The Demography of Families: A Review of Patterns and Change

Abstract

The authors review demographic trends and research on families in the United States, with a special focus on the past decade. They consider the following several topics: (a) marriage and remarriage, (b) divorce, (c) cohabitation, (d) fertility, (e) same-gender unions, (f) immigrant families, and (g) children’s living arrangements. Throughout, the authors review both overall trends and patterns as well as those by social class and race–ethnicity. The authors discuss major strands of recent research, emphasizing emerging themes and promising directions. They close with a summary of central patterns and trends and conclude that recent trends are not as uniform as they tended to be in earlier decades, making the description of family change increasingly complex.

We say the home is in transition. So it is. It is moving away from what it was toward something that it is to be....it is not an institution usually at an equilibrium….It is always on the move….The home problem, therefore, is not one that we can finally solve. (Groves, 1925, pp. 228–229)

A family life educator born in 1877, Ernest Groves observed that the family as an institution was in transition and never static, words that were just as apt then as they are today. Sixty-five years later, family demographer Larry Bumpass asked “What’s happening to the family?” (Bumpass, 1990). The general public, researchers, and the media remain keenly interested in this question. Reporting results from a new survey at that time, the 1987 to 1988 National Survey of Families and Households, Bumpass discussed shifts in family life. He wrote of the rise in divorce, the growth of cohabitation, and increases in nonmarital childbearing. Twenty years after Bumpass, in a 2010 review of family demographic trends in the United States, Cherlin (2010) concluded that we are experiencing a growing disconnect between families and households. That is, given trends in cohabitation, divorce, repartnering, and nonmarital childbearing, families have become less likely to reside in the same household. The most obvious example is that many fathers and children live apart. Despite a rapidly changing social landscape in the past decade, particularly with respect to the aftermath of the Great Recession, these themes have remained central to the study of the family.

There are at least two types of studies in the field of family demography. One is the careful tracking and measurement of families at either a given time point or trends over time. This includes the documentation of family structure; family transitions such as marriage, marital dissolution, and childbearing; and socioeconomic characteristics of families (e.g., income and education levels). This documentation motive is important, providing solid knowledge on changes in the family landscape. The second is a focus on the social, cultural, and economic correlates, precursors, and consequences of family structure and transitions. Although a large portion of this review is geared toward the documentation of recent changes in families, we also incorporate recent work focused on describing and understanding the precursors and consequences of family structure and transitions.

Conceptually, a subtheme of our review draws on the intersections between family patterns and inequality. The “diverging destinies” framework, articulated by McLanahan (2004), indicates that families are increasingly dissimilar across social class lines, often proxied by educational attainment. Throughout our review, we use various terms such as social class, economic (dis)advantage, highly or less educated, and so on. The economically advantaged experience more stable family forms such as marriage, whereas the less privileged undergo more family instability. We broaden the diverging destinies framework to also consider family variation by race and ethnicity, another aspect of inequality and a topic of longstanding interest and importance in family research.

Our review is organized around trends and patterns in basic family demographic measures and events: marriage and remarriage, divorce, cohabitation, fertility, same-gender unions, immigrant families, and children’s living arrangements. Although 50 years ago research in family demography often fit neatly into one of these categories, a hallmark of more current research is that it tends to crosscut categories (e.g., the implications of changes in marriage for children’s living arrangements, the dissolution of same-gender marital and cohabiting unions), making the placement of recent research into categories not clear cut. Because of the broad scope of this review, we largely restrict ourselves to the United States. Given the varying nature of each topic, not all sections adhere to an identical outline, but are organized roughly as follows: We begin with a discussion of trends and basic patterns, next turning to variation in trends and patterns by social class and race–ethnicity; we also highlight major research themes and explanations for changes in families with a special focus on recent research and new directions.

We aim to make this review accessible to a variety of audiences. One audience consists of scholars interested in families (or in one or more of our focal areas) who seek a summary of recent trends and research. We also hope this review is helpful to another audience: those wanting a succinct grounding in basic trends and research findings to inform qualitative scholarship.

Marriage and Remarriage

Marriage is highly regarded and a goal for most men and women in the United States. The majority of young people see marriage in their futures, and this has remained virtually unchanged for the past several decades. Among high school seniors in 1976, 74% expected to marry in the future. In 2014, 79% reported that they expected to marry at some point (Anderson, 2016).

Despite high levels of aspirations for marriage, there are trends signaling a retreat from marriage. Figure 1 shows that the percentage of adults living with a spouse has been falling steadily for decades. In 1967, 70% of adults were living with a spouse compared with only 51% in 2018, and declines over the past decade matched those in the 1990s and 2000s (U.S. Census Bureau, 2018a).

FIGURE 1.

Median Age at First Marriage (1890–2018) and Percent of Adults Living With Spouse (1967–2018).

Note: Dashed lines between years are linearly interpolated between data points.

Sources: Median age at first marriage: U.S. Census Bureau (2018b). Percent of adults aged 18 years and older living with spouse: U.S. Census Bureau (2018a).

Part of the decline in the percentage of married adults is due to large increases in age at first marriage. As shown in Figure 1, men and women are entering first marriages substantially later in life. Even the past decade has seen considerable change. In 2010, the median age at marriage was 28.2 for men and 26.1 for women. Just a few years later, in 2018, the analogous figures were 29.8 for men and 27.8 for women (U.S. Census Bureau, 2018b, Table MS-2). These numbers are historically unprecedented, at least since solid data have been collected that allow for estimates. Although some of the decline in the percentage of adults living with a spouse is due to delays in marriage, there have also been declines in the percent who ever marry. Estimates using data from 1988 suggest that 87% of women married at some point in their lives compared with 84% in 2000 to 2005 and 80% in 2005 to 2010 (Schoen, 2016).

Population-level marriage trends obscure substantial differences by race and ethnicity and by education. One important research emphasis over the past few decades has been a growing Black–White gap in marriage in which African Americans have lower marriage rates than Whites (Lichter, McLaughlin, Kephart, & Landry, 1992; Raley, Sweeney, & Wondra, 2015; Wilson, 1987). In 1970, among 40- to 44-year-old women, 95% of White women had married as had 92% of African American women. In contrast, data from 2012 indicate that fewer than two thirds of 40- to 44-year-old African American women had ever married compared with 88% of White women (Raley et al., 2015). Generally, race–ethnic differences in the percentage of women who have ever married have grown. For example, Hispanic, American Indian/Native Alaskan, and African American women all have increasingly different percentages of women marrying when compared with White and Asian/Pacific Islander women and compared with one another (Raley et al., 2015).

Trends in the proportion of women ever marrying by education align with the observation of McLanahan (2004) that the economically advantaged are increasingly likely to experience stable family forms such as marriage relative to the less advantaged. Considering education alone, in 2016, 27% of women without a high school education were currently married compared with nearly 60% of those with a college degree or more. Women with high school degrees or some college fell in between, with 45% currently married (Allred, 2018). Among White women, college graduates were once less likely to have married than women with less education, but this gap gradually closed since 1960 due to increases in the percentage of White college graduate women marrying relative to women with less education. Black college graduate women have also become more likely to ever marry relative to their less-educated counterparts (Isen & Stevenson, 2011).

Overall, trends by race–ethnicity and education since the 2000s represent a continuation of trends begun decades ago rather than a distinct new period of change (Isen & Stevenson, 2011; Raley et al., 2015). Demographers and other social scientists in both the past decade as well as in prior decades have emphasized changes in the economy as a major reason for declines in marriage (Autor, Dorn, & Hanson, 2018; Cherlin, 2014; Cherlin, Ribar, & Yasutake, 2016; Schneider, Harknett, & Stimpson, 2018). People appear to tie marriage “readiness” to having a comfortable income, little to no debt, and a secure job. That is, marriage has come to be seen as a luxury good, a step to be taken after one has achieved a comfortable level of economic stability (Addo, 2014; Cherlin, 2004; Gibson-Davis, Gassman-Pines, & Lehrman, 2018; Huang, Smock, Manning, & Bergstrom-Lynch, 2011; Smock, Manning, & Porter, 2005). Those markers of economic stability have become less available to those without college degrees as the number of well-paid, skilled manufacturing jobs has declined. Scores of studies have shown that the least educated, typically those with the worst economic prospects, are less likely to marry than the most educated (e.g., Addo, 2014; Gibson-Davis et al., 2018; Ishizuka, 2018; Kuo & Raley, 2016; Lundberg, Pollak, & Stearns, 2016; Oppenheimer, 2003; Schneider et al., 2018; Smock et al., 2005; Smock & Manning, 1997; Xie, Raymo, Goyette, & Thornton, 2003). In addition to economic factors, marriage market availability and the increase in mass incarceration have also been leading explanations for trends and variation in marriage patterns (e.g., Charles & Luoh, 2010; Cohen & Pepin, 2018; Lichter et al., 1992; Raley et al., 2015; Schneider et al., 2018; Wilson, 1987; but see Lopoo & Western, 2005).

Given the relationship between economic well-being and marriage, one might have expected the Great Recession (which officially began in 2007 and ended in 2009 although unemployment rates did not reach prerecession levels until 2015) to have reduced marriage (Schneider, 2017). However, marriage rates continued their previous downward trajectory over this period with little evidence of perturbation; that is, marriage rates declined at the same rate as before the recession when we would have expected the recession to have led to a steeper decline (Cherlin, Cumberworth, Morgan, & Wimer, 2013; Morgan, Cumberworth, & Wimer, 2011). Research using state-level variation in the severity of the recession also finds little evidence of a recession effect, with the possible exception of small negative effects on the likelihood that women of low socioeconomic status marry (Morgan et al., 2011; Schneider & Hastings, 2015). One possible reason for the recession’s lack of apparent effect is that it is not only economics that matter for marriage. A key explanatory framework used to understand temporal changes in family patterns is the second demographic transition. The second demographic transition emphasizes that ideational factors (e.g., norms, attitudes) alongside economic changes are vital for understanding why family life is decreasingly centered around marriage and supportive of other family forms (Lesthaeghe, 2014; Lesthaeghe & Neidert, 2006; Lesthaeghe & van de Kaa, 1986).

The future may well be characterized by a continued retreat from marriage. Consider the percentage marrying by age 40 for Generation X (i.e., those born between 1966 and 1972; Martin, Astone, & Peters, 2014). For Generation X women, 82% married by age 40. Martin and colleagues projected that the percent married will decrease to 75% and 70% for early (born 1980) and late (born 1990) Millennials, respectively. The study also projects growing divergence by education in the percentage who marry with slow but steady declines in marriage for men and women with college degrees and faster declines for those without college degrees. As the authors state, “The singles are coming” (Martin et al., 2014, p. 4).

Although the percentage of men and women currently and ever married has declined, particularly among the less advantaged, remarriage is common because divorce remains common. About 34% of ever-married men aged 60 to 69 in 2008 to 2012 had had two or more marriages as had 30% of ever-married women (Lewis & Kreider, 2015). There is also evidence of racial-ethnic and educational variation. McNamee and Raley (2011) estimate the number of years it takes for 25% of women in various subgroups to remarry using data from the mid-2000s. Black and Latina (both native and foreign born) women take longer to remarry than White women. By 3.8 years after marital dissolution, 25% of White women have remarried. For both native- and foreign-born Latinas, the comparable number is slightly more than 5 years. For Black women, the data do not allow for an estimate: One quarter is never reached during the observation period. Remarriage rates are highest for those with some college or a bachelor’s degree, and least common for high school dropouts (Payne, 2018).

There also appears to be a retreat from remarriage. Between 1950 and 2017, remarriage rates declined by more than one half (Schweizer, 2019). Importantly, this decline is not concentrated in the more distant past; the trend has continued into recent years (Schweizer, 2019; see also McNamee & Raley, 2011; Payne, 2018; Sweeney, 2010). It holds for diverse groups such as Asian, Hispanic, White, Black, and native- and foreign-born men and women. To some extent, postdivorce cohabitation may be replacing remarriage. Entrance into cohabiting unions after divorce happens much more quickly than entrance into remarriage (McNamee & Raley, 2011). Within 1.7 to 2.8 years after divorce for all racial-ethnic-nativity subgroups, 25% have entered a cohabiting union. This compares to 2.8 to 5.2 years for remarriage. Taken together, these findings are suggestive of a retreat from remarriage, with cohabitation offsetting some of this decline (Brown & Wright, 2017). An important implication of both remarriage and postdivorce cohabitation is that they create stepfamilies; most individuals who remarry or enter a postdivorce cohabitation have children from prior relationships.

Divorce

As Figure 2 shows, divorce rates increased slowly in the United States (with the exception of short period fluctuations) from the late 1800s through the early 1960s before surging upward beginning in the 1960s. Since the late 1970s, divorce rates—the number of divorces per 1,000 married women—have remained high. Estimates of the lifetime percentage of marriages that end in divorce from 1980 indicate that 43% of women’s marriages would dissolve compared with 43% to 46% in 2005 to 2010 (Schoen, 2016; Schoen & Standish, 2001). Vital statistics data from state administrative records suggest that divorce rates declined from the late 1970s through the mid-1990s, but the deteriorating quality of these data may partially account for this (Kennedy & Ruggles, 2014). However, data from the American Community Survey (ACS), an in-depth U.S. Census Bureau survey with data on divorce since 2008, lend additional support to the argument that divorce rates have declined. Furthermore, in the past decade, ACS data indicate a steeper decrease in divorce rates than in previous decades.

FIGURE 2.

Refined Divorce Rate (Divorces per 1,000 Married Women): 1890 to 2017.

Note: Dashed lines between years are linearly interpolated between data points. Sources: 1890 to 1920, decennial divorces per 1,000 married women aged 15 years and older, National Center for Health Statistics (1973, Table 1); 1920 to 1995, annual divorces per 1,000 married women aged 15 years and older, Haines (2006); 2008 to 2017, annual number of women aged 15 years and older who divorced in the previous 12 months and annual number of married women aged 15 years and older, American Community Survey, Ruggles et al. (2018).

Declines in divorce among young adults are largely responsible for recent declines in divorce rates. The divorce rate among younger adults declined between 1990 and 2010 and has continued to decline since then (Cohen, 2019; Kennedy & Ruggles, 2014). One group that has not seen declines in divorce but, rather, an increase, is older Americans, a shift coined the “gray divorce revolution.” Between 1990 and 2010, the divorce rate among those aged 50 and older more than doubled, albeit beginning from a low baseline (Brown & Lin, 2012). The gray divorce revolution is largely driven by Baby Boomers who were also more likely to divorce than their predecessors at younger ages (Brown & Lin, 2012; Brown & Wright, 2017). Cohen (2019) suggested that the decline in divorce rates among younger adults might foreshadow lower divorce rates in the future. Moreover, given ever-later ages of first marriage, those who do marry young may be an increasingly select group of highly committed couples. In addition, the aging of the population implies increasing numbers of older married couples who tend to have lower (although increasing) divorce rates than younger married couples. Kennedy and Ruggles (2014) showed that this compositional shift has dampened divorce trends—divorce rates would have been substantially higher since the 1980s if not for the aging of the married population. Thus, the increasing selectivity of couples into marriage and the aging of the married population are vital factors to consider when interpreting trends in divorce.

Similar to other demographic patterns, trends in marital dissolution have diverged by education. This divergence began among those married in the 1980s when education differences in divorce existed but were substantially smaller and continued among those married into the early 2000s (Martin, 2006; Schwartz & Han, 2014). Recent estimates indicate stark educational differences in marital dissolution as measured by either separation or divorce, whichever occurred first. The probability of marital dissolution after 20 years of marriage in 2006 to 2010 was 22% for college graduates but 51% for those with some college, 59% for high school graduates, and 61% for those without a high school diploma or GED (Copen, Daniels, Vespa, & Mosher, 2012, Table 5). The highly educated have increasingly stable marriages relative to their less-educated counterparts in a number of other countries as well, including Austria, France, Spain, Sweden, the Netherlands, Korea, and Taiwan (Cheng, 2016; de Graaf & Kalmijn, 2006; Garriga & Cortina, 2017; Härkönen & Dronkers, 2006; Matysiak, Styrc, & Vignoli, 2014; Park & Raymo, 2013).

There is some indication, however, that the educational divergence in marital dissolution is not the same across populations in the United States, for instance, education differences in marital dissolution have not grown as quickly and are not nearly as large for Black women as for White women (Kim, 2012). Differences in marital dissolution between Black and White women, however, have increased. Before 1920, race differences in divorce were relatively small, but they increased substantially between the 1960s and 1990s (Bramlett & Mosher, 2002; Raley et al., 2015; Ruggles, 1997; Sweeney & Phillips, 2004). Estimates from the late 2000s of the lifetime percentage of marriages that dissolve indicate that Black women’s marriages are more likely to dissolve than other women’s; 63% of Black women’s first marriages are estimated to dissolve within 20 years compared with 46% of White women’s, 47% of Hispanic women’s, and 31% of Asian women’s (Copen et al., 2012).

An area that has received renewed attention during the past decade because of interest in the effects of the Great Recession is the link between economic cycles and divorce. Since the early 2000s, the relationship between divorce rates and unemployment rates has been negative; that is, when unemployment went down, the divorce rate went up, and when unemployment went up (as during the Great Recession), divorce rates went down (Cohen, 2014; Schaller, 2013). As is the case for marriage, these effects were small, and evidence suggests that on balance the recession may have delayed marital dissolutions rather than led people to forgo them altogether (Cherlin et al., 2013; Schaller, 2013). One study finds that when men become unemployed while separated from their wives, legal divorce tends to be delayed. Thus, the Great Recession may have delayed divorce among those already separated (Tumin & Qian, 2017), potentially explaining some of the decline in divorce rates since 2008 (Figure 2). The convention among family demographers is to study marital dissolution starting from the time of separation rather than relying on the date of a legal divorce. Those in economic straits may only divorce if remarriage is on the horizon. Whether better economic times will result in an uptick in divorce or whether these declines are part of a longer term trend driven by declines in divorce among younger adults and increased selection into marriage remains to be seen.

Cohabitation

The United States is past the time when being unmarried meant being “single.” People are waiting longer to marry and if they marry at all, they are not delaying living with an intimate partner (Kuperberg, 2014; Lamidi & Manning, 2016). The proportion of adults living with a different-gender unmarried partner has increased steadily since cohabitation measurement began in the 1960s, with a speed-up in the trend beginning in the late 1970s (Smock, 2000; U.S. Census Bureau, 2018a). Thus, the central story of different-gender cohabitation during the past decade is that the prominence of living together outside of marriage continues to increase.

There are other ways to grasp the cohabitation trend. It is often demonstrated by examining the percentage of people who report having ever cohabited. Questions asking whether one has ever cohabited are preferable to describe people’s cohabitation experience. Because cohabitation tends to be a short-term state, with marriage or dissolution occurring within a few years, snapshot measures of current cohabitation status underestimate the prevalence of cohabitation in people’s lives. Among women aged 25 to 29, 49% had ever cohabited in 1995, rising to a striking 73% in 2011 to 2013. Although those with college educations are somewhat less likely to have cohabited, the majority of women in all education groups had cohabited by their mid to late 20s (Lamidi & Manning, 2016).

Cohabitation is not solely the province of the young. Among older adults, defined as those aged 50 and older, cohabitation has also increased. In 2000, there were approximately 1.2 million men and women aged 50 and older in cohabiting relationships (Brown, Bulanda, & Lee, 2005). By 2014, that number was 3.2 million (Hemez & Brown, 2016). We do not know the percentage of older adults who have ever cohabited. To date, data sources do not provide the requisite information to calculate that figure.

Another way researchers gauge the increasing prominence of cohabitation is via the proportion of marriages that begin as cohabiting relationships. Among women who first married between 1980 and 1984, two fifths lived with their spouses before marriage. Among those marrying between 2010 and 2014, a far higher 70% lived with their spouses beforehand. Figure 3 shows that, consistent with other demographic patterns, there is an educational divide, with more-educated women more likely to forgo premarital cohabitation than less-educated women. Among those marrying between 2010 and 2014, 85% of those without a high school degree lived with their spouse before marriage compared with 63% of those with at least a bachelor’s degree (Hemez & Manning, 2017). Figure 3 also shows that a clear education gradient in cohabitation has emerged across cohorts. Racial-ethnic differences in the percentage of adults who cohabit prior to marriage are small: The numbers are 70%, 77%, and 73% for non-Hispanic Whites, non-Hispanic Blacks, and Hispanics, respectively. The major point is that most people who marry are cohabiting first.

FIGURE 3.

Percentage of Women (15–44) Who Cohabited With Their First Husband by Marriage Cohort and Educational Attainment.

Sources: 1980 to 1984 marriage cohort, 1988 National Survey of Family Growth (NSFG); 2010 to 2014 marriage cohort, 2011 to 2015 NSFG, reproduced from Hemez and Manning (2017, Figure 3). H.S. = high school.

Recent research has examined social class differences in other aspects of cohabitation. Sassler, Michelmore, and Qian (2018), for example, investigate the pace of entry into cohabitation. Women from more advantaged backgrounds are not only less likely to cohabit, but among those who do, the pace of entry into coresidence is slower than for less-advantaged women. In addition, college-educated women who had cohabited were more likely to transition into marriage than less-advantaged women (Sassler et al., 2018; see also Sassler & Miller, 2017). There is also evidence of educational divergence in the role of cohabitation in people’s lives in Canada. Wright (2018) found that Canadians without a bachelor’s degree are more likely to have a child within cohabitation and that marriage chances for the less educated have sharply decreased.

Although most cohabiting couples marry or dissolve their relationships fairly quickly, within 2 to 3 years, recent studies suggest a trend toward lower chances of marriage and higher levels of dissolution. Mernitz (2018) studied two cohorts of cohabitors. The later cohort represented those born between 1980 and 1984, and the earlier cohort consisted of those born between 1957 and 1964. Only 32% of the more recent cohort married within 5 years of the start of cohabitation compared with 41% of the earlier cohort. There was also a slight cohort differential in terms of breaking up: 52% of the more recent cohort broke up within 5 years of living together compared with 49% of the earlier cohort (see also Guzzo, 2014; Kuo & Raley, 2016; Lamidi, Manning, & Brown, 2019).

Thus, it appears that cohabitation has become less a stepping stone to marriage, and this is especially so for those who are not economically advantaged. This is consistent with numerous studies across the past 2 decades or so showing that the well-off are more likely to transition from cohabitation to marriage than those in less salutary economic situations (e.g., Ishizuka, 2018; Smock & Manning, 1997; Xie et al., 2003). In fact, serial cohabitation—cohabiting more than once—has been rising and is more prevalent among the less educated (Eickmeyer & Manning, 2018; Lichter & Qian, 2008; Lichter, Turner, & Sassler, 2010).

Another focus of recent research is childbearing and childrearing within cohabiting unions. Studies suggests a continued “decoupling” of marriage and childbearing as cohabitation becomes an increasingly normative context for having children. Roughly 62% of nonmarital births were to cohabiting couples in 2006 to 2013, compared to 38% in 1990 to 1995 (Lamidi, 2016). In fact, the overall increase in nonmarital births over time is due to the increase in births to cohabiting couples and not to increases in nonunion childbearing (Lamidi, 2016; Wu, 2017). The share of all births to cohabiting mothers has grown from only 6% in 1980 to 1984 to 26% in 2010 to 2014. The increase cuts across all racial and ethnic groups as well as education levels, although highly educated women remain unlikely to have nonmarital births. At the same time, the share of births to single, noncohabiting women has been stable during the past few decades. It has hovered around 15% to 18% between the early 1980s and the early 2010s. In 2010 to 2014, the percentage was 15% (Wu, 2017).

A new development is emerging evidence from several studies that cohabitation appears to no longer decrease marital stability. Two decades ago, the vast majority of studies showed a positive correlation between premarital cohabitation and marital dissolution (Smock, 2000). Those who cohabited before marriage faced higher risks of divorce. The purported mechanisms were that cohabitors were a select group of the population with more liberal attitudes, including attitudes toward divorce. Another explanation was that there is something about the experience of cohabitation itself that leads people to learn that intimate family relationships are not solely defined by marriage. In the early 2000s, studies emerged showing that the negative cohabitation effect did not apply to all racial and ethnic subgroups. Phillips and Sweeney (2005), for example, found that non-Hispanic White women who cohabited were more likely to experience divorce but that this was not so for non-Hispanic Black or Mexican American women.

In the past decade, several articles were published suggesting that the “cohabitation effect” had by and large disappeared and that those who cohabit before marriage are no longer more vulnerable to divorce than others (Copen et al., 2012; Kuperberg, 2014; Manning & Cohen, 2012; Musick & Michelmore, 2015). Reinhold (2010), for example, found that cohabitation could stabilize remarriages. A possible explanation for these newer findings is that as cohabitation now has a normative presence, it no longer serves as a marker of more or less stable marriages. Research on other countries similarly suggests that as cohabitation becomes more normative, the effects of premarital cohabitation on the risk of divorce risk diminish (Liefbroer & Dourleijn, 2006). Although the weight of evidence is in the direction of little or no association between premarital cohabitation and divorce, one recently published article finds that the positive association between premarital cohabitation and marital dissolution persists (Rosenfeld & Roesler, 2019).

Fertility

Figure 4 shows that fertility rates in the United States declined precipitously after the Baby Boom (i.e., 1947–1964) through the early 1970s and have been relatively stable since then. Fertility rates increased slightly between the mid-1990s and 2007 but have since fallen to record lows (Hamilton et al., 2019). In 2018, the fertility rate was 59.0 births per 1,000 women aged 15 to 44 compared with 69.3 in 2007 (Hamilton et al., 2019; Martin et al., 2017). Low points from previous eras were higher than current levels. For example, Figure 4 shows that fertility rates in the midst of the Great Depression were higher than they were in 2018. Much of the decline in fertility since 2007 is due to declines in births among young women. Births to women age 40 and older increased, but not enough to counteract the decline among women in their teens and 20s. Teen births have been dropping especially rapidly and are at historic lows (Hamilton et al., 2019). Since 2007, fertility rates have also declined across a broad array of race–ethnic groups, including for White, Black, American Indian, Alaska Native, and Asian and Pacific Islander women, and declines were especially large for Hispanic women (Martin et al., 2017).

FIGURE 4.

General Fertility Rate (Births per 1,000 Women Aged 15–44): 1909 to 2018.

Sources: 1909 to 2003, National Center for Health Statistics (2003, Table 1–1); 2003 to 2015, Martin, Hamilton, Osterman, Driscoll, and Mathews (2017, Table 1); 2016, Hamilton, Martin, Osterman, Driscoll, and Rossen (2018); 2017 to 2018, Hamilton, Martin, Osterman, and Rossen (2019).

Despite these declines, the United States still has higher total fertility rates than most European countries, although it has not had above replacement-level fertility (above 2.1) since 2007 and was 1.8 births per woman on average as of 2017 (World Bank, 2019). In 2017, the number of births per woman in the United States was slightly lower than in Sweden, Ireland, and France among a few other Organization for Economic Cooperation and Development countries, but higher than in Belgium, Norway, and Germany, for example. An active area of research during the past decade investigates reasons for cross-national variation in fertility rates, often citing the ease or difficulty with which women combine work and family responsibilities (e.g., Esping-Andersen & Billari, 2015; McDonald, 2013).

The decline in fertility since 2007 is small relative to larger historical shifts but has inspired an active literature on whether the decline might be due to the Great Recession. The consensus is that the Great Recession did reduce fertility somewhat among younger women (for reviews, see Schneider, 2017; Sobotka, Skirbekk, & Philipov, 2011), but there is no consensus about whether the decline represents a temporary postponement or whether the total number of children women eventually have will be negatively affected by the recession (Astone, Martin, & Peters, 2015; Cherlin et al., 2013; Currie & Schwandt, 2014; Seltzer, 2019). Because the recession affected the fertility of younger rather than older women, there is still time for completed fertility rates to rebound (Cherlin et al., 2013). In addition, a key component in the decline in fertility was a large drop in fertility among Hispanic immigrants. Because the recession was associated with considerable reductions in Hispanic immigration—in particular, migration from Mexico (Villarreal, 2014)—the Hispanic immigrant population has aged during this period. Thus, rather than a direct response to economic hardship among those already in the United States, the shifting composition of the immigrant population away from peak childbearing years may be responsible for a substantial portion of the decline in Hispanic immigrant fertility (Cherlin et al., 2013; Parrado, 2011). Despite an official end to the recession in 2009 and increased employment, fertility has continued to decline. Longer term structural changes, especially the decline of manufacturing and construction, also depress fertility; thus fertility rates may be unlikely to rebound soon (Seltzer, 2019).

Unlike notable divergence by education in marriage and divorce patterns, trends in completed fertility differentials by education are less apparent. Women with more education have long had fewer children on average than those with less education, and these differentials have remained relatively stable (Isen & Stevenson, 2011; Martinez, Daniels, & Febo-Vazquez, 2018). Yet there are indications that some change has occurred. Although remaining lower than among those with less education, fertility rates among college-educated women (in particular, those with advanced degrees) have increased (Hayford, 2013; Hazan & Zoabi, 2015; Livingston, 2015; Shang & Weinberg, 2013; Vere, 2007). Some research has pointed to the greater ability of highly educated women to outsource especially as the relative price of child care for highly educated women has fallen, whereas it has increased for the less educated (Hazan & Zoabi, 2015). In addition, among highly educated women, the availability of child care has increased and their spouses participate more in child care (Antecol, 2015; Shang & Weinberg, 2013). The timing of women’s fertility by education has also diverged with college graduates delaying childbirth more than less-educated women (McLanahan, 2004). In recent cohorts, women with at most a high school degree have often completed childbearing by their late 20s, an age at which college graduates are just beginning to have children (Cherlin, Talbert, & Yasutake, 2014).

A large proportion of births occur outside marriage. After increasing steadily since the 1960s, the percentage of births to unmarried women has been relatively stable since 2008, with 40% of all births to unmarried women in 2016 (Child Trends, 2018). Births to unmarried mothers are far more common among mothers with less education. Data from 2009 to 2013 indicate that the proportion of births to unmarried mothers with less than a high school degree was 68% compared with 11% for college-educated mothers (Manning, Brown, & Stykes, 2015). Although education differences in nonmarital and unintended fertility have clearly increased over time (England, Shafer, & Wu, 2012; Hayford & Guzzo, 2016), differences in nonmarital fertility between White and Black women have shrunk somewhat because nonmarital births have increased more for White women than for Black women. Nonmarital births grew for White women from 21% in 1995 to 29% in 2016, whereas they were 70% for Black women in both 1995 and 2016 (although fluctuating somewhat in the intervening years; Child Trends, 2018). At the same time, differences between White and Hispanic women have increased because nonmarital births increased faster for Hispanic women (42% in 1995 to 53% in 2016) than for White women (Child Trends, 2018; Manning et al., 2015). Differences in marital fertility rates and timing have also declined between Black and White women (Hayford, Guzzo, & Smock, 2014) as have total fertility rates for White, Black, and Hispanic women (Sweeney & Raley, 2014).

The rise of nonmarital childbearing and family instability is also linked to an increase in multiple-partner fertility (MPF), which is defined as having children with more than one partner. MPF existed in the past—6% of White married couples experienced MPF in 1955 (Zobl & Smock, 2015)—but it has increased markedly and is now less associated with marital fertility within stepfamilies and more associated with nonmarital fertility (Stykes & Guzzo, 2019). Among parents, recent estimates of MPF generally fall in the 14% to 25% range depending on the data source and age restrictions (Guzzo, 2014; Monte, 2019; Scott, Peterson, Ikramullah, & Manlove, 2013; Stykes & Guzzo, 2019). Among unmarried parents, the estimates are substantially higher (Cancian, Meyer, & Cook, 2011; Fomby & Osborne, 2017; Guzzo, 2014). Monte (2019) provided national prevalence estimates and reported that in cohabiting families with children, MPF is 43.6%. MPF is also more common among socioeconomically disadvantaged men and women, those who had their first births at young ages, and in populations and contexts where union dissolution rates are high (Carlson & Furstenberg, 2006; Monte, 2019; Thomson, Lappegard, Carlson, Evans, & Gray, 2014). There is also evidence that MPF has increased in the recent past and will continue to increase into the future given that younger cohorts of men and women are transitioning into MPF sooner than older cohorts (Guzzo, 2014).

Same-Gender Unions

On June 26, 2015, in a landmark case, the Supreme Court ruled in favor of marriage equality (Obergefell v. Hodges, 2015). Before that time, several states allowed same-sex couples to marry, but the 2015 decision made same-sex marriage legal in all states (we use the term “same-sex” couples here given that this is the language used in the laws, and elsewhere use the term “same-gender” couples). Same-sex marriage increased after Obergefell v. Hodges, but pinning down the effect of the decision is methodologically difficult. This stems from the fact that same-sex marriage was available in several states and the District of Columbia prior to Obergefell v. Hodges and that different data sources and assumptions affect estimates.

In 2016, there were 486,994 married same-gender couples, which were relatively evenly split between female and male couples (roughly 251,000 and 235,000, respectively; U.S. Census Bureau, 2016). By 2017, estimates based on nationally representative data from a Gallup poll in combination with Census data indicate that the total number of married same-gender couples had increased to 547,000 (Romero,2017).Some scholars would argue that this is too low; a study also based on Gallup survey data suggested that even in 2015 there were as many as 780,000 same-gender married couples (Gates & Newport, 2015). Focusing only on married same-gender couples also underestimates the number of coresidential same-gender couples given that, as is true for different-gender couples, many same-gender couples are cohabiting. In 2017, there were roughly 380,000 same-gender cohabiting couple households. Similar to married same-gender households, these are fairly evenly split between female and male couple households (U.S. Census Bureau, 2017a).

Same-gender couples tend to be more socioeconomically advantaged on a number of dimensions than different-gender couples. For example, married same-gender couples have higher levels of schooling than their different-gender counterparts. In 34% of same-gender married couples, both spouses have a college degree or more; for different-gender married couples, both spouses have at least a college degree in only one quarter of cases. Married same-gender couples are also more likely to have both partners employed (57% vs. 49%), are somewhat more likely to self-identify as White (84% vs. 81%), less likely to self-identify as Black, Asian, or Hispanic, and have higher median household incomes. In 2017, the median household income for different-gender married couples was $88,683 versus $98,889 for same-gender married couples. Cohabiting same-gender couples are also more likely to be college graduates, be employed, self-identify as White, and have higher median incomes than different-gender cohabitors (U.S. Census Bureau, 2017a).

Notably, there is a substantial gender disparity in the incomes of same-gender couples. In 2017, the annual income for female married couples was roughly $90,000 compared with a substantially higher $111,000 for male married couples (U.S. Census Bureau, 2017a). Fisher, Gee, and Looney (2018) used federal tax records on annual income in 2015 and found similar results: Female married couples had median adjusted gross incomes of $90,531 compared with $109,788 for male married couples and $79,966 for different-gender married couples. The gender disparity among same-gender married couples does not appear to be accounted for by employment or by education. For the two groups, the percent in which both spouses are employed is almost identical and there is only a trivial difference in educational attainment: In 36% of male married households and 33% of female married households, both spouses have at least a bachelor’s degree. The male income advantage among same-gender couples also holds among same-gender cohabiting couples (U.S. Census Bureau, 2017a).

Same-gender couples, married or unmarried, are less likely to be raising children than different-gender couples. Only 19% of married and 13% of unmarried same-gender couples are raising children compared with 39% of married and 38% of unmarried different-gender couples (U.S. Census Bureau, 2017a). In addition, childrearing is much more common among female than male couples. More than twice as many female couples have children (24%) than their male counterparts (9%; U.S. Census Bureau, 2017a; see also Gates, 2015). Childrearing is also more common among African American, Latino, and Native American/Alaskan than among White same-gender couples. For example, 40% of African American same-gender couples have children younger than age 18 in the home compared with 16% of White same-gender couples (Gates, 2012).

Compared to different-gender couples, the education gradient in parenthood for same-gender couples is much larger. In 2009, unmarried same-gender couples with a high school diploma were somewhat less likely than comparably educated different-gender couples to have children in the home (32% vs. 41%), but same-gender couples with college degrees were dramatically less likely to have children in the home (10% vs. 47%; Gates, 2012, Figure 3). Thus, same-gender couples with children in the home tend to have less education, lower median incomes, and higher poverty rates than different-gender parents despite the fact that same-gender couples as a whole have higher incomes and education than different-gender couples (Albelda, Badgett, Schneebaum, & Gates, 2013; Fisher et al., 2018; Gates, 2012, 2015; Rosenfeld, 2010). When differences in couples’ education and other characteristics predictive of poverty are controlled, one study using data from 2010 to 2013 found that children in female couple households have similar rates of poverty as their counterparts in different-gender couple households (Brown, Manning, & Payne, 2016), but others found that even after controlling for similar characteristics, female couple households were more likely to be poor (Albelda et al., 2013; Schneebaum & Badgett, 2019). Data that include information on sexual orientation show that controlling for education, demographic, marital status, and health measures, lesbian women and gay men are not more likely to be poor than heterosexuals, but bisexual women and men are more likely to be poor (Badgett, 2018).

There are also indications of a decline in parenthood among same-gender couples since 2006 stemming from a decrease in births from prior different-gender relationships; this mirrors trends in the general population (see Figure 4). Although the majority of children living with same-gender parents are biological or stepchildren of one of the partners, the rates of adoption and fostering are much higher among same- than different-gender couples (Gates, 2012, 2015). Yet the decline in births from prior relationships outpaced increases in adoption, fostering, and reproductive technologies, leading to an overall decline in parenthood among same-gender couples since the mid-2000s (Gates, 2015).

Relationship dissolution patterns among same-gender couples have been actively studied during the past decade. Perhaps the most important observation is that to date there are no studies that directly compare dissolution rates among legally married same-gender and different-gender couples in the United States. A fairly consistent finding is that same-gender couples (including both cohabitors and married couples), and in particular, female couples, have higher dissolution rates than different-gender couples (e.g., Joyner, Manning, & Bogle, 2017; Kolk & Andersson, 2018; Lau, 2012; Wiik, Seierstad, & Noack, 2014). One reason is that despite the current legality of same-gender marriage, same-gender couples remain less likely than different-gender couples to be married, and marriage is a more stable relationship type than cohabitation or dating (Joyner et al., 2017; Rosenfeld, 2014). Tracking trends in dissolution patterns is an important area of future research.

Immigrant Families

In 2017, 44.4 million people in the United States were immigrants, often termed foreign born, representing 13.6% of the U.S. population (Radford, 2019). Immigration to the United States surged in the 1980s through the early 2000s along with the share of the population that is foreign born (Office of Immigration Statistics, 2017, Table 1; Radford, 2019). Since the Great Recession, legal immigration has leveled off, and the size of the undocumented immigrant population has remained relatively constant or declined somewhat, depending on the estimate (Massey, 2012; Office of Immigration Statistics, 2017, Table 1; 2018, Figure 1; Radford & Noe-Bustamante, 2019).

Despite the leveling of immigration into the United States, the share of the population that is foreign born has grown since the 1970s and reached 14% in 2017 (compared with a high of almost 15% in 1890; Radford, 2019). During this period, the composition of immigrant flows have changed. In the early 2000s, most immigrants arriving in the United States were from the Americas, Mexico in particular, but immigration from Mexico has declined since the Great Recession, and immigration from Asia has increased (Massey, 2012; Passel, Cohn, & Gonzalez-Barrera, 2012; Radford & Noe-Bustamante, 2019; Villarreal, 2014). As of 2017, the largest groups of immigrants are from South and East Asia (27%), Mexico (25%), other countries in Latin America (25%), and Europe or Canada (13%; Radford & Noe-Bustamante, 2019).

What are the basic family patterns of immigrants? First, marriage is common in this population. Among those at least 18 years old in 2017, nearly 61% of those born outside of the United States were currently married. This compares to 48% among those born in the United States (Radford & Noe-Bustamante, 2019). It should be noted that these figures are based on a marital status question so do not capture past experience, only present circumstances. Part of the reason marriage prevalence is higher for the foreign born is because they are younger and in prime ages for marriage.

Second, marital dissolution is not unusual. Life table estimates using data from 2006 to 2010 suggest that 24% of married foreign-born Hispanic women will separate (due to marital discord) or divorce within 10 years of marriage compared with 32% of marriages in the general population (Copen et al., 2012). The percentage of the foreign born currently divorced or separated has increased slowly during the past several decades, from 9% in 1990 to 11% in 2017 (Radford & Noe-Bustamante, 2019). These numbers conceal a good deal of heterogeneity. For example, almost 20% of those from the Caribbean were separated or divorced compared with 7% of those from South and East Asia (Radford & Noe-Bustamante, 2019).

Tougher immigration enforcement associated with deportation is also a cause of family separation (Amuedo-Dorantes & Arenas-Arroyo, 2019; Amuedo-Dorantes, Pozo, & Puttitanum, 2015). For example, Amuedo-Dorantes and Arenas-Arroyo (2019) found that increases in immigration enforcement between 2005 and 2015 raised the likelihood that undocumented mothers lived without their spouses by 20% and the likelihood that Hispanic U.S.-born children lived without their parents by 19%. Family separations associated with deportation are associated with increased poverty, fear, stress, anxiety, and depression (Dreby, 2012, 2015; Hagan, Rodriguez, & Castro, 2011; Rodriguez & Hagan, 2004; Suárez-Orozco, Bang, & Kim, 2011). Nonmigrant mothers that are left behind in their origin country also experience distress when their spouses migrate to the United States (Nobles, Rubalcava, & Teruel, 2015).

Third, immigrants play a large role in fertility in the United States. After increasing steadily for decades, the number of U.S. births to foreign-born mothers declined between 2007 and 2014, a trend that also held for births to undocumented immigrants (Livingston, 2016; Passel & Cohn, 2016). As noted in the fertility section of this article, birth rates among immigrant women fell more rapidly than those for U.S.-born women in the wake of the Great Recession, thereby contributing to an overall decline in fertility rates. Nevertheless, births to foreign-born women still comprise a large portion of all births in the United States. In 2014, 23% of U.S.-born babies had foreign-born mothers. Most of these births were to women born in Latin America (54%), but the share of births to women born in Latin America has declined since 2010 while the share of births to Asian mothers has increased alongside migration from Asia (Livingston, 2016).

Recent research has also shown that births to foreign-born women are compensating for declining births among U.S.-born women in a number of states (including North Carolina, Florida, Washington, and Nebraska among others; Livingston, 2017). A combination of low birth rates, population aging, and increased mortality among the non-Hispanic U.S. population means that Hispanic migration and fertility have been important in offsetting population decline in many locations. Furthermore, because immigration to the United States from Mexico has declined and a smaller share of new immigrants are from the Americas (Radford & Noe-Bustamante, 2019) this means that Hispanic immigration contributes proportionately less to U.S. population growth than in the past, thereby increasing the relative contribution of Hispanic fertility (Johnson & Lichter, 2013, 2016). If immigration declines in the future, the relative contribution of Hispanic fertility versus in-migration to U.S. population growth might increase further; yet a countervailing factor could be the tendency for fertility to be higher among recent immigrants and lower among second and third generation immigrants in the United States (Carter, 2000; Choi, 2014; Parrado, 2011), a pattern also found in Europe (e.g., Dubuc, 2012; Kulu et al., 2017).

Another area of research concerns trends and patterns of intermarriage between immigrants and the native born; intermarriage has often been used to gauge the extent to which immigrants form intimate social ties with the native born. During the past decade, researchers have increasingly focused on differences in immigrant–native intermarriage rates by country of origin rather than classifying immigrants into panethnic groups (Alba & Foner, 2015; Kalmijn, 2012; Lichter, Qian, & Tumin, 2015; Qian, Glick, & Batson, 2012). A major finding from this research is that immigrants from some countries are more likely to marry across ethnic lines the longer they have been in the United States and across successive generations. Other immigrant groups, however, sustain much lower intermarriage rates across generations despite high education and earnings (Lichter et al., 2015).

Scholars have also increasingly taken advantage of new survey data (questions included since 2008 in the ACS) on the timing of marriage and migration. Prior research has generally focused on marriages occurring since immigrants’ arrival in the United States. Recent research, however, shows that this approach ignores substantial proportions of intermarriages that occur either before the immigrant spouse migrated to the United States or occurred in the same year of migration. Data covering the 2010 to 2014 period show that 46% of immigrant women married to native-born U.S. citizens married prior to or in the same year as migration (Balistreri, Joyner, & Kao, 2017, Figure 1). These findings suggest that migration and marriage are often intertwined processes (Balistreri et al., 2017; Stevens, Ishizawa, & Escandell, 2012). Unlike in the 2000s when researchers documented declines in intermarriage between foreign-born Hispanics and native-born Whites (Qian & Lichter, 2007), little research has examined trends in intermarriage between the foreign and native born in the current decade. Given changing migration patterns, this is a fruitful avenue for future research.

Children’s Living Arrangements

Children’s living arrangements reflect the family transitions of their parents. As we have underscored, change continues to occur in many realms relevant to children. Parents divorce, have children in cohabiting relationships or on their own, and enter new marital or cohabiting relationships.

Taking the long view, children’s living arrangements have changed substantially. For example, in 1970, 12% of all children lived in single-parent families, the vast majority of which were mother-only families. By 2018, 27% did so (U.S. Census Bureau, 2018c, Table CH1). A more specific portrait is as follows: In 2016, the proportion of all children living in a two-biological parent married household stood at 60%. Roughly 3% lived in a two-parent biological cohabiting family, 9% in a stepparent family, 24% in a single-parent family, and the remainder in other arrangements. Of the 9% of children living in a stepfamily, approximately 55% are in step-parent-married families, and 45% are in cohabiting stepfamilies. Among the 24% of children living in single-parent families, the vast majority live with their mother; only 3% live in a single-father family (Eickmeyer, 2017b, 2017c). These numbers are virtually identical to those for 2010, indicating little change in the past decade (Eickmeyer, 2017a).

Although only 9% of children were living in a stepparent family in 2016, stepkin are a more prominent feature of U.S. families than this rather low percentage suggests. The formation of stepfamilies in childhood also extends into individuals’ adult lives. A recent study estimates that 30% of U.S. households contain a tie to an adult stepchild or stepparent (e.g., adults with a stepchild in the household or with a stepparent in another household; Wiemers, Seltzer, Schoeni, Hotz, & Bianchi, 2019). The prevalence of stepkin ties is substantially higher among younger households (i.e., head of household is younger than 55 years old) than older households (37% vs. 19%), consistent with the trends covered in this review.

Given the growing diversity of family forms, children are also increasingly likely to live with a grandparent. In 1970, the share was 3.2%, rising to 8% in 2015. Although these percentages are small, almost 30% of all children in the United States have coresided with a grandparent at some point during childhood (Amorim, Dunifon, & Pilkauskas, 2017). Although the majority of children living with a grandparent also live with their parent(s) (i.e., a three-generation household), 25% live with their grandparent without a parent in the household. This latter configuration is termed a skipped-generation household.

There is substantial variation along the lines of social class and race–ethnicity. The general upshot is that economics and family structure play out in a way that disadvantages children with less-privileged backgrounds and underrepresented minorities. Consider children in two biological-parent families; this includes both married and cohabiting couple families. Among those in married families, 53% have a parent with at least a college degree compared with 15% of children in cohabiting families (Eickmeyer, 2017a). Data collected in 2017 are also indicative of a social class divide (U.S. Census Bureau, 2017b). Among parents with a bachelor’s degree or more, 88% are married. In contrast, among parents without a high school degree, 67% are married (see also Stykes & Williams, 2013). In addition, skipped-generation households tend to be more economically vulnerable than three-generational households. More than 30% of children in skipped-generation households live in poverty compared with 17% in three-generational households (Wu, 2018).

White and Asian families are more likely to be in family structures associated with low poverty rates. In 2018, 70% of White children, 85% of Asian children, 36% of black children, and 61% of Hispanic children lived in two-parent married families (data constraints preclude information for detailed Hispanic and Asian subgroups; U.S. Census Bureau, 2018c, Table C3). Poverty rates for children vary as follows: 12% for non-Hispanic White, 26% for Hispanic, 12% for Asian, and 30% for Black children (U.S. Census Bureau, 2018d). There are exceptions to these general patterns. For example, some Asian groups are economically advantaged, but Asian children are more likely to live in three-generation households than Black and Hispanic children (Amorim et al., 2017; see also Pilkauskas & Cross, 2018).

Although cross-sectional snapshots of children’s living arrangements are useful, family researchers are more engaged with documenting and understanding children’s experiences as they unfold over time. Although this has been a major research theme for more than 2 decades, it continues to be a very active area of current research. It is one thing to say that a certain percentage of children live in a particular family structure, but another to track their experiences during the course of their childhoods. This interest stems not only from the goal of demographic accuracy and detail but also from a large number of studies, suggesting that family structure instability is typically not advantageous for children in the United States (e.g., Brown, 2010; Bzostek & Berger, 2017; Lee & McLanahan, 2015).

Given the trends we have reviewed here, it is unsurprising that many children experience a good deal of family instability. There are two concepts and measures that researchers are using to study these issues. Family instability generally refers to parents and children, occurring when a parent leaves or a new parent figure moves into the household. The child may also experience other changes such as moving in (or out) with a grandparent or other relative, but these changes have received less attention overall in the instability literature. The other concept is family complexity; studies of family complexity are largely an achievement of the past decade (Carlson & Meyer, 2014; Meyer & Carlson, 2014). Typically this work focuses on complex sibling composition (half siblings, stepsiblings) or integrates both sibling and complex-parent configurations (single-parent, stepparent; Fomby, Goode, & Mollborn, 2016; Fomby & Osborne, 2017; Manning et al., 2015). Complexity in sibling composition is driven by MPF in which one parent or stepparent has children from more than one relationship (Guzzo, 2014).

Two recent studies examine the prevalence of family instability or complexity for children over time. Using data on children born between 1985 and 2010, Rackin and Gibson-Davis (2018) found that family instability has plateaued for children with highly educated mothers, but has increased for less-advantaged children. Contrasting children’s family situations in 1996 and 2009, another study concluded that there has been a plateau in family complexity (Manning et al., 2015).

Research on family instability and complexity and their linkages to child well-being will likely continue in the foreseeable future. One emerging direction is that researchers are increasingly taking account of other key social, health, and demographic concepts and measures when studying children’s family experiences and well-being. Some examples include a study of father loss and telemore length published in the journal Pediatrics by an interdisciplinary team including family demographers, a developmental psychologist, and a pediatrician who specializes in genetics and children’s health (Mitchell et al., 2017). Fowler, Henry, and Marcal (2015) examined both family and housing instability and their connections to adolescent criminal activity (see also Bosick & Fomby, 2018; Cavanagh, Stritzel, Smith, & Crosnoe, 2018; Gaydosh & Harris, 2018; Mitchell et al., 2015). Perkins (2019) provided a fresh perspective on the concept of instability by focusing on household instability involving nonrelatives. Her work showed that studies on only family instability miss roughly 20% of children’s experiences of instability. Moreover, household instability has a negative association with children’s educational attainment on par in magnitude with family instability.

Discussion

In 1990, Larry Bumpass asked “What’s happening to the family?” concluding that families have undergone revolutionary changes and that these changes were unlikely to reverse (Bumpass, 1990). Trends in the 1970s and early 1980s certainly pointed in this direction—declining marriage rates, increases in cohabitation and nonmarital births, declining fertility, and large increases in divorce. In contrast, Bianchi and Casper (2000; p. 1) interpreted the mid to late 1990s a period of “a ‘quieting’ of family change,” with smaller changes in many family patterns than in earlier decades.

The current decade has similarly witnessed smaller changes or stabilization in some trends and continuation of others, making it more difficult to provide a general characterization of family change during this period. With regard to marriage, trends suggesting a retreat from marriage appear to have continued at a generally similar pace as in previous decades. The average age at marriage for men and women has continued its upward trajectory and is now at unprecedented levels, and the proportions of adults living in marital relationships has continued to decline while the proportion living in cohabiting relationships has increased. In addition, fewer cohabitors are transitioning to marriage.

Unlike the 1970s when divorce rates increased rapidly, divorce rates have remained high but have declined since the early 1980s. Although a number of scholars have made this observation, there has been some doubt about the validity of a decline given the deteriorating quality of vital statistics data on divorce (Kennedy & Ruggles, 2014). New analyses using data from the ACS, however, show a continuation of previous declines and a somewhat sharper drop since 2008 (Figure 2; Cohen, 2019), lending support to the decline interpretation. One of the main drivers of this decline has been a drop in the divorce rate among younger adults, suggesting potential future declines (Cohen, 2019). In contrast, the divorce rate among older Americans has increased—the “gray divorce” revolution. Also contributing to the decline in divorce is the aging of the married population (Kennedy & Ruggles, 2014) because despite “gray divorce,” older married couples remain less likely than younger couples to divorce. In addition, fewer couples are getting married. Thus, the married population, particularly young married couples, may be increasingly composed of those who are highly committed and less likely to divorce. Potential selection and age composition effects are important to keep in mind when interpreting trends, but it is also noteworthy that when Bumpass (1990) was reflecting on basic demographic trends, including trends in divorce, they all pointed in the same direction. This is less true today.

Fertility trends have shown more stability than change during the past decade. Since the large decline in fertility after the Baby Boom, fertility rates have been relatively stable. Still, the small drop since 2007 has brought U.S. fertility rates to record lows. Notably, although U.S. fertility is high relative to many developed nations, it is below replacement and moving down in the rankings. It appears that the Great Recession may have played a role in the recent decline along with declining Hispanic immigration (Cherlin et al., 2013). Although the number of U.S. births to foreign-born mothers declined substantially between 2007 and 2014, a substantial proportion of U.S. births occur to foreign-born women, and births to foreign-born women offset declining birth rates in many areas of the United States (Livingston, 2016, 2017). Whether fertility rates will rebound is an open question.

We have seen a striking change during the past few decades in the context in which fertility occurs, but this trend may be “quieting.” The percentage of births to unmarried women increased steadily since the 1960s but has been relatively stable since 2008 (Child Trends, 2018). This is also true for the share of nonmarital births to cohabiting women—after increasing steadily, the percentage of nonmarital births to cohabiting women has plateaued in the past decade at roughly 62% (Lamidi, 2016). Although recent trends in nonmarital fertility may have “quieted,” there is some evidence that instability in children’s living arrangements has continued to increase, at least for less economically privileged children. MPF has risen and is common among disadvantaged women (Guzzo, 2014; Rackin & Gibson-Davis, 2018). Yet there is also evidence of a possible plateau in children’s experiences of family complexity (Manning et al., 2015). Discrepancies in findings and interpretations are sensitive to the time points being compared. Looking over the longer term and not only this past decade, children’s family experiences have changed dramatically. More work on family complexity is needed, especially studies including very recent data.

The 2010s has also seen a growing literature on same-gender unions, the most recent work catalyzed by the landmark Supreme Court ruling in favor of marriage equality. Unsurprisingly, now that it is legal in all states, there has been an increase in same-gender marriage. A growing number of studies in the past decade have examined relationship quality or dissolution among same- and different-gender couples. To our knowledge, there is no research to date directly comparing dissolution chances among legally married same- and different-gender couples in the United States. This will be an important avenue for future research, once same-gender marriage has been in place nationally longer. In addition, researchers should continue to track and update trends in the basic demographic characteristics of same-gender couples (e.g., Black, Gates, Sanders, & Taylor, 2000). Furthermore, as population-representative data on gender identity and sexual orientation become more widely available, demographers will be better able to describe families including members of gender and sexual minorities (Baumle, 2018). Studies of transgender individuals have already begun to appear in demographic journals (Lagos, 2018).

A major theme throughout our review has been “diverging destinies” (McLanahan, 2004). There is evidence for the continuation of this trend. The highly educated experience more marriage, more cohabitations that lead to marriage, less divorce, less family instability, and less nonmarital fertility. There are also distinct family patterns by race and ethnicity, with Whites and Asians experiencing less nonmarital childbearing and more marriage than Hispanics and African Americans. Part of these differences stem from inequalities in socioeconomic status. The more privileged population subgroups, such as Whites and Asians, have more institutionalized family arrangements.

In closing, we make two observations. First, one cannot understand or contextualize social phenomena without studying families. Whether one is interested in inequality, race–ethnicity, emerging adulthood, mortality, dating, or immigration, families cannot be ignored. To take an example, social inequality is reproduced and maintained across generations. Thus, family change is necessary for studying, and vital for understanding, social stratification and inequality (e.g., Bloome, 2017; Maralani, 2013).

Second, the family is continually changing, as observed by many family scholars and by educator Ernest Groves almost a century ago, whom we quoted at the start of this review: “We say the home is in transition. So it is…The home problem, therefore, is not one that we can finally solve” (1925, pp. 228–229). Family change may be becoming even more complex in that trends in the recent past do not always point in the same direction or appear as uniform as in earlier decades.