Divorce, Repartnering, and Stepfamilies: A Decade in Review

Abstract

This article reviews key developments in the past decade of research on divorce, repartnering, and stepfamilies. Divorce rates are declining overall, but they remain high and have risen among people older than age 50. Remarriage rates have declined, but the overall proportion of marriages that are remarriages is rising. Transitions in parents’ relationships continue to be associated with reduced child well-being, but shifting patterns of divorce and repartnering during the past decade have also reshaped the family lives of older adults. We review research on the predictors and consequences of these trends and consider what they reveal about the changing significance of marriage as an institution. Overall, recent research on divorce, repartnering, and stepfamilies points to the persistence of marriage as a stratified and stratifying institution and indicates that the demographic complexity of family life is here to stay.

In his presidential address to the Population Association of America, demographer Larry Bumpass (1990) argued that no change had so significantly altered U.S. family life as the fact that most marriages would not last a lifetime. The high and rising rates of divorce that prevailed during the 1970s and 1980s concerned family scholars for several reasons. First, many worried that the importance of marriage was waning, although continued high rates of remarriage during this period suggested an enduring attachment to marriage as an institution. A second concern was how rising divorce rates would affect the nation’s youth, as parents’ divorce was associated with poorer child outcomes and repartnering after divorce often creates stepfamilies. A third concern was how divorce might contribute to socioeconomic and gender-based inequality. Divorce is a stratified and stratifying life event: It varies across groups in both its likelihood of occurring and its consequences. Divorce is more common among the least educated and tends to bring more severe economic consequences for women than men—with many women not recovering economically unless they remarry. Divorce and remarriage thus become mechanisms for the transmission of inequality both within and across generations.

The family landscape has shifted in the 30 years since Bumpass penned his address. This article reviews what we have learned in the past decade about divorce, remarriage, and stepfamilies. Given the vast breadth of our charge (which, in recent Journal of Marriage and Family decade-in-review issues, has spanned two separate articles), we necessarily take a bird’s eye view of our topic and focus particularly on recent scholarship on the United States. We begin by reviewing the demography of divorce and remarriage and ask what factors contribute to these trends. Many have identified women’s employment as an important contributor to rises in divorce, but recent research shows that women’s employment is no longer associated with marital dissolution (Killewald, 2016), and the positive association might have always been due to women increasing labor force participation in anticipation of divorce (Özcan & Breen, 2012). We next consider what we have learned about how divorce and remarriage affect the well-being of adults and children and about the changing dynamics of stepfamilies. Of note, in the United States, divorce continues to be financially devastating to mothers and children (Tach & Eads, 2015). Finally, we consider what change and variation in patterns of divorce, remarriage, and stepfamilies suggest about the future of marriage and family life, highlighting important unanswered questions for research.

The Demography of Divorce in the United States

The United States has long had one of the highest divorce levels in the world, with nearly half of marriages expected to end in divorce or separation (Kennedy & Ruggles, 2014; Kreider & Fields, 2002). Yet aggregate divorce rates have declined in recent decades: At its peak in 1980, the annual U.S. divorce rate was 22.8 per 1,000 married women, but by 2016, that number had declined to 16.7 per 1,000 married women (Hemez, 2017). This trend has varied considerably by age, however. Divorce rates for younger married couples have declined since the 1990s (Cohen, 2019; Kennedy & Ruggles, 2014). Recent divorce rates among Americans aged 50 and older dramatically increased during these years—doubling between 1990 and 2010—a phenomenon that Brown and Lin (2012) dub the “gray divorce revolution.” Despite these disparate trends, the divorce rates remain considerably higher for younger than for older adults.

What we do not yet know is whether recent divorce trends reflect intrinsic shifts in marital stability or changes in the composition of populations of married couples (e.g., Lin, Brown, & Cupka, 2018). For example, marriage rates among young adults have declined in recent years, as more couples choose to cohabit rather than marry. Declines in first marriage have been especially steep among less-educated women and those who have had premarital pregnancies (Gibson-Davis & Rackin, 2014; Schneider, Harknett, & Stimpson, 2018), populations that also tend to have elevated divorce rates. This may reflect a broader shift in the pool of younger married couples toward those with the most stable relationships. Among older Americans, the share of marriages that are remarriages for one or both partners has increased over time. Remarriages tend to be less stable than first marriages (Bumpass & Raley, 2007), and thus this shift may have contributed to the rise in “gray divorce.” The extent to which recent growth in later-life divorce reflects aging and life course variation versus the unique experiences of specific cohorts (e.g., the Baby Boom cohorts, born 1946 through 1964) remains an important open question. Whatever the underlying causes, recent data suggest that the rise in gray divorce abated during the past decade (Cohen, 2019).

Divorce rates tend to be higher for some groups than for others. For example, as U.S. divorce rates rose during the 1970s, they did so more steeply for Black women than for Hispanic and non-Hispanic White women (Castro Martin & Bumpass, 1989). The White–Black divorce differential continued to be large during the 1980s, but a substantial wave of immigration reduced Hispanic women’s divorce rate, as immigrant populations tend to have lower divorce rates than the native born (Bean, Berg, & Hook, 1996; Raley & Bumpass, 2003). More recent research suggests that overall racial and ethnic differentials in divorce rates remain large (Cohen, 2019; Copen, Daniels, Vespa, & Mosher, 2012). Asian and foreign-born Hispanic women have the lowest divorce rates, whereas Black, U.S.-born Hispanic, and Native American women have among the highest rates of divorce. The reasons for these differences remain insufficiently understood, but likely involve complex and dynamic relationships among individual, family, and community-level characteristics, including beliefs, norms, and structural factors such as economic opportunities (e.g., Barr & Simons, 2018; Raley, Sweeney, & Wondra, 2015).

Scholars in the past decade have also considered union stability among same-sex couples. Many European countries allow same-sex couples to marry (e.g., the Netherlands) or otherwise formalize their relationships and obtain rights and legal protections similar to those of married couples (e.g., Denmark, Norway, Sweden). The recent push for extending marriage rights to same-sex couples in the United States, which were achieved in 2015, suggests both the continued symbolic importance of marriage and the value of institutional supports accompanying marriage. Thus far, however, few studies have investigated instability among same-sex couples who are married. This is most likely due to a lack of available data and generally small samples sizes of same-sex couples (Manning & Joyner, 2019). The data we do have generally suggest that same-sex cohabiting unions in Europe tend to be less stable than different-sex cohabiting and marital unions combined, but that among cohabitors in the United States, same-sex and different-sex couples have similar levels of stability (Bennett, 2017; Lau, 2012; Manning, Brown, & Stykes, 2016; Rosenfeld, 2014). In settings where same-sex partnerships were more recently legally recognized through civil unions or marriage, such as in the United States, same-sex couples in formalized relationships sometimes have higher levels of stability than different-sex married couples (e.g., Badgett & Herman, 2013; Ross, Gask, & Berrington, 2011). This may reflect selectivity among the early adopters of registered partnership formation, who include a greater share of long-term couples who have been waiting to formalize their relationships (Manning & Joyner, 2019). Results vary across studies (and country settings) regarding whether differences exist between female and male same-sex couples in levels of instability (Andersson, Noack, Seierstad, & Weedon-Fekjær, 2006; Kalmijn, Loeve, & Manting, 2007; Lau, 2012).

Predictors of Divorce

In 2010, Amato wrote a decade-in-review article on divorce that identified the following nine consistent predictors of divorce: teen marriage, poverty, unemployment, low educational attainment, premarital cohabitation, premarital fertility, interracial marriage, previous divorce, and parental divorce (Amato, 2010). The past 10 years of research have advanced our understanding of two of these predictors in particular. One line of research has spurred new questions about the association between premarital cohabitation and divorce, and another has given us new information about the relationship between employment and divorce. Other recent work has also focused on the relationship between education and divorce risk. Most of this work considers marital separation in addition to legal divorce both because of the short average duration between separation and divorce and because of variability across individuals and groups in the timing of obtaining a legal divorce. Although our focus here is particularly on the dissolution of marriages, researchers are also increasingly interested in the dissolution of cohabiting unions.

Premarital Cohabitation

The positive association between premarital cohabitation and divorce is in some ways counterintuitive. Living together should help couples evaluate their relationship and determine whether they are a good match. So how do we account for this association? One perspective is that people who choose to cohabit before marriage have distinctive characteristics that predispose them to divorce and that these characteristics account for their increased incidence of divorce (Impicciatore & Billari, 2012; Manning & Cohen, 2012; Smock, 2000). Premarital cohabitation may also increase the risk of divorce by truncating the marital search: couples may begin cohabiting without giving much consideration to long-term compatibility and then marry out of inertia or in response to a pregnancy (Stanley, Rhoades, & Markman, 2006). A third explanation reinforces the salience of age: the younger a couple was when they began living together, the more likely they are to eventually divorce (Kuperberg, 2014). This effect explains a substantial portion of premarital cohabitors’ higher divorce rate: These couples tend to have begun living together at younger ages than couples who got married without living together first. This age effect holds even without premarital cohabitation: Early marriage is associated with divorce and contributes to a higher risk of divorce among conservative Protestants (Glass & Levchak, 2014).

Some recent studies conclude that the association between premarital cohabitation and elevated risk of divorce may have weakened over time (Manning & Cohen, 2012; Musick & Michelmore, 2015; Reinhold, 2010). These results are consistent with a “diffusion perspective,” which suggests that as cohabitation moves from being uncommon to a modal experience, it becomes less associated with particular individual characteristics that may predispose couples to divorce. (For evidence consistent with this perspective for union stability in Europe, see Liefbroer & Dourleijn [2006].) Recent work on the United States suggests, however, that premarital cohabitation is associated with a reduced risk of divorce in the first year of marriage and an increased risk at longer marital durations. When we take this variation into account, we see that the link between premarital cohabitation and divorce has not weakened but instead has remained stable over time (Rosenfeld & Roesler, 2019).

Women’s Employment

Recent research has investigated whether women’s employment increases the risk of divorce and whether this effect has weakened for recent cohorts. At the individual or couple level, wives’ employment sometimes predicts divorce (Teachman, 2010), but this association is sensitive to context. For example, Sayer, England, Allison, and Kangas (2011) find that women’s employment is associated with an increased likelihood that wives will leave their husbands, but only for wives who report low marital satisfaction. Other research suggests that the association between wives’ employment and divorce is limited to marriages begun before 1975 (Killewald, 2016). Similarly, wives who earned more than their husbands had an elevated risk of divorce in the 1960s and 1970s, but not in the 1990s (Schwartz & Gonalons-Pons, 2016). When wives’ employment and earnings are positively associated with divorce, at least part of the reason may be that women increase their participation in the labor force in anticipation of the end of their marriage (Özcan & Breen, 2012; Tamborini, Iams, & Reznik, 2012).

Although it is not clear that wives’ employment weakens individual marriages, growth in women’s labor force opportunities and rising risks of divorce have likely changed marriage as an institution. The growing risk of divorce encouraged women to get more education and to stay employed after getting married. Dual employment became normative, and this change in social norms has changed gender arrangements within marriage. Wives do much less housework than before, and husbands do a little more (Bianchi, Sayer, Milkie, & Robinson, 2012). Wives’ employment and earnings are no longer stigmatized and are not today associated with marital disruption, and shifts in the gendered expectations of marriage may partly account for the weakening association between women’s earnings and divorce (Schwartz & Gonalons-Pons, 2016). Supporting this perspective, wives’ relative contributions to housework were associated with a reduced risk of divorce before 1975, but have not been associated with divorce more recently (Killewald, 2016). Yet, although wives’ roles within marriage may be changing, husbands’ full-time employment continues to be strongly associated with a lower risk of divorce (Killewald, 2016).

Education

In the United States, as well as in many other societies, education is negatively associated with divorce risk (Aughinbaugh, Robles, & Sun, 2013; Harkonen & Dronkers, 2006; Lundberg, Pollak, & Stearns, 2016; Martin, 2006; Matysiak, Styrc, & Vignoli, 2014; Prince Cooke & Baxter, 2010; Raymo, Park, Xie, & Yeung, 2015). By age 46 in the United States, roughly 21% of men and 32% of women with a bachelor’s degree will have divorced at some point, compared with 44% of men and 52% of women who have less than a high school diploma (Aughinbaugh et al., 2013). In the 1980s through early 2000s, these education-related differences in U.S. divorce rates were expanding, especially for Whites (J. Kim, 2012; Martin, 2006; Schwartz & Han, 2014), but we know little about these gaps in more recent years. In Europe, a negative association between education and divorce is more common in countries and time periods where divorce itself is relatively more prevalent, perhaps because the social and economic costs of divorce become more affordable (Harkonen & Dronkers, 2006; Matysiak et al., 2014). Among married or cohabiting mothers in the United States, the risk of separation is exceptionally high for those without a college degree compared with their counterparts in many Western European countries. By contrast, college-educated American mothers were no more likely to separate from their spouses or cohabiting partners than similarly educated European women (Musick & Michelmore, 2018).

Age at marriage is strongly negatively associated with divorce risk (Kuperberg, 2014), as previously noted, and education is associated with a lower risk of divorce partly because most young adults delay marriage until after they have completed their schooling. This means that those with more education tend to be older when they get married, but age at marriage accounts for only a small portion of the difference that education makes in risk of divorce; the association between education and income likely matters as well. Education is associated with higher family income, which is negatively associated with financial hardship. The family stress model theorizes that education decreases financial strain and that this reduced financial strain improves the quality of family life (Conger, Conger, & Martin, 2010). Some studies lend empirical support to these theorized connections between a couple’s income, the quality of their marriage, and their risk of divorce (e.g., Boertien & Härkönen, 2018; Raymo, Fukuda, & Iwasawa, 2013), but these same studies show that educational differences in marital quality are small and, similar to age, account for only a small portion of educational gradient in divorce. Moreover, many studies find no association between income and divorce (e.g., Killewald, 2016; Schwartz & Gonalons-Pons, 2016). The Great Recession was associated with declines in divorce, at least in the short run (Amato & Beattie, 2011; Cherlin, Cumberworth, Morgan, & Wimer, 2013; Cohen, 2014). Husbands’ lottery wins are associated with increased marital stability, but not because of an increase in marital satisfaction (Boertien, 2012).

Overall, the argument that increased educational attainment reduces divorce risk by reducing financial hardship and stress, and increasing marital quality, stands on weak empirical ground. In fact, it is not even clear that low earnings increase divorce risk, and unemployment might (temporarily) keep couples together. Boertien and Härkönen (2018) argued that rather than reducing stress and increasing marital quality, education, earnings, and wealth increase the barriers to divorce. They posit that many educated, higher earning couples stay together because leaving the marriage would threaten their wealth and consumption patterns (e.g. home ownership). Future research might further investigate this question by examining the associations between wealth, consumption patterns, and divorce. Lundberg et al. (2016), on the other hand, suggested that intensive investments in children (e.g., time and financial expenditures), which they argue are becoming increasingly more common among the better educated, are facilitated by long-run relationship stability. Again, this remains an important question for future research.

Demography and Predictors of Remarriage and Repartnership

Remarriage rates have fallen by roughly half since the mid-20th century, with the steepest declines observed since the 1980s (Schweizer, 2019). Between 2008 and 2016 alone, remarriage rates dropped from 33 to 28 remarriages per 1,000 divorced or widowed adults (Payne, 2018). Despite declining remarriage rates, increases in the population available for remarriage (i.e., individuals with a first marriage that ended in widowhood or divorce) mean that the share of people who have ever remarried by age 40 or 50 has varied only modestly across mid-century birth cohorts (authors’ tabulations). In fact, the share of marriages that are remarriages for one or both partners has actually grown in recent decades, driven in part by concurrent declines in first marriage. In 1960, only 13% of married adults were in their second or higher order marriage, but by 2013 this number had risen to 23% (Livingston, 2014). In recent years, about 40% of all new marriages were remarriages for one or both spouses (Lewis & Kreider, 2015; Livingston, 2014), suggesting to some that Americans remain strongly attached to marriage even when their first (or second) marriage dissolves (Cherlin, 2009).

As first marriage and divorce rates have recently dropped, however, the share of individuals available for remarriage is also beginning to decline. In 1996, 32.6% of women and 26.8% of men ages 35 to 39 were eligible for remarriage, meaning they had been married but were no longer in their first marriage (Kreider & Fields, 2002). In 2008 to 2012, these numbers fell to only 28.5% of women and 22.0% of men (Lewis & Kreider, 2015). Overall, declines in both remarriage rates and the proportion of adults eligible for remarriage portend a gradual decline in the proportion of adults who experience remarriage, as more recent cohorts gradually replace the high-divorce Baby Boom generation. The proportion ever remarried has not yet declined substantially, however, because the population is aging and older people are more likely to be remarried than younger people. Nonetheless, the trends suggest that shrinking proportions of Americans will remarry in future cohorts, although robust rates of cohabitation suggest that postmarital partnerships may remain common.

Unfortunately, we know relatively little about recent trends in postdivorce repartnering more broadly defined through either marriage or cohabitation, and thus the extent to which nonmarital cohabitation will compensate for decline in remarriage remains unclear. In 2002, more than two thirds of remarriages to women younger than age 45 were preceded by cohabitation (Teachman, 2008), and among women who divorced when older than age 50, slightly less than half of those who repartnered between 1998 and 2014 did so through cohabitation (Brown, Lin, Hammersmith, & Wright, 2019). We lack information on recent patterns of repartnering more generally, however. Age restrictions on the National Survey of Family Growth and the Health and Retirement Surveys make it difficult to analyze repartnering patterns across all ages, but an analysis that combined these two data sources could provide excellent basic information useful for understanding family change.

Although remarriage and repartnering remain relatively understudied, as also noted in the 2010 decade-in-review article on this topic (Sweeney, 2010), some recent research considers differences in who is likely to remarry or repartner. In the United States, men are more likely than women to remarry, although this gender difference seems to be lessening over time (Livingston, 2014; Payne, 2018). These gender differences in remarriage also vary with age: The male advantage in remarriage tends to be largest at older ages (Livingston, 2014; H. Wu, 2017), and the male advantage at midlife and beyond is even greater when we define repartnering more broadly to include either marriage or cohabitation (Brown et al., 2019).

Looking beyond the United States, where studies tend to investigate repartnering broadly to include marriage or cohabitation, a large body of work indicates that mothers in multiple countries have lower rates of repartnering than fathers (e.g., Di Nallo, 2019; Gałçzewska, Perelli-Harris, & Berrington, 2017; Gray, 2015; Ivanova, Kalmijn, & Uunk, 2013). However, some evidence indicates that parenthood is only associated with reduced repartnering when parents live with their children (e.g., Beaujouan, 2012; Vanassche, Corijn, Matthijs, & Swicegood, 2015). Studies from multiple countries indicate that childless men and women have similar likelihoods of repartnering (e.g., Beaujouan, 2012; Di Nallo, 2019; Gray, 2015; Ivanova et al., 2013). Leveraging a change in custody law in Belgium, Schnor, Pasteels, and Van Bavel (2017) demonstrated that Flemish mothers with sole custody were less likely to repartner than those with shared custody, even though mothers with sole custody also tended to have other characteristics (e.g., high family orientation) associated with higher rates of remarriage. A recent analysis of maternal repartnering in the United Kingdom finds that fathers’ involvement in parenting has no effect on mothers’ repartnering for those who were married or cohabiting at the time of the birth (Berger, Panico, & Solaz, 2018).

A number of studies also consider how remarriage patterns vary across groups defined by education, race, and ethnicity. Education is associated with slower transitions into first marriage, largely because of marriage delay associated with school enrollment, but education is a weak predictor of remarriage, especially among women (Shafer & James, 2013). Among those who divorce in early adulthood, educational differentials in remarriage may be growing such that the least educated are the slowest to remarry (McNamee & Raley, 2011). The least-educated women are the quickest to repartner (McNamee & Raley, 2011), however, perhaps because their limited economic resources make forming a new union especially attractive (Shafer & James, 2013). Education may also shape the remarriage and repartnering process in other ways as well, such as affecting the preferred characteristics of new partners. In Belgium, for example, fathers’ education levels are associated with postdivorce fathering arrangements, such as the likelihood of repartnering with a childless woman or becoming a stepfather (Schnor, Vanassche, & Van Bavel, 2017). In the United States, non-Hispanic White women remarry and repartner more quickly than Latina and Black women (McNamee & Raley, 2011), and patterns of racial and ethnic intermarriage that characterize first marriages are even stronger in remarriage. That is, White–Black intermarriages are even less common, but White–Asian and White–Hispanic marriages are more common in remarriage than in first marriages (Choi & Tienda, 2017). The greater tendency of Whites to remarry Asians and Hispanics than Whites may be related to the more constrained marriage markets for those who are previously married (Qian & Lichter, 2018).

Consequences of Divorce and Repartnering

Adult Well-Being

Divorce is strongly associated with adult well-being, but this association varies by different types of well-being, gender, life course stage, and social context. These variations reveal the importance of social context in shaping the consequences of divorce. Although divorce’s near-term economic consequences are not as large as they once were, divorce is still associated with substantial short-term declines in economic resources for mothers and children (Osborne, Berger, & Magnuson, 2012; Tach & Eads, 2015). In the 1980s, mothers’ household incomes were 42% lower 1 year after divorce; by the 2000s this number had fallen to 33% (Tach & Eads, 2015). Increases in women’s labor force participation and earnings leading up to and following divorce are responsible for both a predivorce increase and postdivorce recovery in household income (Tamborini, Couch, & Reznik, 2015).

Divorce can also have long-term effects on economic well-being. Even though women’s labor force participation rates and earnings increase for many years following a divorce (Tamborini et al., 2015), divorce is still associated with lower wealth accumulation among older women (Addo & Lichter, 2013) and with elevated poverty rates. Among women who divorced after age 50 and did not repartner, 27% are in poverty, compared with 12% of men of the same relationship status (Lin, Brown, & Hammersmith, 2017). Men’s advantaged position in the labor market compared with women means that for women, repartnering can be a more effective route to economic recovery than increased labor force participation, at least in Europe and especially for mothers (Jansen, Mortelmans, & Snoeckx, 2009). In the United States, remarriage can have the added benefit of providing access to the second spouse’s Social Security benefits, provided the remarriage lasts long enough (Lin et al., 2017).

The comparatively severe economic consequences of divorce for women stem from the fact that men continue to be the primary breadwinners in most families. Women are more often consistently employed than they used to be, and the gap between women’s and men’s wages has decreased (in part because of increases in women’s education). Despite these gains, men still outearn women, and fathers especially outearn mothers.

Divorce is also associated with health and perceived well-being. A common framework for understanding this association separates selection from two types of causation: crisis and resources. Divorce might contribute to declines in well-being because of short-term turmoil surrounding the crisis of marital disruption, and it might also affect long-term health outcomes by depriving individuals of material, psychic, and social resources that derive from marriage. The past decade has produced substantial evidence to support the crisis model, particularly regarding mental health. Many high-quality panel studies using multiple interviews to estimate fixed effects models identify short-term negative consequences for mental health and life satisfaction following divorce, especially for parents (Hewitt, Turrell, & Giskes, 2012; Kalmijn, 2017; Leopold, 2018; Osborne et al., 2012; Williams & Dunne-Bryant, 2006). A fixed effects approach can adjust for unobserved factors that are fixed over time, such as a preexisting predisposition toward depression. Consistent with a crisis perspective, these studies generally show recovery over time. Incidentally, however, a study of Swedish men found that remarriage was not associated with reduced depression (Hiyoshi, Fall, Netuveli, & Montgomery, 2015).

It is less clear whether divorce contributes to short-term declines in physical health. Mortality rates increase immediately after divorce (Rendall, Weden, Favreault, & Waldron, 2011), but mortality does not always result from poor physical health, for example in cases of accident or suicide. It is also difficult to rule out selection factors because mortality does not lend itself to a fixed effects approach. A fixed effects analysis of Norwegian administrative records detects a steep, short-term increase in work sick days in the year of divorce, with some recovery over time for men and childless women, but not for mothers (Dahl, Hansen, & Vignes, 2015). It is not clear that this reflects poorer physical health because people may take sick days for mental health reasons as much as physical health. Moreover, divorce sometimes leads to short-term improvements in physical health (Leopold, 2018; Monden & Uunk, 2013; Williams & Umberson, 2004).

The crisis model seems to apply better to mental health and general well-being than to physical health (Kalmijn, 2017). It may be that the association between divorce and poor physical health (Liu & Umberson, 2008) is due entirely to selection, or it may be that marriage provides resources that benefit health in ways that have not been easy to detect in most fixed effects analyses. The marital resource model predicts that poor health develops over time as the benefits of marriage or the costs of singlehood accumulate (Hughes & Waite, 2009). Most fixed effects analyses may be better at identifying large, short-term swings in health than at picking up slowly developing processes that take decades to result in measurable health declines. Recent research using marital biographies links divorce to poorer health trajectories later in life even for those who remarried, although those who remarried were healthier than those who did not remarry (Hughes & Waite, 2009; O’Flaherty, Baxter, Haynes, & Turrell, 2016). Data limitations make it difficult to assess the long-term consequences of marital biographies using rigorous controls for selection factors. Supporting a marital resource interpretation, however, a recent fixed effects analysis of survey and administrative data in the United States finds that divorce has long-term effects on work disability for men who do not remarry. These negative consequences did not appear until 20 years after divorce (Couch, Tamborini, & Reznik, 2015). These findings suggest that it is premature to rule out the marital resource model.

Contextual factors seem to play an important role in conditioning the effects of divorce and repartnering on health. For example, in the United States the negative association between divorce and self-reported health increased from the early 1970s to the early 2000s (Liu & Umberson, 2008), and as mentioned previously, the negative health consequences of divorce appear to be greater for parents (Dahl et al., 2015; Kamp Dush, 2013; Williams & Dunne-Bryant, 2006). Some also argue that the negative effects of divorce may be weaker in social contexts that have stronger welfare systems or that place less emphasis on marriage than the United States (Kalmijn, 2017).

Offspring Well-Being and Attainment

During the past decade, scholars have continued to move away from studying divorce and remarriage as isolated events in children’s lives and toward considering cumulative histories of instability and diversity in children’s family and living arrangements. The overall level of children’s family instability in the United States grew only modestly in recent years, as measured by mothers’ transitions into and out of cohabiting and marital relationships (Brown, Stykes, & Manning, 2016; Rackin & Gibson-Davis, 2018). Growth in family instability occurred primarily among Black children and those with less-educated mothers, further widening existing family instability gaps by race and mother’s education (Brown et al., 2016; Rackin & Gibson-Davis, 2018). Cohabitation plays a central role in the instability experienced by contemporary U.S. children, increasing estimates of children’s family instability by 80% for Whites, 50% for Hispanics, and nearly doubling estimates for Black children (Brown et al., 2016). Estimates of instability in children’s family environments increase further when we look beyond mothers’ partnerships to consider broader sources of instability in children’s households, such as the arrival or departure of nonparents. These increases are particularly strong for Black and Hispanic youth and those with less-educated mothers (Perkins, 2017; Raley, Weiss, Reynolds, & Cavanagh, 2019).

An extensive body of work makes it clear that the more family structure transitions children face, the lower their level of well-being on average (for recent reviews, see Cavanagh & Fomby, 2019; Hadfield, Amos, Ungar, Gosselin, & Ganong, 2018). This pattern holds across multiple domains of well-being, including problem behavior (e.g., Cooper, Osborne, Beck, & McLanahan, 2011; Fomby & Mollborn, 2017; Fomby & Sennott, 2013; Mitchell et al., 2015), health (Bzostek & Beck, 2011; Smith, Crosnoe, & Cavanagh, 2017), and emotional well-being (e.g., Bzostek & Berger, 2017; Lee & McLanahan, 2015), as well as socioeconomic attainment and relationship stability in adulthood (e.g., Amato & Patterson, 2017; Bloome, 2017; Fomby, 2013; Fomby & Bosick, 2013). A range of explanations have been offered for the association between family instability and reduced offspring well-being—generally focusing on changes in economic resources, parenting, and emotional stress—but empirical support for such mechanisms remains surprisingly modest (Cavanagh & Fomby, 2019). Moreover, the extent to which preexisting selectivity in the unmeasured characteristics of parents and children who experience family instability can explain these associations remains an important outstanding question.

A growing body of evidence points to heterogeneity in the effects of family instability on offspring well-being. Instability seems to have particularly negative consequences for the well-being of White youth and those who are socioeconomically advantaged (Bernardi & Radl, 2014; Cavanagh & Fomby, 2019; Fomby & Cherlin, 2007; Perkins, 2019; Ryan, Claessens, & Markowitz, 2015; see also an earlier landmark study by L. L.Wu & Martinson, 1993) or born to married parents (Bzostek & Beck, 2011; Bzostek & Berger, 2017; Ryan & Claessens, 2013). The reasons for these differences are not yet well understood but may reflect the relatively lower likelihood of experiencing a transition among these groups, and the fact that events that are unanticipated may tend to have particularly deleterious consequences, or the greater selectivity in background characteristics of parents and children from those groups who experience instability (Bernardi & Radl, 2014; Cavanagh & Fomby, 2019; Turney, 2017). Other possible explanations include group differences in the average economic consequences of divorce or the nature of relationships with nonresident biological parents.

The cumulative instability approach to studying children’s family structure environments has an appealing conceptual and empirical parsimony. It is relatively straightforward to create measures of number of transitions that are theoretically grounded in an “instability and change” perspective (e.g., L. L.Wu & Martinson, 1993). Another strength of this approach is its broad focus on instability and change in general, which directs attention beyond a parent’s marriage or divorce to include additional sources of instability in children’s environments.

Yet this approach also has important limitations. Studies often fail to distinguish transitions that involve marriage from those that do not, implicitly assuming that transitions into and out of unions affect children’s well-being similarly. This is inconsistent with a large literature indicating that a parent’s marriage is associated with positive outcomes for children (e.g., Ribar, 2015)—although any causal underpinnings of this association remain an active topic of debate—and with evidence that children’s access to economic and parenting resources varies depending on the nature of the transition (e.g., Osborne et al., 2012). In addition, as noted previously, the associations between instability and poorer well-being may vary by transition type (see also Lee & McLanahan, 2015; Ryan et al., 2015).

Greater attention to types of transitions and heterogeneity in family instability’s effects on offspring well-being and hypothesized intervening mechanisms (e.g., economic and parenting resources)—either across children from different groups or across domains of well-being—may provide important clues about underlying causal pathways. We also encourage scholars to consider instability in children’s larger family environments, including fathers and siblings, and to pay greater attention to broader complexity in family and household composition, including the extent to which children spend time living in multiple households. Qualitative research might be particularly helpful in understanding variability in these processes by child and family characteristics. This literature might also benefit from revisiting the family science research reviewed in Buehler (2020) on risks, strengths, and resiliency processes in families and their implications for child development to identify the distinct mechanisms that might connect parental relationship formation and dissolutions to child well-being.

In an effort to better adjust for the potentially selective characteristics of parents and children who experience divorce, recent research has also applied more rigorous statistical methods to study the association between divorce and children’s well-being. These studies tend to confirm prior findings that divorce is associated with reduced child well-being, but they often estimate a weaker negative impact than more traditional observational studies (McLanahan, Tach, & Schneider, 2013). For example, using matching methods and growth curve models, H. S. Kim (2011) found that divorce negatively affects children’s math test scores and interpersonal social skills. Brand, Moore, Song, and Xie (2019) use propensity-score analysis to show that divorce reduces educational attainment most for children whose parents were least likely to divorce. (But see Grätz [2015], which finds a different pattern in Germany using a sibling-based fixed effects model.) Taken together, these studies increase our confidence that divorce tends to have a causal negative effect on well-being. However, all such studies rely on assumptions that are not directly testable, particularly assumptions about unobserved determinants of children’s well-being that correlate with family structure histories. It is possible, however, to conduct sensitivity analyses to gauge the likely robustness of estimates of varying assumptions about the unobserved variables, and this sort of work could advance our understanding of the factors that shape divorce’s impact on kids. Better understanding the nature of preexisting selectivity in the characteristics of adults and children who experience family instability remains essential for identifying causal effects of family transitions on offspring well-being.

Finally, a growing number of studies consider the relationship between postdivorce custody arrangements (e.g., joint vs. sole custody) and child well-being. Although the results tend to vary across studies, some evidence points to modestly improved outcomes for children associated with joint custody arrangements (for recent reviews, see Baude, Pearson, & Drapeau, 2016; Steinbach, 2019). As shared custody arrangements are increasingly common for children after divorce (e.g., Cancian, Meyer, Brown, & Cook, 2014), this is an important area for future work. Broadly, additional attention is needed to the complex and dynamic family and household environments of children, including the fact that many children spend time in multiple households.

Stepfamilies

Stepfamilies result from remarriage or repartnering when at least one partner has a child from a prior relationship, which leads to tremendous variability and complexity in specific stepfamily structures. The Survey of Income and Program Participation provides data about stepfamilies from children’s perspective: In 2009, 7.5% of U.S. children younger than age 18 (5.6 million children in total) lived with a cohabiting or married stepparent. This figure is up only slightly since 1991, when 7.0% of children lived with a stepparent (Kreider & Ellis, 2011). The term blended family refers to households that include stepparent, stepsibling, or half-sibling relationships. In 2009, roughly 1.7% of children younger than age 18 lived with at least one stepsibling—that is, the biological child of a stepparent. Roughly 10.8% lived with at least one half-sibling—the child of a biological parent and another adult (Kreider & Ellis, 2011). Estimates suggest that 17% to 19% of same-sex couple households contain children (Gates & Cooke, 2011), but data limitations make it impossible to tell what proportion are children from a prior union (Ganong & Coleman, 2018).

U.S. society continues to lack clear norms about the expectations and obligations associated with stepfamily relationships (Cherlin, 1978; Ganong & Coleman, 2017). Perhaps because of this uncertainty, people use a variety of terms to refer to step-relationships, for example, referring to a stepfather as a “my mother’s husband.” Some evidence suggests that the likelihood of using the “step” label varies according to the nature of the child’s relationship with different family members and may also vary over time within families. For example, adolescents who report feeling close to their biological mothers are more likely to use the stepfather label, whereas those who report closeness with their nonresident fathers are less likely to use the term stepfather (Thorsen & King, 2016). Importantly, how adolescents label their stepfather appears to be more strongly related to relationships with biological parents than with the stepfather himself, and step-grandchildren are more likely to call a step-grandparent a “grandparent” when the step-grandparent fulfills traditional grandparent role expectations (Chapman, Coleman, & Ganong, 2016). In addition to providing important information about the nature of stepfamily relationships, variability in labels used to refer to step-kin presents a challenge to family scholars attempting to enumerate or otherwise study step-relationships, as stepkin relationships may be overlooked when only identified through the use of step-labels (Seltzer, 2019).

Growing attention over the past decade has focused on the nature of ties between step-kin. These studies confirm prior findings that step-relationships are generally not as close as biological relationships, but also they document considerable variability in relationships between stepparents and stepchildren. For example, the quality of relationships between children and their biological mothers, and between mothers and stepfathers, are associated with the quality of the children’s relationships with their stepfathers and with adolescents’ perceptions of belonging in their stepfamilies (Jensen & Howard, 2015; Jensen & Shafer, 2013; King, Amato, & Lindstrom, 2015; King, Boyd, & Thorsen, 2015; King, Thorsen, & Amato, 2014). Stepchild–stepparent relationships are also influenced by factors such as the stepchildren’s ages, stepchildren’s and stepparents’ genders, relationship duration, and the nature of custody arrangements (Becker, Salzburger, Lois, & Nauck, 2013; Ganong, Coleman, & Jamison, 2011; Kalmijn, 2013).

Research during the past decade has increasingly considered dynamics involving older parents and their adult stepchildren. Residential proximity tends to be associated with an increased likelihood that parents and their adult children will routinely exchange time and money. Adult stepchildren and stepmothers are less likely than their biologically related counterparts to live together or near each other (Seltzer, Yahirun, & Bianchi, 2013). Just having step-kin increases the availability of kin, but after taking this general increase into account, scholars find that households with step-relationships are less likely than other households to participate in intergenerational transfers of resources, especially time (Wiemers, Seltzer, Schoeni, Hotz, & Bianchi, 2019). Dutch stepparents at mid-life and later life, however, were more likely in 2009 than in 1992 to consider their stepchildren part of their personal network (Suanet, van der Pas, & van Tilburg, 2013), suggesting that step-relationships may be becoming more “institutionalized” over time. Little work has investigated the extent to which stepparent–stepchild ties endure beyond the end of a biological parent’s union, although Noël-Miller (2013) finds contact between stepparents and adult stepchildren to be largely conditional on the continuing marital tie with the stepchild’s biological parent.

The recent rise in “gray divorce” may have contributed to a growing interest in later life stepfamily relationships. Stepfamilies formed in later life may face many unique challenges, including issues of inheritance, elder care, and complex long-term networks of previous relationships and kin ties (Papernow, 2018). Among married adults older than age 50, the share in a remarriage (rather than a first marriage) increased from 19% in 1980% to 30% in 2015 (Lin et al., 2018), and among all coresidential couples in which one partner was older than age 50 in recent years, a remarkable 41% had at least one child from a previous relationship. Of these later life stepfamilies, 87% involved married partners, whereas the remaining 13% involved unmarried partners who were cohabiting (Lin et al., 2018). Stepfamilies, especially those in which the parents are cohabiting but unmarried, tend to be less economically and socially advantaged than married families without children from previous relationships. After adjusting for these compositional differences, relationship quality tends to be similar across family types (Lin et al., 2018).

Complexity in stepfamily structures can also be multigenerational and attention to step-grandparenthood has also increased in the past decade. There are two paths to step-grandparenthood: an individual may partner with someone who has a grandchild from a prior relationship or may have a child who partners with someone who has a child from a prior relationship. The likelihood of having any step-grandchildren has increased across U.S. birth cohorts and is more common among those without a college education and among African Americans (Yahirun, Park, & Seltzer, 2018). Although evidence points to generally weaker ties between step-grandparents and step-grandchildren than biological grandparents and grandchildren, the nature of step-grandparent–step-grandchild relationships varies considerably (Chapman et al., 2016; Ganong & Coleman, 2018).

Conclusions and Directions for Future Research

What have we learned about changing family life from the past decade’s scholarship on divorce, repartnering, and stepfamilies? We began this review by observing that overall divorce rates are declining, particularly for today’s young adults. This portends an eventual decline in the percentage of marriages that end in divorce (Cohen, 2019). For at least two reasons, however, we suspect that couple relationships are not returning to former levels of stability.

First, marriage tends to be more common among economically advantaged couples, and some of the greater stability observed among marriages relative to other couple relationships relates to this preexisting selectivity (Tach & Edin, 2013). Second, longer run trends in relationship stability—broadly defined through marriage or cohabitation—remain less clear (Cherlin, 2017). It is not possible to understand family change without considering the growth in cohabitation, which differs from marriage in many ways. Cohabitating unions are becoming more stable over time (Cohen & Manning, 2010) but remain distinctly less durable than marriage, at least in the United States (Musick & Michelmore, 2018). A shift from marriage to cohabitation implies increases in relationship instability.

For decades, scholars of social change have pointed to high remarriage rates as evidence of Americans’ strong attachment to the institution of marriage (Cherlin, 2009). Remarriage rates continue to be high but are now declining. Is this a sign that Americans’ attachment to marriage is weakening? First marriage patterns suggest not (yet). Young adults are waiting longer to marry, but declines in the proportion of women who have ever married by age 40 have so far been small, except among Black women (Raley et al., 2015). Moreover, recent efforts to same-sex marriage suggests both the continued symbolic importance and institutionally supported benefits of marriage. Declines in divorce may reinstate marriage as a lifelong commitment among those who can marry. This status might keep marriage desirable, even as it shrinks the proportion of the population that gets married. Even if marriage remains a strong social institution, all signs indicate that access to stable marriage will be increasingly limited to those of relatively high socioeconomic status (e.g., Cherlin, 2004; Cohen, 2019), and similar to first marriage, remarriage remains more common among the economically advantaged.

Findings from the past decade suggest that marriage has changed but continues to be gendered. The economic consequences of divorce continue to be much greater for women than men (Kalmijn, 2015; Tach & Eads, 2015), and the consequences of cohabitation dissolution for women’s economic well-being are increasing (Tach & Eads, 2015). Women’s earnings, even when they are higher than their husbands’, are no longer associated with an increased risk of divorce (Schwartz & Gonalons-Pons, 2016). At the same time, husbands’ full-time employment remains essential to stable marriages (Killewald, 2016).

The past decade’s research has also provided new insights into the consequences of divorce and family instability for adults and children. We have particularly strong evidence that divorce has negative effects on mental health of adults—but also that these effects are temporary. Meanwhile, the lack of findings about physical health may reflect studies’ focus on short-term consequences: The physical consequences of divorce may develop slowly.

During the past decade, considerable effort has also been made to better understand the consequences of divorce and family instability for children, more often taking a family life course perspective with a focus on cumulative histories of family change and including more rigorous designs that adjust to varying extents for selectivity in preexisting characteristics of parents and families. Taken together, the findings suggest that while some of the association between family instability and child well-being is likely spurious, the case that divorce and family instability reduce children’s well-being is strong. At the same time, the magnitude of these consequences can vary considerably across individuals and groups. Findings indicating that the consequences of family instability tend to be largest for youth who are least likely to experience it (e.g., socioeconomically advantaged youth and youth born to married parents) are particularly compelling and warrant further research.

Finally, our understanding of changes in family stability and its consequences would benefit from more attention to the changing legal landscape of family life (e.g., the legalization of same-sex marriage) and more descriptive and qualitative research that addresses the complexity of contemporary partnerships and family configurations, including children’s coresidence with single or repartnered fathers, siblings, and the extent to which children may reside in multiple households at any given time. Likewise, recent scholarship highlights the considerable variability in stepfamily forms and in relationships between stepparents and stepchildren, including as stepparents age into later life. Attention to later life families offers many opportunities for advancing family knowledge, including our understanding of longer term consequences of instability and the extent to which relationships between children and stepparents endure after marital ties end.

Greater attention to nonmarital relationships, cohabitation in particular, has been a strength of the past decade’s research on divorce, repartnering, and stepfamilies. At the same time, marriage remains important in its own right. Whereas at the time of Bumpass’s address to the Population Association of America some were concerned that rises in divorce foretold the diminishing relevance of marriage, marriage endures as a core social institution. Yet marriage (and remarriage) is changing as it becomes less common and increasingly selective of socioeconomically advantaged groups. This may have spurred modest declines in divorce rates while levels of relationship instability (including cohabitation) remain high. Nonetheless, among those who marry, divorce also persists as both a stratified and stratifying life event, deepening existing social inequalities by gender and socioeconomic status. We encourage future researchers to investigate variation across individuals and contexts in these associations to better understand how to reduce these negative impacts.