the disadvantaged. Several large, federally funded demonstration projects are under way to evaluate whether education programs can facilitate healthy relationships and ultimately, marriage, among populations in which marriage rates are low and nonmarital childbearing high (Dion, 2005). Efforts to encourage marriage are grounded in family scholarship on children's living arrangements and their relationship to well-being. But the mechanisms linking marriage, poverty, and child outcomes are still not entirely understood. Indeed, scholars and policy makers alike continue to debate the role of marriage in ameliorating poverty and enhancing child well-being (Amato & Maynard, 2007; McLanahan et al., 2005). Research conducted over the past decade not only has shed additional light on this subject but also has pointed to directions for future research that are likely to inform family policy in new and meaningful ways.

Family scientists have expanded the conventional approaches to conceptualizing and measuring family structure in response to the increases in cohabitation and unmarried childbearing coupled with the rise in complex stepfamilies (cf. Bumpass & Lu, 2000). These changes bring to the fore the heterogeneity among two-parent families, which itself offers leverage on what about marriage is uniquely beneficial for children, thereby informing ongoing efforts to disentangle selection versus causation effects (Amato, 2005; McLanahan & Percheski, 2008). Research has demonstrated that neither marriage nor residing with two biological parents is in itself a sufficient condition, as children in married stepfamilies, complex two-biological-parent married families, and two-biological-parent cohabiting families fare worse, on average, than do their counterparts in simple two-biological-parent married families (Artis, 2007; Brown, 2004; Halpern-Meekin & Tach, 2008). The task for future research is to develop more nuanced theory and richer data to decipher the mechanisms driving these differentials.

The growing emphasis on the dynamics of family experiences has revealed that the types and timing of family transitions are linked to child outcomes. Cohabitation and marital transitions have unique effects on children; stability in cohabiting families is not necessarily beneficial for children (Brown, 2006; Cavanaugh, 2008). Family instability during early childhood has enduring consequences (Cavanaugh & Huston, 2008; Heard, 2007b). Future research should adjudicate among the myriad ways family instability has been operationalized— types, timing, and duration, to name a few—to pinpoint why and how instability is consequential for child well-being. Also, greater attention should be paid to explaining variation in family stability and child outcomes among subgroups, including by race-ethnicity and marital status at birth, as evidence continues to mount that such relationships can be distinctive across groups (Fomby & Cherlin, 2007; Heard, 2007b; Manning, Smock et al., 2004).

Indeed, the Fragile Families Study represents a significant contribution to the field not only because it focuses on these understudied subgroups but also because it allows researchers to examine family dynamics among the population that federal marriage initiatives target. This line of inquiry has shown that, although the outcomes of children born to married parents are, on average, better than those for unmarried parents (Gibson-Davis & Brooks-Gunn, 2007; Harknett 2009), marriage following the birth of a child apparently has few immediate benefits for children (Heiland & Liu, 2006). Children born to unmarried parents face considerable family instability during the first years of life, which is linked to well-being (Osborne & McLanahan, 2007). Notably, the family processes undergirding these linkages are not primarily economic but rather relationship based, pointing to an avenue for intervention through relationship skills training that could buffer the potentially negative outcomes associated with multiple partnerships (Osborne & McLanahan, 2007).

Overall, perhaps the most consistent and compelling message to emerge from scholarship over the past decade is that encouraging marriage per se is not enough. This is evident in research on the general U.S. population of children as well as on disadvantaged subgroups. Research on children born to unmarried parents (Heiland & Liu, 2006) revealed no appreciable gains in child well-being following parental marriage shortly after birth. Similarly, cognitive and behavioral outcomes did not vary by family structure among economically disadvantaged preschoolers, regardless of race-ethnicity (Foster & Kalil, 2007). These findings suggest that encouraging marriage among at-risk populations may not translate into improved child outcomes, although firm conclusions necessitate replication with other data sources and for children of varying ages. Researchers and policy makers need to learn more about the enduring consequences of growing up in at-risk families. And greater attention to the mechanisms or family processes that mediate much of the relationship between family structure and child well-being, particularly for at-risk subgroups, may contribute to the development of family policies and programs that enable parents to provide family environments in which children thrive, regardless of family structure (Foster & Kalil, 2007). Since 2000, family scholars have made tremendous strides toward achieving the goal of deciphering the conditions under which family experiences maximize children's well-being. This momentum promises to propel future research in new directions that bring us even closer to this goal.