Reflections on “Family Structure and Child Well-Being: Economic Resources vs. Parental Socialization”

Our Paper

In Thomson, Hanson and McLanahan (1994), we investigated the relative importance of two types of parental resources – time and money – for explaining the association between family structure and children's academic and socioemotional development. Family structure was classified as married-parent, stepparent, cohabiting parent, divorced-mother and never-married mother families. The primary innovations of the study were as follows: (1. to use nationally representative data – the U.S. National Survey of Families and Households (NSFH) – to examine the association between family structure and child well-being, and (2. to include a more diverse set of family structures than had been examined in the past, including cohabiting stepparent families and families headed by never-married mothers. The NSFH had over-sampled single-parent families, cohabiting couples, and stepfamilies, providing for the first time sufficient numbers of observations for these less common but theoretically meaningful family types.

Briefly, our findings showed that cohabiting parents were closer to single parent families than married parent families in terms of economic disadvantage. We also found that differences in economic resources accounted for much more of the disadvantage associated with non-traditional family structures than differences in parenting, especially differences between single parent families and married parent families. Parenting differences, as measured in our data, accounted for only a small part of the differences associated with family structure.

Our article was part of an ongoing stream of research to understand and explain the well-established association between family structure and child well-being. The fact that much subsequent research confirmed many of our findings is due to the substantial body of theory and empirical research upon which our study was built. As members of the design team for the National Survey of Families and Households, we had worked to ensure that all of the elements for the study were in place and did our best to ensure that gaps in prior data were filled. We view this study, however, as one in a series of incremental steps toward a fuller understanding of family transitions and child well-being. It simplified but clearly articulated what had been viewed as the two primary mechanisms that might account for differences among families – time and money. The simplification may be part of the article's popularity as a referent point for subsequent research that has gone well beyond in both theory and methods. The distinctions we were able to make among cohabiting and married stepfamilies and between never-married and ever-married mothers also provided stepping stones to more extensive and nuanced research.

Research on family structure has expanded dramatically during the past two decades, in part because of the proliferation of different family forms and in part because of the availability of new longitudinal studies that follow families and children over time. The new body of research presents a much more detailed and more complicated picture of the types of families in which children grow up and the family conditions and processes that are associated with healthy child development. In the discussion below, we describe some of the most important ideas and concerns that have arisen and how they extend our original findings.

The Importance of Cohabitation

When the paper was written, researchers were focused on single parenthood created by divorce and stepparent families created through remarriage. Since then, divorce rates have leveled off and the flow into single motherhood and new partnerships is driven more by nonmarital births, an increasing proportion of which are to cohabitating parents (Bumpass and Lu 2000; Graefe and Lichter 1999). Accordingly, more recent surveys, especially those that follow parents and children over time, allow researchers to identify not only cohabiting stepfamilies but also to distinguish cohabiting biological parents from married biological parents as well as from cohabiting or married stepfamilies.

Most subsequent research confirms our finding that children living with their mother and her cohabiting partner have the poorest outcomes, or are more similar to children living with single mothers than to children living with a married stepparent (Sweeney 2010). As we found, these differences are not completely explained by the poorer economic circumstances or parental engagement in cohabiting stepfamilies compared with married stepfamilies.

The new research has shown that cohabiting biological parents are in many ways more similar to cohabiting stepfamilies than to married biological parents. Both types of families are economically disadvantaged in comparison to married biological parents or stepfamilies (Jonsson and Gähler 1997; Manning and Lichter 1996). Cohabiting biological parents may also provide lower quality parenting and home environments than married biological parents (Aronson and Huston 2004; Klausli and Owen 2009). And cohabiting biological parents are more likely to separate than married parents (Andersson 2002; Heuveline et al. 2003). Perhaps due to the relatively poorer economic and parenting experiences in cohabiting families, children appear to suffer less from their cohabiting parents' separation than from their married parents' divorce (Jonsson and Gähler 1997; Wu et al. 2008, 2010).

The Importance of Instability

Although economic resources and parental engagement remain important explanations for associations between family structure and child well-being (e.g., Hofferth 2006; Marks 2006; Ram and Hou 2003), new research has demonstrated the importance of family instability per se. Early research on local and purposive samples had shown that most families adjusted to divorce after a year and half (Hetherington, 1992). In our study, we examined but found no differences in children's outcomes by duration in different types of family structures. We did not, however, consider the number of total transitions the child experienced prior to our observation.

Our colleague Larry Wu had documented extensive variation in childhood histories collected for the first time in the National Survey of Families and Households (Wu and Martinson 1993). Subsequent research has demonstrated that – beyond the particular types of transitions children experienced – the cumulative number of changes may be independently and negatively associated with outcomes during childhood and young adulthood (see review in Sweeney 2010). Researchers have also shown that instability in family structure is associated with lower quality parenting (Beck et al. 2010) and that certain types of instability are less harmful than others. Osborne et al (2012), for example, show that moving in with the biological father after a nonmarital birth is positively associated with child well-being. Brown (2006) reported improvements in school engagement after the mother's separation from a cohabiting stepfather. And children living with single mothers do not appear to gain from the stability of that family form (Waldfogel et al., 2010).

Because family structure and instability are intertwined, researchers must be careful to take both into account when reporting the “total effect” of a particular family structure. For example, a divorced or separated mother, by definition, has experienced at least one transition, while a remarried mother is likely to have experienced two transitions. Thus the total effect of remarriage on child well-being is the sum of the effect of the transitions plus the effect of the mother's current status.

The Importance of Complexity

The idea of family complexity is not new. Well before our study, the complexity and ambiguity of stepfamily life was recognized as a source of difficulty for children, parents and stepparents alike (Cherlin 1978). Indeed, in our paper we argued that complexity and ambiguity might lead to lower parental investments in stepfamilies. More recently, research on families formed by unmarried parents has found much higher levels of complexity than were typically seen in stepfamilies in the past (Carlson and Furstenberg 2006). Indeed, recent estimates suggest that nearly half of children born to unmarried parents will live with a half sibling by the time they are 5 years of age and a quarter will live with half siblings sired by two or more fathers (Tach et al. 2011).

Although having a child in a new partnership was thought to strengthen relationships in stepfamilies, it appears that the added complexity is more harmful than any relationship gains. Several recent studies show negative outcomes for children with half siblings or stepsiblings, even if they live with their two biological parents (see Sweeney 2010 review). Tillman (2008) reported that the presence of half siblings or stepsiblings accounted for a great deal of the poorer outcomes among children living with a parent and stepparent. Similarly, studies focusing on nonmarital unions find that family complexity is associated with less parental cooperation, lower levels of father involvement (time and money) and higher levels of child behavior problems (Tach 2011).

The Importance of Context – Welfare Regimes

At the time we wrote the article, research on family structure and child well-being was almost exclusively based on the United States, with a few studies from other English-speaking, market-oriented welfare states (Chapple 2009). In all of these countries, many lone mothers had never been married, economic inequality was relatively high, and single parenthood often meant poverty or a marginal economic position. We attributed the lack of research in other welfare regimes, especially the Nordic countries, to the fact that single mothers received greater transfers and were better off than single mothers in the United States (Casper et al. 1994). With child allowances and highly subsidized childcare and housing assistance, single mothers in these other countries had greater opportunities for employment and could provide higher quality environments for their children. These differences would not, of course, apply to difficulties of children in stepfamilies.

To the contrary, recent research has demonstrated quite convincingly that family structure is strongly associated with children's well-being, even in the very generous Nordic welfare states (Björklund et al. 2007; Breivik and Olweus 2006; Jonsson and Gähler 1997; Steele et al. 2009). In a 24-country study, Låftman (2010) found children living with single mothers had fewer material resources, less parental support, and poorer health than those living with two original parents. More important, the degree of differences was not consistently related to the proportion of children living with single mothers, the welfare state regime, or single mothers' employment rates. Other cross-national analyses have also failed to find a systematic relationship between a country's welfare provisions and the relationship between family structure and child well-being (Chapple 2009; Marks 2006).

One might also expect cohabitation to be less selective and have weaker effects on children in contexts such as the Nordic countries where births to cohabiting couples were common even when our article was published and the legal status of cohabitation differs little from that of marriage when couples have shared children. While the relative risk of separation for cohabiting and married parents is lower in these countries than in other contexts, children born to cohabiting parents are more likely to experience disruption than those born to married parents (Andersson 2002; Heuveline et al. 2003). Furthermore, differences in achievement between children living with cohabiting or married parents are also found in the Nordic context, both for biological and stepfamilies (Jonsson and Gähler 1997).

Although studies outside the United States have all incorporated the ‘money’ part of our analysis, most have not had or have not used data on parental practices or the child's cumulative experience of family change. An exception is research from the U.K. Millennium Cohort Study that collected data on both instability and complexity. Analyses of these data show that although overall levels of instability and complexity are lower in the United Kingdom than in the United States, the family structure gradient is similar. Mothers who are married at birth have the lowest levels of instability and complexity, single mothers have the highest levels and cohabiting mothers fall in between (Kiernan et al 2011).

The Importance of Selection

Since we wrote the article, we have gained a much greater appreciation of how difficult it is to identify causality. The basic problem is that people choose their family structures and thus the factors that affect their choices may also affect parental resources as well as child outcomes. In the absence of an experiment, obviously unethical, researchers cannot be sure that the association between family structure and child well-being is causal rather than spurious.

Increasingly high-quality longitudinal data have made the task of inferring causal relationships a little easier by allowing researchers to control for a much larger set of potential confounders, including child well-being prior to a disruption. Unfortunately, the latter is not always appropriate because the predivorce measure may be picking up the effect of a pending divorce. Research has also applied a variety of econometric and statistical methods to adjust for confounders, including instrumental variables, family fixed effects, sibling fixed effects and propensity score matching. Each method removes potential unobserved sources of the association between family structure and child well-being, but each is also based on assumptions that are often difficult to support (for reviews see Sigle-Rushton and McLanahan 2004; Chapple 2009).

For example, Gruber (2004) used divorce laws as an instrument for divorce, under the assumption that laws would be independent of couple characteristics underlying divorce. The fact that liberal laws are associated with poor child outcomes was viewed as evidence of a causal effect of divorce. As the author notes, however, states that pass liberal divorce laws are likely to differ from other states in many ways that might affect child well-being. Studies that control for family differences by observing siblings in the same family may find no differences between siblings who experienced parental separation or divorce before a particular outcome is measured and those who experienced the separation or divorce at a later point in time (e.g., Björklund et al. 2007). Unfortunately, such results are not generalizable to families in which all children did or did not experience their parents' separation. And because most siblings are relatively close in age, the underlying conflicts that precipitate separation or divorce could be the true source of difficulty for both siblings who did and siblings who did not experience the event before the outcome is measured.

Overall, studies that attempt to control for selection generally find that the causal effect of family structure is smaller than the estimate typically obtained from the observed differences, and in some cases disappears. On balance, ­however, the consensus is that family structure has a small causal effect on the well-being of parents and children.

Future Directions

Several avenues for future research look promising. First, the work on instability is in its early stages, and much more research is needed to understand which types of instability are most important and under what circumstances they matter most. Some studies, for example, have found that instability in noncoresident unions is just as stressful for mothers as instability in coresident unions (Cooper et al. 2009). Other studies have found that the effect of instability on mothers' parenting may differ depending on mothers' education and the type of parenting examined (Beck et al. 2010). In order for researchers to fully investigate these issues, our surveys will need to include more detailed information on partnership changes, including changes in noncohabiting romantic relationships.

Similarly, recent work on family complexity suggests that this topic is likely to yield important insights into family functioning. As noted above, complexity appears to be associated with both family income (receipt of child support) and parenting practices (parental cooperation). There is also evidence from qualitative work that fathers' involvement with a child from a previous relationship leads to jealousy and mistrust in his current partnership (Monte 2007). These issues need much more attention if we are to understand not only the effect of complexity on child well-being but also the feedback effect on union stability.

In addition to the work on instability and complexity, we expect to see much more research on cross-national comparisons of the links between family structure and child well-being. In the past decade several new birth cohort studies have become available that will permit researchers to examine a variety of family structure and the effects of early exposure to family instability. Included here are birth cohort studies in Australia, the United Kingdom, France and Ireland. Longitudinal data registers in the Nordic countries also afford possibilities for studying family structure and family change across generations and cohorts.

Finally, we believe there is much to be gained from paying more attention to heterogeneity in the effects of family structure on child well-being. We have long known that while the average effect of divorce is negative, for some families it may actually improve family functioning and child well-being. Work by Amato (1993), for example, shows that in families with high levels of conflict, divorce improves child outcomes. More recently, Jaffee et al. (2003) have found that children are better off not seeing their fathers in cases where these men are violent or antisocial.

Much can also be gained from focusing on differences in children themselves. Research has been mixed with respect to differential effects of different family forms for boys and girls. Recent evidence suggests that boys may be more sensitive to family instability than girls, especially when the outcome examined is externalizing behavior (Cooper et al. 2011). Although gender differences are not found in all studies, the issue is important and may help explain the growing gender gap in school performance that has been documented in the United States and other countries (McDaniel 2010). We also are intrigued by recent work suggesting some children are more ‘reactive’ to their environment than others because of their genetic makeup (Boyce and Ellis 2005). If the “orchid-dandelion hypothesis” is correct, then our previous estimates of the effects of family structure are quite misleading, underestimating the effect of family instability for one group of children while overestimating the effect for others.

Conclusion

We are, of course, glad that so many scholars have found our article a useful reference and stepping stone for their own work. It is humbling to realize where we were in 1994, how far the field has come in the past two decades and what the future holds. The fact that single-, step- and cohabiting-parent families continue to grow and the fact that they are associated with poorer outcomes for parents and children means that the implications of today's and tomorrow's research on these topics are enormous for both individuals and the societies in which they live.

Notes

References