Family Instability in the Lives of American Children

Abstract

Scholars have long looked to family composition to understand child well-being. Family instability, or the experience of repeated changes in parents’ union status during childhood, represents a recent advance in this field that takes into account the dynamic nature of contemporary family organization and considers its implications for children’s adjustment and development. We review some of the structural and cultural factors that have contributed to rising levels of family instability and highlight the emergence of national data to measure it. We then review the perspective that guides much of the scholarship on family instability and critically assess the contributions of this work to the understanding of child well-being. We close by suggesting new directions for research, with a call for work that broadens the conceptualization and measurement of contemporary children’s family systems and home environments as well as the mechanisms that explain why—or whether—instability matters.

INTRODUCTION

One of the basic functions of families is to develop and promote the well-being of children (Parson 1949). In sociological research, family structure, or the presence and union status of children’s parents, has long stood as a marker of families’ capacity to achieve these goals through the provision of time, money, and emotional support. As divorce rates increased across the 20^th century, especially during the 1970s and 1980s (Martin & Bumpass 1989; Ruggles 1997), scholarship on divorce and its implications for child well-being increased significantly (Cherlin 1981; Furstenberg et al., 1983; McLanahan & Bumpass 1988). Collectively, this work highlighted the negative implications of divorce for children born in married parent families and helped establish the sociological study of the changing American family.

Building on these insights, Wu and Martinson (1993) introduced the concept of family instability. The authors observed that across childhood, a substantial fraction of youth experienced not just a single change in family structure like a divorce, but multiple changes in parents’ martial or romantic relationship status. Wu and Martinson posited that this family instability, along with the cascading set of disruptions that repeated family structure change may introduce into children’s lives, was a distinctive feature of contemporary family experience. Moreover, they hypothesized that family instability would contribute to population variation in child development and to the intergenerational transmission of (dis)advantage net of the family structure in which a child lived at any given point in time. Their proposition drew upon family stress theory and emerging empirical evidence on the increasing fluidity of families with children, setting family instability, defined as the experience of multiple changes in parents’ union status, apart from prior research on family change, which had focused on single events such as divorce or remarriage.

The concept of family instability emerged against a backdrop of significant economic and cultural shifts in American life that unfolded over the preceding fifty years and reshaped patterns of family composition. A rapidly expanding pool of quantitative data captured the complexities of children’s family environments through retrospective reports and prospective research designs focused on parents and young people themselves. These factors established family instability as a point of social science inquiry. Since the early 1990s, scholarship on family instability expanded to include more than 130 articles on the topic published in leading family, demography, sociology, and child development journals. Further, this approach has broadened how scholars think about the contexts in which children develop, informing research on other domains of family life such as family complexity (Cancian, Meyer, & Cook 2011) including multipartner fertility (Carlson & Furstenberg 2006), relationship churning (Halpern-Meekin & Turney 2016), sexual minority families (Moore & Stambolis-Ruhstorfer 2013), household instability (Raley et al., 2018), and income volatility (Hill et al. 2013).

The goals of this review article are 1) to highlight family instability as a distinctive component of contemporary family life, and 2) to illustrate how theory, data, and methods to assess the family as a dynamic and socially embedded institution have advanced sociological insight into the role of family in shaping child well-being. In the interest of space, this review does not consider the implications of family instability for mothers’ and fathers’ well-being beyond their roles as parents and is limited to research about families in the United States. We draw heavily on quantitative research, reflecting the methodological orientation of family demographers and developmental psychologists who have conducted most of the research in this area. We begin by identifying the contextual factors that have contributed to the organization of contemporary family life and to the rise of family instability. We then present the instability and change hypothesis and the theories that guide it. Next, we describe data advances that measure family structure dynamically. We then review empirical research on the link between family instability and child well-being across the early life course, highlighting conceptual and methodological innovations. We close by suggesting new directions for this research.

Except where noted, we draw upon literature identified through a systematic review of published research using the following criteria: (1) publication date between January 1993 and June 2018; (2) publication in one of the following peer-reviewed general interest or field journals relevant to scholarship on sociology of the family or child development: American Journal of Sociology, American Sociological Review, Child Development, Demography, Developmental Psychology, Journal of Marriage and Family; or Social Forces; and (3) matched on one or more of the following search terms: partner instability, family instability, family transitions, family structure instability, family structure transitions, family structure change, relationship instability, marital instability. We filtered the resulting list to exclude articles that focused only on a single union status transition such as divorce, cohabitation dissolution, or remarriage or that considered outcomes pertaining to parents, rather than children. The resulting index of 57 published work appears as Online Appendix 1 at [URL].

MACRO LEVEL CHANGES

Well-documented demographic shifts in union formation behaviors (e.g., marriage, divorce, cohabitation, remarriage) and fertility behaviors (e.g., nonmarital fertility) have translated into rapid change in the composition and stability of American families with children (Furstenberg 2014). At the midpoint of the 20^th century, most children were born into marital unions and about three-quarters remained in “traditional” families—defined as families with two biological parents married to each other, full siblings only, and no other household members—through childhood and adolescence. Contemporary children’s family structure histories are far more diverse, and far more distinct across social class and race (Furstenberg 2007). Life course estimates, for example, suggest that contemporary youth experience about one family structure transition by age 12, on average. Just over a quarter experience two or more transitions by age 12, with Black children experiencing significantly more changes than White or Latinx children. These differences are also pronounced by family structure at birth, with lower proportions of children born in married-parent families experiencing subsequent transitions compared to children born in cohabiting or single-parent families (Brown, Stykes & Manning 2016).

These changes in family composition and stability reflect a dynamic and reinforcing set of structural conditions and cultural forces that shape everyday life (Ruggles 2015). We briefly describe some of these relevant shifts, highlighting how these changes contributed to a distinctive era of family diversity and change.

Economic Changes

The extraordinary growth of the post-war U.S. economy that persisted from 1950 until the early 1970s contributed both to the Baby Boom—defined by early, nearly universal marriage and high fertility—and to married women’s entry into the labor force to meet increasing employer demand. This latter shift marked the transition to dual-earner households in which both members of a couple worked outside the home. Early on, women’s time in paid work occurred when their roles as mothers and workers were more compatible, either before they had children or after all their children started formal schooling. Poor women, especially women of color, balanced work and family responsibilities simultaneously (Amott & Matthaei 1991). Still, work patterns in this period were largely predicated on high and stable male employment, as men with high school degrees, especially White men, were more likely to be gainfully employed with a salary and benefits that could support a household. Beginning in the 1970s, as jobs in the service industry increased, economic growth slowed, and women continued to increase their educational attainment, women’s labor force participation rose sharply and women became more continuously attached to the labor force (Presser 1999; Ruggles 2015).

Although the movement of women into paid work is often considered the core of a constellation of changes contributing to contemporary family patterns (Bianchi 2000), significant shifts in men’s work lives matter. Simply put, men’s stable employment continues to be a precondition for marriage and marital stability (Sweeney 2002; Schneider & Hastings 2015; Killewald 2016). The labor force participation rates of working-age men have declined over the past 60 years (Ruggles 2015), a decline that has accelerated in the past 20 years (Tuzeman 2018), and median annual earnings have stagnated since the mid-1970s (Semega, Fontenot, & Kollar 2017). These changes in work and earnings have been most pronounced among men without a college degree (BLS Report 2017; Tuzeman 2018). Moreover, work practices have shifted (Kalleberg 2011). Jobs have become more precarious as factors like globalization, technological changes, and deregulation have disrupted the social contract between employers and employees in ways that transfer risk onto workers. There is also growing inequality in job rewards. Collectively, these changes have altered the economic foundation of marriage (Sweeney 2002) and contributed to growing heterogeneity in union formation behaviors (Kuo & Raley 2016).

Cultural Changes

Concurrently with changes in work, ideational shifts around gender, family, and fertility have also been profound. These changes are consistent with the Second Demographic Transition theory (see Zaidi & Morgan 2017). Advocates of this approach posit that the declining significance of political and religious authority and the rising significance of norms that emphasize self-realization, autonomy, and achievement alter how individuals think about happiness, love, and fulfillment and begin to animate decisions around marriage and divorce. As a consequence, Americans have become increasingly more tolerant of nonmarital sex and childbearing, cohabitation, and, more recently, same-sex marriage (Powell et al. 2016, Thornton & Young-DeMarco 2001; Treas, Lui, & Gubernskaya 2014). Still, Americans continue to valorize marriage, view divorce as a negative event, and anticipate that their own marriages will endure (Axinn & Thornton 2000; Gibson-Davis 2009; Edin & Reed 2005).

This tension—enduring support for marriage as a social and cultural institution coupled with increasing acceptance of family formation through nonmarital fertility, cohabitation, and legitimation of an increasingly diverse set of family forms (Powell et al. 2016)—reflects what Cherlin (2004, 2009) describes as the deinstitutionalization of marriage. More specifically, marriage has moved from the companionate model marked by a clear gendered division of labor, women’s economic dependence on men, and an expectation of love and support (Burgess & Locke 1945), to a more individualized model where expectations about fairness and individual emotional fulfillment are more central (Cherlin 2009; Coontz 2006). Moreover, marriage has become more selective, less a marker of conformity, and more a marker of prestige (Cherlin 2004; Edin & Reed 2005).

Social Location and Family Change

Changes in work, culture, and ideas about personal happiness have played out differently across social locations. Consistent with Sara McLanahan’s diverging destinies framework (2004), contemporary family structure trajectories increasingly differ by class and race. For both women and men, those with lower levels of education and people of color are more likely to be in jobs marked by lower wages, fewer benefits, and more uncertain tenure (Kalleberg 2011; Boushey 2017). Moreover, the economic prospects of Blacks, especially Black men without a high school degree, have been further minimized by rising levels of mass incarceration (Western & Muller 2014). Recent estimates suggest that about 27% of Black men have been incarcerated by age 35, that proportion increases to about 68% for Black men without a high school degree. Incarceration not only keeps men from gaining key educational and work experiences, but also makes them less hirable upon release (Pager 2003). This added penalty has marginalized the economic position of many Black men, especially those with less than a high school degree, making the economic bar for marriage and union stability more difficult to meet.

These patterns are often framed as evidence of the growing bifurcation in the American family system (Furstenberg 2014). Consistent with Cherlin, we suggest instead an emerging trifurcation of families, with education as a marker of advantage (Kuo & Raley 2016): the truly advantaged (e.g., those with a college degree) marry later, stay married, and bear and raise their children in marriage; the truly disadvantaged (e.g., those with no high school degree), largely forego marriage and even cohabitation due to economic concerns, bearing and raising their children outside of marriage (Gibson-Davis 2009); and the moderately advantaged adults (e.g., those with a high school degree or some college) have lost ground economically, compared to their parents and grandparents, yet maintain a precarious hold on ‘traditional’ family forms. Cherlin (2014) argues that this demographic group, especially the White working class, exhibits the most pronounced recent changes in family formation as well as considerable heterogeneity.

CONCEPTUALIZING FAMILY INSTABILITY

An early and organizing innovation undergirding much of the research on family instability is the instability and change perspective, first articulated by Wu and Martinson (1993). This perspective focuses on children’s response to the stressors that accompany changes in a parent’s marital or romantic trajectories (e.g., divorce, remarriage, or cohabitation). Like other research on specific types of family change, the instability and change perspective recognizes that the loss or addition of a parental figure can disrupt a child’s sense of security and create ambiguity in household rules, family relationships, and parental expectations about behavior (Wu & Martinson 1993; Teachman 2003). It may also create changes in family income and parents’ employment patterns (McLanahan & Sandefur 1994). Residential moves, which often accompany family structure change, can exacerbate the stress of family transitions by breaking neighborhood and school ties, extended family relationships, and children’s friendships (McLanahan & Sandefur 1994). Distinct from other perspectives, however, the family instability and change perspective takes as its starting point the observation that family structure change is repeatable, and its direct and indirect influence on children may compound over multiple transitions, particularly where children have limited time to adjust to new circumstances. Although many children never experience family structure change, those who experience one family transition are at risk of experiencing subsequent transitions and the concomitant stresses (Martinson & Wu 1993). Young people who experience multiple family transitions more often respond to this stress in ways that are linked with compromised well-being compared with those who experience no such transitions or only one (Cavanagh & Huston 2006).

Like earlier perspectives seeking to explain why divorce was problematic for children, this literature assumes that a stable home environment is an optimal context for raising children (Bloome 2017). What sets the instability and change perspective apart, however, is its focus on the family system and children’s and parents’ responses to the collective set of changes that unfold over time. Moreover, the family changes that can trigger stress are not limited to divorce or remarriage. In contrast, other perspectives about family structure and child well-being have focused on the consequences of a single event or a specific type of family change. For example, social control theory has been used to highlight how changes in parental monitoring and closeness following divorce or remarriage undermine the home environment and may explain differences in children’s behaviors (e.g., Matsueda & Heimer 1987; Maccoby & Martin 1983), socialization and social learning theories point to the lessons that young people learn when living with a single parent or married parents that can encourage different kinds of behaviors and attitudes (e.g., Caspi & Elder 1988), and a focus on economic change highlights income drops and financial stress following a divorce as key mechanisms linking family structure to child well-being (e.g., McLanahan & Sandefur 1994).

At its core, the instability and change perspective draws upon insights from these more static sociological explanations as well as insights from family stress theory (Minuchin 1985; see Hadfield et al. 2017). From this perspective, family stress is expected to occur when family demands are greater than the family’s ability to absorb them (Patterson 1993; Boss & Greenberg 1984). Such demands can be triggered by a partner change, the birth of a new child, a new job, or a health crisis. In other words, not all change is negative—change is expected and can be desirable—but demands greater than the family (or a family member) can handle introduce stress and dysregulation into the home. In the case of partner instability, the repeated movement of partners and sometimes other children in and out of the home can outstrip a family’s capacity to absorb these changes, making it increasingly difficult to reach a new equilibrium after each new change (e.g., Obsorne & McLahanan 2007). Such instability can disrupt the ways rules, patterns, and expectations are understood and enforced in families, and these disruptions can shape how children develop. To be sure, Amato’s (2000) divorce-stress-adjustment perspective also takes into account changes families undergo before, during, and after a divorce but still, divorce (and marriage) remain central to this framework. In its strongest form, the instability and change perspective is agnostic about the type of family change.

Selection into Unstable Family Trajectories

Selection offer a competing explanation for the observed association between family instability and child outcomes (Biblarz & Raftery 1999; McLanahan & Percheski 2008; Wu 1996). Although no social group is immune to family instability, there are important social, structural, and individual differences in the relative likelihood of experiencing family instability, including maternal attributes like educational attainment, personality differences, and early behavior (Fomby & Cherlin 2007). These differences, in turn, are related to how parents raise their children and organize the home, and to how children develop. Similarly, children themselves play a role in the stability of unions and parenting behaviors (Morgan, Lye, & Condran 1988; Crouter & Booth 2003). Thus, family instability and children’s compromised development may be spuriously related through common factors that are antecedent to family change. Given this, the observed links between partner instability and child development that do not account for selection often overestimate the family instability effect, reflecting a mixture of covariance and causality (Fomby & Cherlin 2007; McLanahan & Percheski 2008; Wu 1996). To varying degrees, family scholars have addressed this issue in much of the research on instability as we highlight below.

Data Collection on Family Instability

The available data infrastructure has expanded substantially to measure broad shifts in family structure. The first set of improvements relates to cohabitation. The National Survey of Families and Households (NSFH), a longitudinal study designed to understand “the causes and consequences of changing family and household structure” in a US-representative sample (Sweet, Bumpass, & Call 1988:5), collected cohabitation and marital histories from women and men of all ages in 1987–88 and oversampled cohabiting couples. These attributes allowed scholars to follow unions over time and measure changes in union status with greater precision, both retrospectively and prospectively (e.g., Bumpass & Sweet 1989). In the same period, ongoing national studies began capturing information on cohabitation and marriages of shorter duration. Since 1995, the National Survey of Family Growth (NSFG) has collected complete retrospective reports of cohabitation and marriage from women age 15–44 years and added data collection on men in 2002 (Potter et al. 1997). The 1979 National Longitudinal Study of Youth (NLSY79) began asking about the presence of a cohabiting partner in the household in 1987, and since 1990, has recorded the start date for cohabitation with a current spouse or partner (e.g., Oppenheimer 2003). Since 2002, NLSY79 has also recorded the start and end dates of all between-wave cohabiting unions that lasted three months or longer.

The US Census included a question about unmarried partners in the 1990 Census, a similar one was added into the Current Population Survey (CPS) in 1995. These represent important changes but likely underestimated counts of cohabitation because they were only asked of the household head (Casper & Cohen 2000). In 2007, CPS introduced a measure that identified all cohabiting partner in the household, regardless of whether they describe themselves as “unmarried partners” to the household head in the roster (Kennedy & Fitch 2012). Unmarried partners have been included in the Survey of Income and Program Participation (SIPP) household roster since 1996.

These innovations spurred scholarship on cohabitation and on transitions into and out of different union types (e.g., Rindfuss & VandenHeuvel 1990, Axinn & Thornton 1992, Teachman et al. 1991).Complemented by information on women’s and men’s fertility histories, these data allowed researchers to parse out the timing of union transitions and children’s exposure to partner change. Bumpass and Lu (2000), for example, used NSFH and NSFG to document increases in cohabitation and estimate children’s exposure to it. Raley and Wildsmith (2004) used the 1995 NSFG to highlight how a focus on martial transitions only underestimated family instability. They found that measuring changes in mothers’ cohabiting unions increased estimates of family instability by 30% for White children and by over 100% for Black children, compared to estimates based only on changes in maternal marital status. Some data sources also included retrospective reports of respondents’ own family structure histories. Wu and Martinson’s (1993) influential study of young women’s fertility behavior used NSFH to measure respondents’ childhood exposure to family instability and their own risk of a nonmartial birth.

The second set of improvements related to the emergence of national longitudinal studies focused on children. The NLSY79 Children and Young Adults (CNLSY/NLSY-YA) survey began in 1986 as a longitudinal second-generation study, following children born to female NLSY79 respondents, and has enabled a variety of rigorous research designs to assess the association between maternal union status change and children’s well-being across the early life course (e.g., Fomby & Cherlin 2007). Beginning in 1995, the National Longitudinal Study of Adolescent Health (Add Health) asked young people and their parents, typically their mothers, questions that allowed for the construction of retrospective family structure histories from the perspective of children since birth (e.g., Brown 2006; Heard 2007; Fomby, Mollborn, & Sennott 2010). The Early Childhood Longitudinal Study birth and kindergarten cohorts have allowed scholars to observe parents’ union status across childhood and study the implications of partner status for child development (e.g., Kim 2011; Mollborn 2016; Sun & Yi 2011). The Fragile Families and Child Wellbeing (FFCW) study, designed to focus on families at the greatest risk of experiencing nonmartial fertility, cohabitation, and partner instability, dramatically expanded this field of study, exploring the implications of family instability from early childhood through adolescence (e.g., Carlson, McLanahan, & England 2004; Lee & McLahanan 2015).

Other, smaller-scale studies typically lack representativeness but include richer or more frequent measures of family structure, child development, parenting, and biometric indicators (e.g., Crosnoe et al. 2014; Ackerman et al. 1999; Coe, Davies, Sturge-Apple 2017) to better capture the tempo of change as well as the physiological correlates of response to associated stress. Collectively, retrospective and prospective data on adults and children have been instrumental for measuring family instability and exploring the interplay among family instability and child development.

Measuring Family Instability

Family instability is measured in a variety of ways. Capaldi and Patterson (1991) provided the earliest descriptive evidence of an association between family instability and child behavior. They used a convenience sample of boys who had ever lived with both biological parents and created a measure of subsequent partner change by 4^th grade (0 transitions = intact family, 1 = single mother, 2 = stepfather, 3+ = multiple transitions). In introducing family instability as a topic of sociological inquiry shortly thereafter, Wu and Martinson (1993) used a count of the number of coresident parents’ union status changes experienced since the respondent’s birth, controlling for family structure at birth and last observation. Some scholars have also included changes in non-coresidential dating relationships in their counts of family instability (Osborne & McLanahan 2007; Cooper et al. 2011). More recently, Turney and Halpern-Meekin (2016; 2017) have considered relationship churning or the repeated exits and entrances of the same partner. Others, mostly developmental psychologists, operationalized family instability as a count of a coresident parent’s partner transitions plus children’s exposure to changes in primary caregiver, residential moves, and the death of a close family member (Ackerman et al. 1999; Coe et al. 2017; Sour et al. 2015). Collectively, these measures prioritize changes in parents and parent figures in children’s lives. The focus is on the number of changes, rather than the type of family structure change children experience.

A growing subset of research on family instability has focused on transition type (e.g., Lee & McLanahan 2015; Mitchell et al. 2015). This work is guided by the expectation that parents’ entrances into and exits out of unions are not symmetric in their influences on children’s behavior. Dissolution, especially of the union between the child’s biological parents, is viewed as especially consequential: it is associated with decreased parental time and economic resources for children as well as increased levels of stress in the home (Lee & McLanahan 2015). (Re)partnering, on the other hand, may have offsetting effects. A new partner can increase the money and time available to children and reduce maternal stress and depression (Osborne, Berger, & Magnuson 2012) but can also disrupt existing family roles and routines and introduce new sources of stress in the home. It might also complicate children’s relationships with the nonresident biological parent. Moreover, children appear to respond differently to the formation or dissolution of marital compared to cohabiting unions (Brown 2006).

CHILDREN’S ADJUSTMENT TO FAMILY INSTABILITY

Scholarship on the implications of family instability for children across the early life course has focused on three domains: social and emotional behaviors, academic performance and cognitive functioning, and young people’s sex, fertility, and union formation behavior. Consistent with the life course perspective, this work starts from the position that parents’ romantic lives are intimately linked with their children’s adjustment (Elder 1998). These indicators are markers of child well-being at key developmental periods (e.g., childhood, adolescence, and young adulthood) and can also signal later socioeconomic attainment. We describe key findings within each developmental domain, review common explanations for these linkages, and then discuss population heterogeneity across findings. A more comprehensive index of related empirical research is available as an online appendix at [URL].

Socioemotional Behavior

The developmental domain most consistently linked to family instability is children’s socioemotional and problem behavior, typically operationalized as internalizing and externalizing behavior in early and middle childhood and as depression and delinquency in adolescence. In fact, in one of the earliest studies of family instability, Capaldi and Patterson (1991) documented the association among repeated family structure change, behavior problems, and social adjustment during middle childhood among boys in a small school-based sample.

Subsequent work has explored potential confounders, mediators, and moderators of these linkages (e.g., Adam & Lansdale 2002; Cavanagh & Huston 2006; 2008). Using data from FFCW, Osborne and McLanahan (2007) documented that each change in maternal coresidential and nonresidential partnership status since a child’s birth increased mother-reported aggressive and anxious/depressive behaviors at age 3 in unadjusted models. These associations were reduced to non-significance after accounting for contemporaneous maternal stress and parenting quality. Fomby and Cherlin (2007) used the prospective two-generation design of NLSY79/CNLSY to measure the association between cumulative (coresidential) family instability and maternal and child reports of externalizing behaviors and delinquency in middle childhood and early adolescence net of maternal selection into unstable unions. Indicators of maternal selection were measured for mothers in the earliest waves of NLSY79, prior to the birth of their children. Change in maternal union status increased externalizing behavior scores and delinquency scores among White children but not among Black children.

These early findings point to the association between maternal relationship transitions and children’s socioemotional behaviors across the early life course, but concerns about selection and the asymmetries between romantic partners’ exits and entrances continue to motivate scholarship in this research area. Many of the observed and unobserved correlates of family instability—notably changes in household income, mothers’ employment status, and parenting behaviors—co-occur with partner transitions and can mediate the impact of family instability on child development. Such dynamic bundling of events can make it difficult to determine the causal significance of family structure transitions per se (McLanahan & Percheski 2008).

To better account for co-occurring changes as well as the expected differences associated with partner exits and entrances, some scholars have focused on transition type, typically measured within a narrower observation window. For example, using FFCW, Lee and McLanahan (2015) measured instability as both counts and transition types and estimated a host of multivariate models (i.e., random effects), fixed-effects, and marginal structural models that addressed different types of confounds to predict children’s socioemotional and cognitive outcomes at age 9. Transition type better explained variations in children’s socioemotional development compared with counts of family structure changes, but the type of transition that was most salient to children’s well-being varied by race/ethnicity. Exits out of two-parent families were most significant for White children and entrances into two-parent families mattered most for Latinx children.

Cognitive Achievement and School Performance

Family instability is also implicated in children’s schooling and cognitive achievement. Children exposed to more maternal coresidential and dating instability in early childhood reported lower verbal ability (PPVT score) and more externalizing problems and social problems at age 5 (Cooper et al. 2011). Heard (2007) found that maternal transitions during early childhood were negatively associated with measures of school functioning in adolescence. Paternal transitions also mattered, but transitions away from mother, especially early ones, had more severe consequences for schooling behaviors. Adolescents who experienced family instability were also less likely to complete higher-level mathematics in high school compared to peers in stable families, making them less prepared for college (Cavanagh et al. 2006). Finally, family structure changes across the early life course were negatively associated with the odds of earning a college degree, a key marker of status attainment (Fomby 2013).

Collectively, these findings suggest a link between changes in family structure and children’s academic careers, one that shapes the intergenerational transmission of (dis)advantage. Still, accounting for parental selection into unstable unions generally weakens the observed associations between family instability and indicators of cognitive achievement. Lee and McLanahan (2015) found that transitions out of two-parent families had a negative association with children’s verbal scores, but these effects were more modest than those tied to socioemotional behaviors. Fomby and Cherlin (2007) observed no significant association between family instability and children’s cognitive functioning once static maternal selection characteristics were taken into account, both for Whites and Blacks. Using fixed-effect models, Aughinbaugh and colleagues (2005) found little longitudinal variation in children’s test scores by exposure to maternal union transitions.

Overall, the relationship between family instability and school performance appears to operate less through changes in cognitive functioning, a finding that is consistent with the divorce literature (McLanahan 1999), and more through changes in young people’s school setting and their behaviors, orientations, and attitudes about schooling, which can make educational attainment more challenging.

Romance, Sex, and Family-Building Behaviors

Family instability is also consistently associated with young people’s own romantic, sexual, and family-building behaviors in adolescence and young adulthood. Wu and colleagues (1993; 1996) identified a positive relationship between cumulative family transitions and risk of premarital birth for young White and Hispanic women. Hofferth and Goldscheider (2010) built upon these findings to consider the timing and union context of first birth in a more recent cohort of young women and men. For women, early instability was positively associated with the timing of their transition to parenthood and with the likelihood of being a married or single parent. For men, frequent instability lowered the likelihood of transitioning to married fatherhood and increased the chance of transitioning to cohabiting fatherhood. These findings echo work by Cavanagh and colleagues (2008), who found that family instability was especially consequential for boys’ romantic involvement in adolescence.

Repeated family structure change is predictive of the timing, type, and stability of young people’s first coresidential unions. Young women in both older (NLSY79; Teachman 2003) and more recent (Add Health; Ryan et al. 2009) cohorts who experienced more instability in childhood transitioned into cohabiting unions more quickly than others. Young men in the more recent cohort also cohabited earlier than others. Using latent class analysis to capture the timing and sequencing of key life course transitional events by age 24, Fomby and Bosick (2013) found that family instability was positively associated with membership in classes characterized by low college attendance and early union formation, childbearing, or entry into full-time work. Teachman found no significant link between cumulative family change and one’s own divorce in an older cohort (2002), but Amato and Patterson (2017) found support for the intergenerational transmission of instability by around age 30 in a more recent cohort.

What Explains these Associations?

The instability and change perspective and the broader family literature have guided research to explain why repeated family structure change is negatively linked with well-being across the early life course. Most research has focused on family income, stress and mental health (maternal and child/adolescent), parenting and parent-child closeness, and changes in other developmental domains as explanatory mechanisms. In the empirical record, each domain is associated with repeated family structure change, but evidence that they mediate the association between family instability and child well-being is modest. For example, Amato and Patterson’s (2017) study of the intergenerational transmission of relationship instability included a comprehensive set of mediators and no item explained more than 1% of the association. Here, we briefly review current knowledge.

Economic Resources

Family structure is tightly linked to economic resources. As a selection mechanism, women raised in low-income families are less like to marry and more likely to have a nonmarital birth and to experience multipartner fertility than other women, all factors which reduce the likelihood of entering or remaining in a stable union (Cancian et al. 2011; Gibson-Davis 2009). After a child’s birth, the entry or exit of a biological parent or a parent’s partner in the household affects the amount of money coming into the home, the labor force participation of the resident parent, and the number of people relying on the family’s income (McLanahan & Sandefur 1994). In turn, low income is associated with child well-being across the early life course (Duncan & Brooks-Gunn 1999). Thus, scholars have long emphasized household income as a key mechanism linking family structure and child well-being (McLanahan & Sandefur 1994).

Wu (1996) used prospective data from NSLY79 to consider whether income volatility, or frequent rises and drops in family income, confounded the observed link between family instability and women’s transitions to nonmarital births. Instead, income volatility and family instability operated independently of one another. This finding, combined with concerns about collinearity between family structure change and income volatility, has limited the amount of work that considers both income and family instability. Still, work by Bloome (2017) and Ryan and colleagues (2015) demonstrates that family instability and income may not have independent or additive effects on child well-being but can be interactive, with the short- and longer-term effects of family instability stronger among those from families with middle or higher incomes.

Stress and Mental Health

How mothers respond to change in union status potentially mediates the link between family instability and children’s well-being across the early life course. Collectively, related research, much of it done with FFCW, suggests that coresidential and dating transitions are associated with higher levels of parenting stress (Beck et al. 2010; Cooper et al. 2009; Halpern-Meekin & Turney 2016) and short-term increases in depression for mothers (Meadows, McLanahan, & Brooks-Gunn 2008). Some research suggests that maternal exits from relationships with a child’s biological father and entrances into relationships with a social father are each associated with higher levels of reported parenting stress, compared to mothers in stable coresidential relationships (Cooper et al. 2009), while other work has identified exits from marital or cohabiting unions as a distinctive driver of increased parenting stress and declining perceived social support (Osborne et al. 2012).

Despite these associations, there is only modest evidence that maternal stress and mental health mediate the relationship between family instability and children’s well-being (e.g., Carlson & Corcoran 2001; Cavanagh & Huston 2006, 2008; Fomby & Osborne 2017). As noted above, Osborne and McLanahan (2007) found that the link between instability and children’s externalizing behaviors was reduced to non-significance after contemporaneous maternal stress and parenting quality were taken into account. Work conducted on less disadvantaged populations, however, shows little or no attenuation in the association between family instability and children’s behavior (Cavanagh et al. 2006; Cavanagh & Huston 2008; Fomby et al. 2016).

Only a handful of studies have considered child and adolescent mental health and stress as a potential mediator of later behavior (Amato & Patterson 2017). For example, Cavanagh and colleagues (2006) considered adolescent depression and parenting practices at the start of high school as a mechanism linking family instability and students’ course-taking patterns at the end of high school. Although both factors were associated with course-taking, neither offered leverage to explain the associations between family instability and young people’s behavior.

Scholarship on younger children’s well-being, however, does point to child stress as a potential mediator. Suor and colleagues (2015) explored the interplay among family instability, changes in children’s basal cortisol levels (a key biomarker of stress response), and their cognitive functioning at age 4. Using growth mixture models with data from a longitudinal sample of about 200 children, they found distinct patterns of basal HPA (hypothalamic pituitary adrenal) axis activity that were tied to early reports of family environment. They argued that the immediate caregiving environment likely calibrates children’s stress response to maximize functioning in anticipated conditions. Family instability was associated with extreme (very high or low) basal cortisol trajectories, which in turn were associated with lower cognitive abilities compared to moderate trajectories. Using a convenience sample of about 240 preschool-age children, Coe and colleagues (2017) found that the association between family instability and teacher reports of children’s school adjustment problems at the start of formal schooling was explained by child callousness, or emotional insensitivity. Those exposed to more family instability were less able to process and respond to emotional events in ways that made them more vulnerable to behavioral problems later on. Collectively, these studies point to children’s responses to stressors as a mechanism shaping later behavior. Future work on larger, representative samples is needed to confirm these associations and move theory forward.

Parenting Practices and Parent-Child Closeness

The entrance or exit of a coresident romantic partner or other household members can complicate and compromise relationships between parents and children. King (2009), for example, found that the mother-child relationship weakened when a cohabiting, but not a married, stepfather entered the home. Hence, changes in parent-child closeness, parental supervision, parental involvement, and parenting behaviors, including harsh parenting and sensitivity potentially mediate the association between family instability and child well-being. However, to date there is little evidence in support of any such mediation (e.g., Amato & Patterson 2017; Amato & Sobolewki 2001; Beck et al. 2010; Cavanagh et al. 2006).

Change in Other Domains

Scholars have also considered concurrent changes in domains outside of the family such as child care, school, and neighborhood, to explain the impact of family instability on children’s and young adults’ outcomes. Much of this work has focused on young children. Crosnoe and colleagues (2014), for example, found that family structure transitions were associated with changes in the type and quantity of early care, as well as the number of care arrangements used. Their results point to the collateral changes in children’s ecology set in motion by family instability. Fomby and Mollborn (2017), building on the concept of children’s developmental ecologies or the interrelated features of children’s everyday environments that affect their well-being (Mollborn 2016), considered how frequent changes in multiple settings shape development. They found that frequent and persistent exposure to change across domains was more strongly linked to teacher-reported behavior scores than was change in any single domain, such as family structure, child care, maternal employment, or residential mobility.

Heterogeneity and Family Instability Effects

The empirical record on family instability highlights how exposure to family instability shapes key developmental outcomes that can contribute to growing inequality. Still, a theme running through the findings and the broader literature on families is one of heterogeneous effects. In other words, family instability may be causally linked with poor outcomes for some children, but inconsequential or even beneficial for others (Turney 2015). Here, we highlight findings that suggest differences in associations related to age, gender, race/ethnicity, and social class.

Age.

Exposure to early family instability (by age 5) appears to have more lasting behavioral consequences than does instability experienced later (e.g., Bzostek & Berger 2017; Fomby 2013; Fomby & Bosick 2013; Heard 2007; Ryan et al. 2015). For example, Cavanagh and Huston (2008) considered the effect of early and concurrent instability on children’s social development at the end of elementary school. Early instability had a lasting effect on teacher reports of externalizing behaviors and peer competency, as well as child reports of peer loneliness. Family instability experienced during middle childhood added little to these associations.

Two explanations are plausible. First, attachment theory (Bowlby 1982) and other theories of child development point to early childhood as a developmentally sensitive period (Shonkoff & Phillips 2000). Children are most dependent on their parents during this development stage, so family structure changes may be more consequential. Such changes alter the template upon which children develop later social relationships and competencies, leaving young people less able to navigate social relationships with peers and teachers and to regulate behaviors at home and in school. The second explanation is consistent with a selection argument: parents who break up when children are very young might differ from parents who stay together or break up later in terms of relationship skills like conflict resolution and dispositions linked with children’s own social behavior.

Gender.

Although gender differences in the likelihood of experiencing divorce are modest (Diekmann & Schmidheiny 2004), many studies find that boys exposed to family instability fare worse than girls in terms of externalizing and attention behaviors (e.g., Cavanagh & Huston 2008; Cooper et al. 2011) and adolescent romance (Cavanagh et al. 2008). Part of the gender difference is likely tied to the social construction of gender and the ways boys and girls interpret and compensate for changes in parental relationships (Maccoby 1998). Girls, for example, often have intimate friendships with peers whom they can turn to for support (Giordano, Longmore, & Manning 2006). Because boys have fewer intimate friendships, they may have fewer resources to draw upon, so they may cope with changes in their home environments in ways that can increase the likelihood of engaging in externalizing or delinquent behaviors. This need for connection may also encourage romantic and sexual relationships at earlier ages (Giordano et al. 2006).

Differences in how parents raise boys and girls also may explain this discrepancy. Parental monitoring and autonomy-granting are linked with delinquency (Chen 2010), sexual debut (Browning, Leventhal, & Brooks-Gunn 2005), and early parenthood (Hofferth & Goldschieder 2010), but these associations differ by gender (Kerr & Stattin 2000). In the context of family instability, parenting behaviors that connect children to adults may become less consistent and supportive (Buchanan et al. 1996). These shifts can increase opportunities for young people, especially boys, to engage in problem behaviors.

Race and ethnicity.

Black and Latinx youth are more likely to experience family instability compared to their White peers (Brown et al. 2016; Raley & Wildsmith 2004). But unlike White youth, Black and Latinx youth report no difference or only modest differences in socioemotional behavior (Fomby & Cherlin 2007; Fomby et al. 2010), age at sexual debut (Wu & Thompson 2001), timing and marital context of childbearing (Wu & Martinson 1993), and academic performance (Fomby & Cherlin 2007) compared to peers raised in stable families.

Socioeconomic status.

Ryan and colleagues explored whether the association between family change within and across developmental windows and children’s socioemotional behaviors were conditioned by family income in the 3 years prior to the child’s birth (Ryan et al. 2015) or by family structure at birth (Ryan & Claessen 2013). In terms of income, family instability was more consequential for children born to middle and high-income parents; no change in problem behavior was identified for children born to low-income parents (Ryan et al. 2015). Bloome (2017) considered the interplay between family instability and intergenerational income mobility and found that all children exposed to family instability saw more downward income mobility than did those raised in stable homes; this effect was especially strong for children raised in middle or upper income families. Family instability was also more consequential for behavior among children born into married-parent families than others (Ryan & Claessen 2013). Bzostek and Berger (2017) found that family instability was not associated with children’s later socioemotional behaviors among those born to cohabiting parents or single parents. Instead, family instability was associated with lower functioning among those born to married parents.

Collectively, these findings related to race/ethnicity and socioeconomic status represent a puzzle. Much of the research on family instability seeks to understand how families contribute to or ameliorate increasing inequality among children. Yet these findings suggest that the impact of family instability on child well-being is weakest for those who are most likely to experience it. We consider three plausible explanations. First, the socioeconomic disadvantage hypothesis posits that the more modest associations observed for Black children or children raised in low-income or nonmarital families occur because many of these children also experience other types of disadvantage compared to other children. In other words, the “saturation of disadvantages” may mean that partner instability is not an event stressor, so it offers no additive consequence for some children. Because children born in married-parent families and White children experience fewer stressors overall, family instability may be especially disruptive, in part because it is infrequent (Turney 2017).

Second, the protection hypothesis asserts that emotional and instrumental support provided to parents and children from a network of kin and kin-like figures can insulate children of color and lower-income children during periods of change. Organization in Black families extends beyond the nuclear household (Newman 1999; Sarkisian 2007) and families of color and lower-income families are more likely to live in extended kin households than are others (Raley et al. 2018). In contrast, Whites and middle and upper-class families less often benefit from these sources of support, meaning that family structure changes, especially multiple transitions, may translate into worse outcomes for children.

A final explanation is methodological. These null findings might reflect problems with the comparison group for Black, low-income, or nonmarital families. Many individuals in these demographic groups have been exposed to family instability. Thus, comparisons within low-income families, for example, may be harder to establish due to ceiling/floor effects. Moreover, vulnerable families, where family instability may be higher and child well-being may be lower, may be underrepresented in longitudinal, school-based, or nationally representative samples (Pettit 2012). Conversely, such surveys may be better equipped to measure the implications of family instability among more advantaged youth (Perreira et al. 2005). In fact, with the exception of Bzostek and Berger (2017), which used the FFCW, most of this evidence regarding social location heterogeneity is found in nationally representative studies. Careful consideration of samples, measurement, and social theory is needed to address these puzzles.

FUTURE DIRECTIONS IN FAMILY INSTABILITY RESEARCH

A substantial body of research has established robust associations between children’s exposure to a coresident parent’s repeated changes in union status and their elevated risks of behavior problems, delinquency, poorer academic achievement, and early family formation. This work has been at the forefront of expanding how scholars think about families. Yet in many ways, research on how family instability influences children’s outcomes is just beginning. We highlight six areas for development.

First, a tension between conceptualizing family instability as the repeated experience of disruption or as the movement between qualitatively distinct states permeates this literature. The first approach, often using counts of family structure change, emphasizes the costs of adaptation and adjustment to changes in family systems as factors that uniquely contribute to child well-being, overlooking the qualitative experiences associated with these changes. The second approach focuses on asymmetries in transition type, considering whether union dissolutions are more harmful to children born to married parents compared to those born to a single parent who eventually repartners. This approach conflates the transition event with the origin and destination states, making it challenging to measure the potentially separate influences of change and status. Parsing out asymmetric effects is useful but does not necessarily translate into a richer understanding of the meaning of family instability for children. Fundamentally, is a focus on family instability another way to think about why not living with two married parents may be detrimental to children or is there something more to be gained by considering cumulative change? We support the latter position and, to that end, encourage more focus on the conditions under which children and adults respond to disruption, uncertainty, and transition.

Second, greater attention to how family change is defined can clarify research on family instability. Much of the research reviewed here continues the focus on changes in a coresident parent’s (usually a mother’s) union status. A related line of inquiry has demonstrated that household entries and exits by extended family members, such as grandparents and aunts, are independently predictive of children’s well-being net of parents’ union instability. This is especially true in families of color, including immigrant families (Van Hook & Glick 2007). These transitions can also introduce family stress, but the mechanisms that explain their impact on children’s development may be distinct. Attention to how repeated union and household transitions are conceptually similar or distinct in terms of their causes and consequences for children can help to refine the way we think about family instability.

At the same time, defining family structure through membership in a child’s physical household imposes an arbitrary boundary on children’s active family systems. Work to date has not considered whether repeated union status change in a nonresident parent’s household spills over to influence children living elsewhere. Further, only one national secondary data source (FFCW) records parental relationships with nonresident romantic partners in each wave.

Third, the family instability perspective can be applied to parents’ periodic absence from the household due to events such as military deployments, labor migration, and long-distance employment postings. Research on union instability has contended with confounders, pre-union dissolution conflict, household income change, and the characteristics of new entrants into a child’s household. Focusing on families that remain ‘intact’ but experience one parent’s periodic absence can control for these factors. Recent work on churning, or parents’ break-ups and reunions with the same partner (Halpern-Meekin &Turney 2016), and on parents’ incarceration (Turney & Wildeman 2013) shed some light on how children respond to stress and disruption associated with one parent’s entrance into and exit from the household. This work can also refine our conceptualization of family instability.

Fourth, measurable mechanisms to substantiate a causal relationship between family instability and child outcomes are relatively underdeveloped. On the one hand, little evidence of mediation has been identified. On the other hand, repeated family change is not equally predictive across child outcomes or across subpopulations, suggesting that family instability is not irreducibly a causal mechanism in shaping well-being. The explanations most rigorously evaluated to date have not anticipated this variability, but making this heterogeneity in associations a subject of inquiry may yield new insights about why (or whether) instability is consequential for child development. Techniques to assess heterogeneous treatment effects offer a methodological advancement that can be applied to existing data (Turney 2015). Relatedly, historical and cross-national comparisons can bring into relief how economic, policy, and cultural contexts condition family instability.

Fifth, much of the research on family instability has been conducted at the micro-level, emphasizing individual-level characteristics that predispose parents to launch unstable union trajectories, interpersonal dynamics in unstable families, and individual-level child outcomes. Macro-level contextual factors like ideational and economic change that may have given rise to union instability remain largely unexplored but can elucidate how social forces influence the probability of experiencing family instability and shape its consequences. We know, for example, that local wage conditions and unemployment rates are associated with a single event like the transition to marriage (e.g., Cherlin, et al., 2016; Kearney & Levine, 2014). How might a more dynamic consideration of local vulnerability in labor markets over time inform our understanding of repeated family structure change? In other words, how might macro-level factors shape family instability? Datasets like NLSY97and the Panel Study of Income Dynamics that follow adults over time can be applied to treat family instability as an outcome in itself.

Finally, the field needs more qualitative research. Questions about meaning and measurement of family instability can be addressed by observing families who undergo change across social class and race. Moreover, such an approach could provide new insights into mediating mechanisms. Parents and children have likely found new ways to create routines, bonds, norms, and expectations in the presence of family instability, multipartner fertility, sibling complexity, and family diversity, yet survey research has not yet established valid and reliable measurements to reflect such change. Qualitative studies that inform theory and measurement will enable researchers to see contemporary family relationship quality and family processes more clearly.

At its core, the scholarship on family instability reflects the family as a resilient institution adapting to broadly evolving economic and cultural contexts. As future work continues to map the etiology of family instability and the conditions under which it may compromise child well-being, we encourage family researchers and policy makers to be cognizant of the conditions that give rise to and result from instability as a distinctive component of contemporary family life.