Abstract

Over the last two decades, the share of U.S. children under age 18 who live in a multigenerational household (with a grandparent and parent) has increased dramatically. Yet we do not know whether this increase is a recent phenomenon or a return to earlier levels of coresidence. Using data from the decennial census from 1870 to 2010 and the 2018 American Community Survey, we examine historical trends in children’s multigenerational living arrangements, differences by race/ethnicity and education, and factors that explain the observed trends. We find that in 2018, 10% of U.S. children lived in a multigenerational household, a return to levels last observed in 1950. The current increase in multigenerational households began in 1980, when only 5% of children lived in such a household. Few differences in the prevalence of multigenerational coresidence by race/ethnicity or education existed in the early part of the twentieth century; racial/ethnic and education differences in coresidence are a more recent phenomena. Decomposition analyses do little to explain the decline in coresidence between 1940 and 1980, suggesting that unmeasured factors explain the decrease. Declines in marriage and in the share of White children most strongly explained the increase in multigenerational coresidence between 1980 and 2018. For White children with highly educated parents, factors explaining the increase in coresidence differ from other groups. Our findings suggest that the links between race/ethnicity and socioeconomic status and multigenerational coresidence have changed over time, and today the link between parental education and coresidence varies within racial/ethnic groups.

Introduction

Since the 1990s, the number of U.S. children living in multigenerational households (also known as three-generation households) has increased substantially, growing from 5.7% in 1996 (Pilkauskas and Cross 2018) to 10% of all children in 2018 (authors’ calculation). We define multigenerational households as those in which a child (under age 18) lives with at least one parent and grandparent. Across all childhood, about one-quarter of American children have ever lived in multigenerational households (Amorim et al. 2017), and a study of a recent cohort found that 24% of children lived in a multigenerational household by age 5 (Pilkauskas and Martinson 2014). Rates of multigenerational coresidence over childhood are even higher among non-White children (Amorim et al. 2017) and among those whose parents do not have a high school education (Cross 2018), reaching about 30% for both groups. Whether these recent trends are part of a longer-term pattern or a return to earlier historical levels is not well understood. Extant research taking a long historical view of multigenerational living arrangements has focused on the living arrangements of the older adults (over age 65) (Ruggles 2003, 2007, 2011) rather than children; thus, we lack long-run historical information with which to contextualize recent trends in children’s experiences of multigenerational living arrangements.

Examining trends in multigenerational coresidence from the perspective of older adults sheds light on important shifts in caregiving, kin support, and opportunities for coresidence (Ruggles 1996, 2003, 2007). Estimating long-term trends in children’s multigenerational arrangements is somewhat more complicated, given that changing longevity, fertility, and generational length all influence the availability of kin and thus children’s family structure. However, the key role that family and household structure play in child development (Demo and Cox 2000) as well as the links between multigenerational coresidence and child well-being (e.g., Augustine and Raley 2013; DeLeire and Kalil 2002; Dunifon 2013) and parental behavior (e.g., Amorim 2019; Chase-Lansdale et al. 1994; Hao and Brinton 1997; Pilkauskas 2014a) make the study of trends in multigenerational coresidence from the child’s perspective an important endeavor.

We examine long-run historical trends in children’s multigenerational coresidence in the United States and factors that explain these trends. We also study key stratifying factors in the experience of multigenerational living arrangements: race/ethnicity and education. Although research has documented pronounced point-in-time differences in grandparental coresidence by race/ethnicity and parental education (Amorim et al. 2017; Cross 2018), we know little about longer-run historical trends in children’s multigenerational coresidence by these demographic characteristics. This is an oversight given that prior research has suggested that family experiences and life trajectories are closely linked with race/ethnicity (Raley et al. 2015) and with mother’s education (McLanahan 2004). Additionally, research has shown significant differences in the associations between multigenerational coresidence and child well-being by race/ethnicity (Dunifon and Kowaleski-Jones 2007; Mollborn et al. 2012; Pilkauskas 2014b) and by socioeconomic status (Amorim 2019; Aquilino 1996; DeLeire and Kalil 2002).

The current study addresses these gaps in the literature using data from the 1870–2010 decennial censuses and the 2018 American Community Survey (ACS). We situate current patterns of children’s multigenerational living arrangements in a broad historical context, allowing us to better understand the extent to which current differences are part of a longer-run historical trend or more recent phenomena. By examining data over almost 150 years, using a child-focused perspective, we can better understand the historical role of grandparents in the lives of grandchildren, extend our understanding of family complexity, and shed light on whether the factors associated with children’s experiences of multigenerational coresidence have changed over time. Finally, studying historical patterns in coresidence by race/ethnicity and education can increase our understanding of the shifting sources of social stratification in children’s lives.

Trends in Multigenerational Households

The increased prevalence of multigenerational households among children over the last two decades is well documented (Dunifon et al. 2014; Ellis and Simmons 2014; Kreider and Ellis 2011; Pilkauskas 2012; Pilkauskas and Cross 2018), but little is known about these trends prior to the late twentieth century. Although no research has investigated historical trends from a child’s viewpoint, prior studies have documented intergenerational coresidence between adult children (regardless of the presence of grandchildren) and older adults (over age 65) (Ruggles 2003, 2007, 2011, 2015).1 Ruggles’ work (2003, 2007) showed that intergenerational coresidence declined from about 65% of older adults living with their adult children in 1870 to about 35% in 1940; by 1990, fewer than 15% of older adults resided in such living arrangements.

Ruggles provided important historical information about the living arrangements of those aged 65 and older, shedding light on intergenerational dependence. However, demographic and economic shifts that occurred over this period likely affected children’s living arrangements differently than those of older adults. In particular, declines in fertility mean that older adults have fewer adult children with whom to live, perhaps resulting in a decline in the opportunity for intergenerational living among grandparent-aged adults (Ruggles 2003). Yet from a child’s viewpoint, declines in fertility mean less competition with aunts, uncles, and cousins for multigenerational coresidence, which increases the demographic opportunity for coresiding with grandparents. This distinction highlights the importance of considering different generational perspectives to gain a fuller understanding of the implications of demographic shifts on multigenerational living arrangements. We build on this earlier work by focusing on patterns in children’s multigenerational living arrangements in a historical context.

Key Factors Linked to Multigenerational Coresidence

The American family experienced major demographic and socioeconomic shifts throughout the twentieth century. We discuss some of the key factors that may affect multigenerational coresidence from a child’s viewpoint.

Longevity

In the United States, longevity increased significantly over the twentieth century. Life expectancy before 1900 was less than age 50 (Ruggles 2003). Average life expectancy had risen to age 63 by 1940, to 74 by 1980, and to 79 by 2017 (Arias and Xu 2019; Penn Wharton Budget Model 2016). Today, children are more likely to have a living grandparent than they were 100 years ago (Bengtson 2001; Margolis and Verdery 2019; Song and Mare 2019).These changes in longevity mean that generational overlap has greatly increased over the last century and a half (Margolis and Wright 2017; Song and Mare 2019), leading to increases in the demographic opportunity for multigenerational coresidence.

Age at Birth

In 1870, the average age of women at birth was over 29 (Ruggles 2003). By 1940, average maternal age at birth had declined to 27; by 1980, it decreased to 25 (Mathews and Hamilton 2002; Ruggles 2003). With increased longevity, the decline in age at birth decreased generational length, increasing opportunities for multigenerational coresidence. However, since 1980, maternal mean age at birth has increased, increasing generational length (age at first birth in 2018 was 27; Martin et al. 2019).

Total Fertility

The number of children that parents (or grandparents) have affects coresidence. Lower fertility means less competition (from aunts/uncles) to coreside with grandparents, increasing opportunities for multigenerational coresidence. The total fertility rate for White women declined from about 4.5 in 1870 to about 2.5 in 1940 (Ruggles 2003).2 Total fertility declined further to 1.8 by 1980 and to 1.7 by 2018 (Hamilton et al. 2019; World Bank 2020).

Relationship Status

In the second half of the twentieth century, changes in nuclear family structure—specifically, increased divorce and nonmarital childbearing, and decreased (and delayed) marriage—increased the likelihood of children living with a single parent (e.g., Cherlin 2010; Ruggles 2015). The rise in single parenthood likely led to increased multigenerational coresidence as the need for grandparent involvement increased (Dunifon 2013), and studies have shown that these parents are most likely to coreside (e.g., Dunifon et al. 2014; Pilkauskas 2012). Research focused on the increase in children’s multigenerational living arrangements after 1996 has also shown that changes in nuclear family structure explained much of the increase in multigenerational households (Pilkauskas and Cross 2018).

Urbanization

Changes in the structure of the economy and in economic opportunity might also affect multigenerational coresidence. In the late nineteenth and early twentieth century, the U.S. population changed from a mostly rural (94%), agriculturally focused population to a largely urban (57%) population (Greenwood and Seshadri 2002). By 1980, only 2% of workers remained in farming (Ruggles 2007), and rates are even lower today (about 1.5%; Bureau of Labor Statistics (BLS) 2019). This increased urbanization, which was accompanied by expansions in education (Katz 1976), also changed economic opportunities for adult children and drove younger generations away from the family farms and into wage-earning jobs in cities and towns (Ruggles 2003, 2007). Thus, as farming declined, and young adults moved to cities, opportunities and need for intergenerational coresidence may have declined (Ruggles 2007).

Immigration

Immigration may increase multigenerational coresidence if this living arrangement is more common in new immigrant communities, familism/filial obligations play an important role in new immigrant populations, or coresidence is driven by language or economic constraints (Aquilino 1990; Choi 2003; Kamo 2000). The immigrant share of the U.S. population declined from roughly 14% to 5% between 1870 and 1970, and then increased to nearly 14% in 2017 (Migration Policy Institute n.d.). However, if the grandparent generation does not live in the United States, immigration might in fact reduce the demographic opportunity for multigenerational coresidence.

Labor Force Participation

Women’s labor force participation also increased dramatically over the twentieth century, especially among women with children, from 47% in 1975 to 71% in 2016 (U.S. Department of Labor n.d.). Although increases in maternal labor force participation may increase multigenerational coresidence if mothers need more assistance with childcare (Bengtson 2001; Hao and Brinton 1997), research has indicated that the economic independence associated with employment may actually reduce coresidence with grandparents (e.g., Aquilino 1990; Choi 2003; Engelhardt et al. 2005; Pilkauskas and Michelmore 2019; Ruggles 2007).3 The relationship between economic well-being and independent living may also have changed over time. In the late nineteenth and early twentieth century, intergenerational coresidence was more common among households with greater wealth; by the mid-twentieth century, this relationship had flipped such that by 1980, older adults with fewer economic resources were more likely to coreside with their children (Ruggles 2007).

Stratification by Race/Ethnicity and Education

Our study focuses on socioeconomic status (proxied by mother’s education) and race/ethnicity as two key factors that stratify the experience of multigenerational coresidence. We study these two factors both because they are linked with the overall prevalence of multigenerational households today (Amorim et al. 2017; Cross 2018) and because demographic characteristics that influence multigenerational coresidence have shifted differently for these groups. Longevity, for example, is closely linked with education and race/ethnicity. Remaining life expectancy at age 25 for those without a high school diploma is nearly a decade shorter than for those with college education (Hummer and Hernandez 2013), and the gap in life expectancy at birth between Hispanic individuals (those with the highest life expectancy) and Black individuals (those with the lowest life expectancy) was seven years in 2016 (National Center for Health Statistics 2018).

Overall, research has indicated that family formation experiences are stratified across socioeconomic and racial lines, creating what is known as the “two-tiered system” of families or “diverging destinies” of children (Furstenberg 2014; McLanahan 2004). For example, the decoupling of marriage and childbearing has been greatest among lower-SES women than among higher-SES women in the United States (Bailey et al. 2013; Edin and Kefalas 2005; Gibson-Davis 2009, 2011), resulting in higher rates of nonmarital childbearing among the less-educated (McLanahan 2004). In addition, whereas the more-educated experience lower fertility levels and older ages at first birth, those with less education have higher fertility and younger ages at first birth—and this was also true in the historical context (Bailey et al. 2013). Family formation experiences also diverge by race/ethnicity: Black women have historically experienced greater marriage dissolution (Raley et al. 2015) and nonmarital childbearing (Wildsmith et al. 2018) than women of other races/ethnicities. Although total fertility rates are relatively similar across racial lines (1.66 for White women, 1.8 for Black women, and 2.0 for Hispanic women; Mathews and Hamilton 2019), teen birth rates remain twice as high for Black and Latina women relative to White women (Sweeney and Raley 2014).

In recent decades, non-White parents and those who are socioeconomically disadvantaged (single, young, economically unstable, or less-educated) are most likely to live in multigenerational households (Amorim et al. 2017; Cross 2018; Dunifon et al. 2014; Pilkauskas 2012; Pilkauskas and Cross 2018; Swartz 2009). Thus, the widening racial and socioeconomic differences in mortality, fertility, and family formation patterns (Furstenberg 2014; McLanahan 2004) may also lead to widening gaps in the experience of multigenerational coresidence. Historical work suggests this was the case for the experience of intergenerational living arrangements of older adults: whereas rates of intergenerational coresidence were similar for Blacks and Whites in 1940, it was twice as common among older Black individuals than among elderly White individuals by 1990 (Ruggles 2007).

We focus on the interaction of racial/ethnic and educational differences in patterns of children’s mutigenerational coresidence over a long period of U.S. history. Previous research has suggested that these characteristics may matter in tandem; for example, racial/ethnic differences in nonmarital and teen births persist across the education distribution, suggesting SES does not fully explain these differences (Raley et al. 2015). Unique historical and structural factors also shape kinship relations within racial/ethnic minorities beyond SES (Carlson and Furstenberg 2006; Stack 1974). Thus, we examine how patterns of children’s living arrangements are stratified along racial and class lines together.

Data and Method

Data

Our study uses the largest national random samples from the U.S. Decennial Census available through IPUMS (https://usa.ipums.org/usa/; Ruggles et al. 2019) for the years 1870–20104 and the American Community Survey (ACS) for 2018, both collected by the Census Bureau. Our analytic sample is composed of all children under the age of 18 living in the United States. For descriptive analyses and figures representing trends over time, all years are used.

The decomposition analysis is divided into periods that represent breaks in historic trends (see upcoming Fig. 1, in the Results section). The first period (1940–1980) is characterized by a large decline in the prevalence of multigenerational households, whereas the second period (1980–2018) is characterized by an increase in the number of multigenerational households. For both decompositions, we rely on data from the first and last year of each period. We do not decompose trends prior to 1940 because we do not have consistent variables for the whole sample prior to this time. We start in 1940 rather than 1950 (the peak) because the 1950 survey asked about the educational attainment for only 20% of the sample, whereas the 1940 survey was available for all household members (and coresidence rates in 1940 and 1950 were similarly high).

Measures

We include measures that refer to both children and their parents. For demographic characteristics, we use the mother’s information if she is in the household (in 97% of cases), and we use the father’s information if the mother is not present (in 3% of cases with a parent present). We include children who live with neither parent (about 3% to 4% of children in any given year) when examining trends for all children, but we exclude them from the decomposition analyses and descriptive analyses that investigate trends by demographic characteristics or focus on multigenerational households.

Multigenerational Coresidence

In each year, we identify multigenerational coresidence based on the relationship to the household reference person.5 A child is coded as living in a multigenerational household if s/he lives in a household where (1) the reference person (middle generation/parent) is the parent of the child (biological, step-, or adopted child) and has their own parent/parent-in-law (grandparent generation) in the household, or (2) the reference person (eldest generation/grandparent) has a grandchild in the household, and that grandchild has at least one parent (middle generation) in the household (using parent pointers provided by IPUMS).6 Because our measures are at the child level—not the household level—some children within the same household may categorized as multigenerational, and others might not.7

Child’s Race/Ethnicity

In every year, children are classified as Hispanic and non-Hispanic White, non-Hispanic Black, non-Hispanic Asian, and non-Hispanic of other race or ethnicity.

Mother’s Education Level

Consistent education information is available after 1940 (except in 1950), and we create four categories of parents’ education: less than high school, high school, some college, and college or more. Because the meaning of education has changed over time as overall levels of education have increased, we also generate an educational ranking measure in which parents are ranked each year by their percentile of educational achievement. Our main analyses divide mothers into two groups: those above and those below the median educational achievement in that year. In extensions (available in Fig. A1, online appendix), we divide the educational distribution into quartiles.

Other Mother/Household Characteristics

Marital status is coded as married, previously married (separated, widowed, divorced), or never married. Employment status is measured as employed, unemployed, or not in labor force. Mother’s age is included as an indicator of age at the birth of the focal child (<20, 20–29, 30–39, 40 or older). Total number of children is a continuous variable indicating number of mother’s coresident children. Citizenship is an indicator of whether the mother is a U.S. citizen. Farm status indicates whether the family lives on a farm. We also create an indicator for father is present, but no mother is present (father primary parent).

Method

First, we estimate the share of children living in multigenerational households over time (weighted to be nationally representative). Next, we use an adaptation of Blinder-Oaxaca decomposition techniques for binary dependent variables (for more information, see Yun 2005). These decompositions identify the underlying sources of change in the prevalence of multigenerational households over time by breaking down the total change into the portion due to changes in the distribution of observable cohort characteristics (explained) and the portion due to changes in the effects or returns of these characteristics (unexplained). We decompose the decrease in the prevalence of multigenerational households observed between 1940 and 1980, then we decompose the increase observed between 1980 and 2018. We do not decompose the trend prior to 1940 because of the lack of consistent data in the earlier period. Detailed information about the decomposition methods is available in the online appendix.

Next, we use life tables to estimate children’s demographic potential to live in a multigenerational household (for a similar approach, see Ruggles 1996). The risk of living in a multigenerational household ultimately depends on the availability of grandparents, which is shaped by their fertility and longevity—factors that cannot be accounted for in the decomposition analyses.8 We focus the life table analysis on matrilineal multigenerational households, which include at least a mother and a grandmother, although they may also include fathers or grandfathers. We focus on matrilineal households because of the analytic complexity involved in estimating kin over multiple generations, the superior data quality on fertility, and the large share of matrilineal households among multigenerational households.9 In the online appendix, we describe the methods for the life tables in detail.

Our final set of analyses focuses on differences by race/ethnicity and education. We examine differences in the share of children living in multigenerational households by the child’s race/ethnicity and parent’s educational attainment (separately and interacted) over time. We then conduct a decomposition by education (top/bottom 50th percentile) within racial/ethnic groups.

Results

Historical Trends in Children’s Multigenerational Coresidence

Figure 1 presents the percentage of children living in multigenerational households between 1870 and 2018 (black line) as well as the number of children living in such households (gray line).10 In 1870, 7.7% of children (1.3 million children) lived in a multigenerational household. This rate was relatively stable, although it increased slightly over the next 40 years to a little more than 8% in 1910. The percentage of children living in multigenerational households started to increase after 1910, reaching a peak in 1950, when 10.1% of children lived in these households. From 1950 to 1980, the percentage steeply declined to a nadir of just under 5% of children. However, since 1980, the percentage of children living in multigenerational households has been steadily increasing. In 2018, 9.9% of children (7.3 million children) lived in such households. Multigenerational coresidence among children increased by 34% between 1870 and 1950, halved by 1980, and then almost doubled by 2018.

Decomposing Observed Trends in Multigenerational Coresidence

The next set of analyses is aimed at understanding the factors that explain the large decline in multigenerational coresidence from 1940 through 1980 and then the subsequent increase after 1980. Table 1 provides descriptive statistics on the measures included in our decomposition by period for all children who live with at least one parent.11 This table demonstrates the large shifts in the demographic makeup of the population of children and families between 1940 and 2018. Parental marriage rates declined from 93% in 1940 to 67% in 2018. Education levels increased markedly: 2% of mothers had a college degree in 1940, compared with 35% in 2018. Maternal employment increased significantly from 12% to 68% over this period. The percentage of White children declined over time (87% to 51%), and the share of children with non–U.S. citizen parents increased from 6% in 1940 to 13% in 2018. The average number of children in the household declined by about one child. Last, the share of children living on farms decreased dramatically, from 29% in 1940 to less than 1% in 2018.

Table 2 shows the results of a pooled Oaxaca decomposition for the two periods (1940–1980 and 1980–2018). These decompositions aim to examine how demographic shifts in parent and child characteristics over time explain the decline in multigenerational coresidence from 1940 to 1980 and subsequent increase from 1980 to 2018. The top part of the table shows the share of children in multigenerational households at the start and at the end of each period, the change in this share during the period, and the portion of the change that is explained and unexplained by the decomposition.12 The “total explained” refers to the share of the overall difference in multigenerational coresidence that can be explained by changes in the included parental and child characteristics. This explained portion can also be interpreted as the change in multigenerational coresidence that would have happened if only the population characteristics had changed over time (and the effects of these characteristics remained constant). The “total unexplained” refers to omitted variables as well as changes in the associations between multigenerational coresidence and the observed independent variables. The bottom part of Table 2 shows the beta coefficients on the explained portions of the decomposition, which can be interpreted as percentage point changes. We also show the percentage change in the prevalence of multigenerational households that is explained by each characteristic (as a percentage of the difference over time).

Multigenerational coresidence decreased by 5 percentage points between 1940 and 1980. The explained component of the decomposition suggests that shifts in parent and child characteristics should have instead resulted in an increase in the prevalence of multigenerational coresidence of about 2.2 percentage points. The unexplained component of the decomposition captures the entire 5 percentage point decrease for the early period (–7.2 percentage points: 2.2 – 7.2 = –5 percentage points). Changes in the included population characteristics did little to explain any of the decrease in multigenerational coresidence between 1940 and 1980.

The beta coefficients for the 1940–1980 period show that nearly every population composition shift predicted an increase in multigenerational coresidence over this period. Decreases in fertility (number of children) and in the share of married parents (relationship status) between 1940 and 1980 each would have predicted about a 1 percentage point increase in the prevalence of multigenerational households. Similarly, the increases in the racial diversity of children and in mothers’ employment each should have increased multigenerational coresidence by about 0.3 percentage points. Only two shifts—increased maternal education and decreased farming—predicted a decline in multigenerational coresidence, but together they would have predicted only a 0.3 percentage point decline.

The decomposition model does a better job of explaining the 5 percentage point increase in the prevalence of children living in multigenerational households between 1980 and 2018. These findings, also in Table 2, show that for the later period, compositional changes explained about 40% of the increase in multigenerational households, and the “total unexplained,” was about 3 percentage points.

The explained increase in multigenerational coresidence is composed of the sum of the contributions of each characteristic included in the model (the beta coefficients), some of which worked against this overall trend. Between 1980 and 2018, declines in marriage contributed to a 3 percentage point increase in multigenerational coresidence. Changes to the racial/ethnic composition of children (fewer White children) increased multigenerational households by 2.6 percentage points. Declines in fertility and the increased share of fathers who are primary parents (i.e., households where mothers are not present) together also contributed to a nearly 1 percentage point increase in grandparental coresidence. Population shifts that worked against the overall increase in multigenerational households included an increase in education (–2 percentage points), employment (–0.3 percentage points), age at birth (–1.4 percentage points), and the share of non–U.S. citizens (–0.8 percentage points). For example, had education not increased, multigenerational coresidence would have risen even more.13

Comparing across the two periods, it is evident that some factors that explained changes in multigenerational coresidence in the earlier period no longer did so in the later period, and vice versa. This can occur in three ways. First, the effect of a characteristic may remain similar over time, but the demographic characteristics of the population may change. For example, older maternal age consistently predicts less coresidence over time; however, because the average age of mothers was younger (27.6 in 1940 to 25.9 in 1980) and then older (28.9 in 2018), changes in the population distribution of maternal age contributed to a rise and then a decline in multigenerational coresidence. Second, a characteristic may change the direction of its influence over time, as is the case for maternal employment. In the earlier period, multigenerational coresidence increased along with employment (perhaps because of greater childcare needs). In the later period, this factor explained a decline in multigenerational coresidence (perhaps driven by greater economic well-being to afford childcare assistance or by greater childcare availability). Last, the compositional shifts experienced by the population may be more significant in one period than the other. For example, farm status mattered much more in the earlier period (when more people were moving from farms to urban settings) than in the later period (when most people already lived in urban settings). Taken together, these changes in the influence of demographic factors suggest that the characteristics we currently think of as important predictors of coresidence may be different in the historical context and may change in the future.

Children’s Multigenerational Coresidence: Adjusting for Grandparent and Kin Availability

As described earlier, many large demographic shifts occurred over the past 150 years in the United States. Although our decomposition analyses account for some of these key shifts—such as fertility of the middle/parental generation, marital changes, and racial/ethnic shifts in the child population—we cannot account for other important shifts experienced by the grandparent generation. In Fig. 2, we summarize results that estimate the changing odds of living in a matrilineal multigenerational household while accounting for changes in kin availability (grandparents’ longevity and generational length). These factors matter because a child must have at least one surviving parent and one surviving grandparent in order to have the opportunity to reside in a multigenerational household. In Table A2 of the online appendix, we present results also adjusting for grandparent fertility.

The solid line in Fig. 2 shows the percentage of children with surviving mothers and surviving grandmothers over time. In 1910 and 1940, 47% and 60% of children were at risk of matrilineal multigenerational coresidence, respectively (a 28% increase). Between 1940 and 1980, the risk of matrilineal multigenerational coresidence increased by 50%, with little change after that. From 1980 through 2018 risk changed little, and more than 90% of children could potentially live in a matrilineal multigenerational household by 2018.

Despite the increased risk, the observed proportion of children residing with mothers and grandmothers (dashed line, Fig. 2) declined by 38% between 1940 and 1980, from 5% to 3%. The dotted line adjusts the observed estimates of coresidence by the availability of kin (the survival of mothers and grandparents). Although the overall pattern of coresidence does not change, the decline between 1940 and 1980 is even steeper, from 8% to 3%, after the estimates are adjusted for availability. Grandparents’ availability does little to explain the decline in multigenerational coresidence or the unexplained portion of the earlier decomposition. For the 1980–2018 period, the observed share of children in matrilineal multigenerational households increased from 3% to 6% (a 110% increase), and accounting for availability of kin (which was quite stable) does little to change the share of children in such households. Thus, in neither period did grandparent availability explain the observed trends in coresidence.

How Patterns of Multigenerational Coresidence Differ by Race/Ethnicity or Education

As described earlier, multigenerational households are more common among non-White households and families with less education (e.g., Cross 2018), yet little is known about whether this has historically been the case. In Fig. 3, we plot the share of children in multigenerational households by racial/ethnic group (except Asian children before 1950 because of very small samples).

The rise in multigenerational coresidence to 1950, decline to 1980, and rise thereafter is similar across racial/ethnic groups with a few notable exceptions. Prior to 1920, all three racial/ethnic groups had very similar rates of coresidence, at about 8%. Black families pulled away markedly from Hispanic and White families, starting in 1920; by 1950, the prevalence of multigenerational households among Black families nearly doubled (from 7.4% to 13.7%). In contrast, Hispanic and White families had a similar rate of coresidence until 1950 (about 10%), after which White children experienced a steeper decline in coresidence (to 3.8% for White and 6.5% for Hispanic children in 1980). Rates of coresidence among Hispanic children increased to match rates among Black children by 2000 (more than 11%). Asian families also experienced a dip and a rise in multigenerational coresidence between 1950 and 2018, but rates of coresidence were much higher among this group (always exceeding 10%). More generally, after 1950, there was a fanning out that occurred by race/ethnicity, and rates of coresidence differed greatly across groups by 2018: 7% of White children, 11% of children of “other” race/ethnicity, 13% of Black and Hispanic children, and 17% of Asian children were living in multigenerational households.

Figure 4 shows how coresidence varies over time by mothers’ education levels. The timeline in the figure starts in 1940 because education was not captured in the earlier census years. Again, we find a similar overall pattern with two notable differences. First, in the early years (1950–1970), the differences in the level of multigenerational coresidence across education groups were small (about 1–2 percentage points), but the gap widened across groups over time. In 2018, 15% of children whose mothers had a high school education lived in multigenerational households, compared with only 6% of children whose mothers had a college degree. Second, although the highest rates of coresidence between 1960 and 2010 were among children whose mothers had less than a high school diploma, this was not always the case. In 1940, mothers with less than a high school diploma were the least likely to live in a multigenerational household. By 2018, we observe a crossover: rates of coresidence among children whose mothers had a high school diploma surpassed those with less education.14

Although examining differences by education level is informative, the meaning of education has changed dramatically over the last 75 years. In 1940, nearly 80% of mothers had less than a high school education; by 2018, this number had declined to 10%. For this reason, we also examine the patterns in multigenerational coresidence using the education distribution (following Bailey et al. 2013), dividing the sample into the bottom and top 50th percentiles, as shown in Fig. 5.15

Starting with panel a in Fig. 5, we see a decline in multigenerational coresidence for both education groups between 1940 and 1980. However, the decline was much steeper for those in the top 50th percentile of education than for the bottom 50th percentile. In 1940, coresidence was more common among the top 50th percentile than among the bottom, but by 1960 it was reversed; coresidence was more common among the bottom 50th percentile than among the top (i.e., the lines crossed). In 1940, 12% of children in the top 50th percentile lived in a multigenerational household, but by 1980 only 4% did so. For these children, the rise in such living arrangements post-1980 was small—an increase of 2.5 percentage points. In comparison, for children whose parents were in the bottom 50th percentile of the education distribution, the decline from 1940 to 1980 was much less steep (a 3 percentage point decrease); the increase post-1980 was more pronounced, from 6% in 1980 to 13% in 2018. Although many studies have documented links between children’s multigenerational coresidence and lower levels of education, this was not always the case historically and emerged only around 1950.

In Fig. 5, we also plot trends by 50th percentiles in education within racial/ethnic groups (we do not plot these figures for Asian children because of small samples). Starting with White children in panel b, this plot mirrors that for the overall sample, although the levels of multigenerational coresidence are about 2 percentage points lower than for the full population of children. Panel c plots trends for Black children. Although Black children also experienced a crossover in which those with less education became more likely to coreside, it occurred later than for White children (in 1980 vs. 1950). Notably, between 1980 and 1990, there was a large increase in multigenerational coresidence for Black children in the bottom 50th percentile of the education distribution, from 10% to 15%. Since 1990, rates in both education groups have remained relatively flat (a difference of approximately 4 percentage points). In comparison, the gap between the top and bottom education groups for White children has been growing since about 1970.

Last, in panel d of Fig. 5, we plot the same trends for Hispanic children. Overall, multigenerational coresidence declined between 1940 and 1980 and then increased, but the educational crossover for Hispanic children did not really occur. Hispanic children in the top 50th percentile were also more likely to live in a multigenerational household than those in the bottom 50th percentile in the early period, but by 1990 the bottom and top percentiles converged.16 Hispanic children are far more likely than Black or White children to have noncitizen parents. In Fig. A2 in the online appendix, we show coresidence among Hispanic children by citizenship status and education. Even when the sample is restricted to Hispanic children with U.S. citizen parents (roughly two-thirds of Hispanic children), patterns by education continue to differ from those for White and Black children.

These results suggest a complex relationship between education and multigenerational family structure by race/ethnicity. Despite some change over time in the share of children in multigenerational households by education for Hispanic children, the link between education and family structure is different (and arguably weaker) than that for White families. For White children, the difference in the prevalence of multigenerational coresidence by education group has been growing since the 1980s. Although we see a big split between the bottom and top 50th percentile among Black children around 1990, unlike for White families, this gap has not grown over time (and may even be narrowing). On aggregate, there appears to be a divergence by education in the likelihood of multigenerational coresidence; however, the extent to which this occurs differs by race/ethnicity.

Decomposition by Race/Ethnicity and Education

Given the differences in trends by race/ethnicity and education, our final analysis is a decomposition within racial/ethnic groups by 50th percentile in education. Doing so allows us to examine whether the factors explaining the trends in multigenerational coresidence differ by these demographic characteristics. As was the case in the full sample, for the period between 1940 and 1980, we again cannot explain the decline in multigenerational coresidence. In fact, the decomposition would predict higher levels of coresidence for every education and racial/ethnic group, suggesting that unmeasured factors explain this change (results available in Table A3 of the online appendix). For that reason, we focus our discussion here on the factors that explain the 1980–2018 increase.

Table 3 shows the effects of compositional change on multigenerational coresidence by race/ethnicity and by 50th percentile in education for the 1980–2018 period. Although multigenerational coresidence increased for all groups over this period, the increase for the top 50th education percentile was much smaller than that for the bottom 50th percentile. For the top 50th percentile, increases were 1.5 percentage points for White children, 1.7 percentage points for Black children, and 3.2 percentage points for Hispanic children. For the bottom 50th percentile, the increase in multigenerational coresidence was about 7 percentage points for White children and Hispanic children and 4.5 percentage points for Black children.

Shifts in population characteristics (total explained) explain a much larger share of the increase for the bottom 50th percentile (across all racial/ethnic groups) than for the top 50th percentile. For Black children with parents in the bottom 50th percentile of the education distribution, the changes in the studied population characteristics explain 90% of the increase in multigenerational coresidence but only 52% of the increase for White children and 26% of the increase for Hispanic children with similarly educated parents. Shifts in the population composition explain less than 20% of the rise for the top education group for any racial/ethnic group. The fact that the unexplained component is so high among children with highly educated parents (and particularly Hispanic children) suggests that the observed increases in coresidence among the more advantaged is not explained by shifts in the modeled characteristics. Rather, the observed increases are due to unobserved variables or to changes in the effects of the measured characteristics over time.

Increased single parenthood, fathers as the primary parent, and decreased fertility all predicted higher levels of multigenerational coresidence for Black and Hispanic children in both education groups and for White children with parents in the bottom 50th percentile for education. Similarly, a number of population shifts (increased employment, age at birth, and decreased farm status) predicted a decline in multigenerational coresidence for these same groups of children.

The factors that influence multigenerational coresidence among White children in the top education group appear to differ from other racial/ethnic and education groups. Although most factors are not significant, all factors are associated with the opposite influence on multigenerational coresidence for the top 50th percentile as compared with the bottom 50th percentile (the sign on the coefficient flips). Only two population shifts explain the increase in multigenerational coresidence, although both are small: increased employment (0.2 percentage points) and age at birth (0.5 percentage points).

Citizenship status is the one factor that mattered differently across racial/ethnic groups and across education groups. Shifts in citizenship had little influence on multigenerational coresidence for Black and White children regardless of education (small or zero coefficients), whereas citizenship status influenced the top and bottom education groups differently for Hispanic children. Between 1980 and 2018, children in the bottom 50th percentile were 13 percentage points more likely to have non–U.S. citizen mothers (from 27% to 40%), whereas children in the top 50th education percentile experienced a decline in non–U.S. citizen mothers by about 3 percentage points (see Table A1 in the online appendix for details). Were this compositional shift among those in the bottom 50th percentile not to have happened, increases in multigenerational coresidence among Hispanics in this education group would have been 1.3 percentage points higher.

Discussion

Using data from the 1870–2010 decennial census and the 2018 ACS, our study is the first to describe historical trends in multigenerational coresidence from the child’s perspective. By studying trends over a long period, we enhanced our understanding of current patterns of children’s living arrangements, factors that shape coresidential arrangements, and changes in key determinants of coresidence over time. We also investigated trends by race/ethnicity and education to better understand how children’s living arrangements have historically differed by these key sources of social stratification.

Although previous studies identified increases in children’s multigenerational households starting in the 1990s (e.g., Dunifon et al. 2014; Pilkauskas and Cross 2018), our results revealed that this upward trend began in 1980. By looking at a longer time horizon, we demonstrated that the current high prevalence of multigenerational living arrangements (10%) represents a return to previous levels, which peaked in 1950 at 10% and then declined dramatically until 1980 to about 5%. These trends remained even after we accounted for changes in the availability of grandparents, which increased dramatically between 1940 and 1980 and have remained relatively stable since 1980.

From the perspective of children, multigenerational living arrangements have doubled in prevalence over the last 30 years. This increase likely has consequences for children’s well-being given that household resources are affected by who is in the home (e.g., Dunifon 2013; Reyes 2018). Still, compared with shifts in the intergenerational living arrangements of older adults (which included 65% of those over age 65 in 1870 and compared with 15% by 1990; Ruggles 2007), the share of children in multigenerational living arrangements has been quite stable in the historical context, ranging between 5% and 10% in prevalence. This stability is especially notable given the large increases in the availability of grandparents with whom children can coreside.

To examine the extent to which changes in population characteristics explain the observed patterns of children’s multigenerational living arrangements, we conducted decomposition analyses. We examined whether key demographic shifts—race/ethnicity, education, fertility (age and parity), urbanization (farm status), relationship status, employment, and immigration—influenced multigenerational coresidence. Between 1940 and 1980, changes in the characteristics examined explained very little of the decline in multigenerational coresidence. In fact, almost all compositional changes examined predicted higher rates of coresidence. For the 1980–2018 decomposition, our analysis explained about 40% of the increase in multigenerational households. In keeping with prior research (Pilkauskas and Cross 2018), increased single parenthood and greater racial/ethnic diversity in children were the biggest drivers of this increase. Additionally, had a number of shifts not occurred (e.g., increased age at birth and higher levels of educational attainment), the increase in multigenerational households in this later period would have been even greater.

Factors that predicted greater coresidence were quite consistent across periods. For example, older maternal age at birth consistently predicts less coresidence, despite changes in age at first birth over time. Some factors became more (single fatherhood and citizenship status) or less (farm status) relevant over time, but the direction of the influence remained stable. Only maternal employment had a different influence across periods; greater employment predicted more coresidence in the early period, and the converse was true for the later period. Changes in norms around female employment and the availability of childcare may have driven this shift.

Why did our decomposition analyses explain so little of the decline in multigenerational households between 1940 and 1980 and only 40% of the rise between 1980 and 2018? The unexplained portion of the decomposition may be driven by both omitted variables and by changes in the influence of observed characteristics over time. In the early decomposition, our analyses suggest that omitted variables (captured by the intercept of the decomposition, not shown) drove more of the unexplained portion than changing influences of measured characteristics. Unmeasured factors that might affect multigenerational coresidence include grandparent characteristics, such as economic well-being or health (Swartz 2009); changes in preferences and economic ability to live independently, driven by changes in housing costs, homeownership policies, or other social policies like Social Security (e.g., Engelhardt et al. 2005) or income support policy (e.g., Pilkauskas and Michelmore 2019); shifts in attitudes (e.g., parenting norms/demands and childcare needs); strength of kin ties; spatial proximity; mobility; or migration patterns (Lesthaeghe 2011; Ruggles 2007; Stack 1974; Swartz 2009). In the later decomposition (1980–2018), much of the unexplained portion of the increase was driven by changes in the effects of the measured characteristics, rather than omitted variables, given that the intercept was small. In supplemental analyses examining correlates of coresidence (available upon request), we found, for example, that the association between being unmarried and multigenerational living arrangements weakened over time. Overall, the findings for both decompositions highlight the need for more research that can examine how these and other factors are linked with living arrangements.

Because we could not include measures of grandparent availability in our decomposition analyses, we also conducted a life table analysis examining matrilineal multigenerational households. For the earliest period (1910–1940), when we could not conduct a formal decomposition, we found that the availability of grandmothers rose 28%. This increase in longevity, coupled with a steep decline in fertility and earlier child-bearing (decreased generational length), may in part explain the slow and steady rise in coresidence through 1950. However, between 1940 and 1980, the availability of maternal grandmothers increased by more than 50%, and fertility and generational length continued to decline, yet coresidence declined. This trend is counter to expectations and emphasizes the importance of other potential unexplained factors, such as shifting norms, economic independence, policies, and mobility.

Our study also looked at key sources of social stratification in the experience of multigenerational arrangements: race/ethnicity and mother’s education. Patterns of multigenerational coresidence by maternal education show that differences in coresidence by education were small until around 1980, at which time trends began to diverge significantly by maternal education level. In recent decades, the children whose mothers had less education were most likely to live in a multigenerational household. This suggests that the relationship between living arrangements and children’s socioeconomic status has become stronger in recent decades. This finding is also in keeping with McLanahan’s (2004) research showing that children’s trajectories differ greatly by maternal education and also with Ruggles’ (2007) work documenting differences in intergenerational coresidence by income.

We also found that the relationship between race/ethnicity and multigenerational coresidence strengthened over time: despite few racial/ethnic differences in the earlier period, trends diverged by race/ethnicity after 1940. First, Black children experienced a sharp increase in multigenerational living arrangements in 1920, perhaps driven by the Great Migration (and also partly driving the overall rise to 1950), although we could not test this hypothesis. Second, Hispanic and Asian children experienced a less dramatic decline in coresidence than their White counterparts after 1950, perhaps because of different norms or preferences for multigenerational coresidence (e.g., Kamo 2000). Together, these results suggest that the factors that influence multigenerational coresidence differ by racial/ethnic groups.

Our descriptive results indicate that trends in multigenerational coresidence are increasingly shaped by both race/ethnicity and education. When we interacted race/ethnicity and education, we found that living arrangement patterns by education varied for each racial/ethnic group. First, both Black and White children experienced a historical shift by education, such that in the early periods, children with more-educated mothers were more likely to live in a multigenerational household; this then reversed in later periods (around 1950 for White children, and around 1980 for Black children). Hispanic children, on the other hand, experienced a convergence in the likelihood of coresidence by educational attainment by around 1990.

The crossover by education for White children is in keeping with previous work by Ruggles (2003, 2007), who found that intergenerational coresidence among the elderly was more common in the early period among families with higher incomes and greater wealth but that this relationship changed after 1950. Changes in opportunities for the middle generation (earning opportunities, urbanization) and declines in farming inheritance prompted these changes (Ruggles 2007). For Black children, although changes in urbanization and farming may explain some of the shift by education, the crossover occurred much later (1980), and the difference has remained steady or narrowed. This suggests that factors different from those affecting White children—such as changing needs of both generations, shifts in norms, or increased longevity for those in the lower education group (and hence availability for coresidence)—may explain the finding. Last, the pattern for Hispanic children may result from greater norms and preferences for coresidence that may be shared across levels of education (e.g., Kamo 2000).

We also investigated whether population shifts that explained the rise in coresidence varied by race/ethnicity and education. As was the case for the full sample, our decompositions for the early period by race/ethnicity and education did not explain the increase in coresidence. Between 1980 and 2018, factors that explained the increase in multigenerational coresidence for both education groups of Black children and Hispanic children were similar: increased single parenthood and decreased fertility contributed to higher prevalence of multigenerational coresidence, whereas increased employment and age at birth contributed to declines in coresidence. The one exception was citizenship status: the results suggest that higher rates of citizenship (and likely grandparent availability) would increase multigenerational coresidence among Hispanic children, regardless of parental education level.

For White children, among those with parents in the bottom 50th percentile of education, the findings were similar to Black and Hispanic children. Yet, White children with parents in the top 50th percentile for education were distinct: only increased employment and age at birth contributed to greater multigenerational coresidence between 1980 and 2018 among White children, whereas shifts in these factors predicted declines in coresidence for all other groups. Age at birth is highest for highly educated White parents (30 in 2018 vs. 28 for highly educated Black/Hispanic parents), making generational length particularly long for this group. Coupled with increased longevity, this finding may reflect an increased need for coresidence driven by the care needs of the grandparent generation (i.e., the “sandwich generation”). Coresident White grandparents are, on average, older than coresident Black or Hispanic grandparents (Pilkauskas 2014b), and thus different processes may shape multigenerational arrangements of White children with highly educated parents.

Our work has several limitations. We were not able to conduct a decomposition analysis for trends prior to 1940 because of a lack of key data. Although our models are parsimonious to avoid issues of endogeneity and we included many key demographic shifts (parity, maternal age at birth, education, relationship status, citizenship status, agriculture/urbanization) that were linked with multigenerational coresidence in prior research (e.g., Pilkauskas and Cross 2018; Ruggles 2003), many other unmeasured factors likely drive coresidence (such as health, policy, or norms). Similarly, although our analyses of kin availability adjusted for maternal grandmother longevity and fertility, we could not account for other important grandparent factors that might influence coresidence, such as healthy grandparenthood (Margolis and Wright 2017) or patrilineal prevalence. Last, our analyses by race/ethnicity were limited. Small samples precluded some analyses among Asian American children, we could not consider variation within racial/ethnic groups, and some groups were not studied (American Indian/Native American children who have rates of coresidence; Pilkauskas 2014b).

Nonetheless, our results provide new information on long-run patterns of children’s multigenerational living arrangements. By examining multigenerational coresidence from a historical perspective, we found that both race/ethnicity and education have become more strongly correlated with multigenerational coresidence than in the past. That the patterns by education differ among racial/ethnic groups also suggests that future research on children’s living arrangements should consider the interaction of these two important socially stratifying demographic characteristics.

Acknowledgments

The authors thank Martha Bailey, Matthew Hall, Daniel Lichter, and Ashton Verdery for input on methods and content.

Authors’ Contributions

All authors made substantial contributions to the conception and design of the study. Data preparation and analyses were primarily performed by Mariana Amorim. Some coding and analyses were conducted by Natasha Pilkauskas. Natasha Pilkauskas took the lead on drafting the article. Mariana Amorim and Rachel Dunifon wrote the first draft of sections of the article. All authors read and edited all drafts of the manuscript. All authors read and approved the final manuscript.

Data Availability

The data sets generated and analyzed for the current study are available through IPUMS USA, usa.ipums.org.

Compliance With Ethical Standards

Ethics and Consent

The authors report no ethical issues.

Conflicts of Interest

The authors declare no conflicts of interest.

Notes

1

Although many older adults likely live with grandchildren, the average U.S. grandparent is 64 years old (authors’ calculations based on the SIPP 2009). Median age at the transition to grandparenthood in the United States is 49 for women and 52 for men (Leopold and Skopek 2015). Nearly three-fourths of multigenerational grandparents are under 65, compared with about one-half of all grandparents (Ellis and Simmons 2014; Stykes et al. 2014). Thus, previous historical work focusing on the elderly likely excludes many multigenerational children.

2

The 1870 and 1940 estimates are for White women; Black women had higher total fertility rates, at about 7.5 and 3, respectively (Ruggles 2003).

3

Economic independence of the middle generation may particularly influence multigenerational coresidence. Ruggles (2003, 2007) showed that the decline in intergenerational coresidence among the elderly in the early part of the twentieth century was largely driven by economic opportunities of adult children, not by the economic independence of the eldest generation.

4

The exception is 1890, for which data are not available. Sample sizes range from 1% to 10% of the population, depending on the year.

5

Because the IPUMS imputed household relationships in 1870, these estimates may be less precise than other estimates.

6

We privilege the parent pointers to identify children’s parents rather than the relationship to the reference person variable. We do this to avoid erroneously calling a child’s aunt/uncle their parent.

7

We follow census definitions of multigenerational households (Ellis and Simmons 2014; Kreider and Ellis 2011). Thus, our estimates of coresidence are lower than those using the IPUMS MULTGEN variable, which categorizes children at the household level. This approach distinguishes the experience of a skipped-generation child (grandparent, no parent) from a multigenerational child (who has their parent present) because research suggests that these children’s experiences are very different (e.g., Dunifon 2018; Pilkauskas and Dunifon 2016).

8

Our decomposition analyses include parental fertility (number of coresident children) and generational length (age at birth), which may influence multigenerational coresidence. However, grandparent availability is not captured in the decompositions, which is shaped by their own fertility, longevity, and generational length.

9

In 2018, about two-thirds of multigenerational households were matrilineal, and historically rates were closer to 50% (authors’ calculations).

10

The pattern (peak in 1950, valley 1980, return by 2018) is nearly identical to the pattern shown in Fig. 1 when plotted at the household level, although the prevalence is about 1 percentage point higher at each observation (available upon request).

11

Table A1 in the online appendix shows these descriptive statistics by race/ethnicity and educational attainment.

12

These analyses are restricted to children who live with at least one parent; thus, estimates of multigenerational prevalence differ slightly from the figures.

13

Income is not included in the decompositions because it was not available in 1940 and for parsimony. An analysis including income in the later decomposition increased the percentage explained (from 1.9 percentage points to 2.2 percentage points) but did not change the other substantive findings.

14

We also observe the education crossover with earlier years of the ACS (e.g., 2014, 2015, or 2016).

15

We divide the sample by the 50th percentile for simplicity, but trends and conclusions hold if we divide the sample into quartiles (see Fig. A1, online appendix).

16

Differences between the top and bottom groups were less than half a percentage point in most years post-1990.

References

Amorim, M. (
2019
).
Are grandparents a blessing or a burden? Multigenerational coresidence and child-related spending
.
Social Science Research
,
80
,
132
144
.
Amorim, M., Dunifon, R., & Pilkauskas, N. (
2017
).
The magnitude and timing of grandparental coresidence during childhood in the United States
.
Demographic Research
,
37
,
1695
1706
. 10.4054/DemRes.2017.37.52
Aquilino, W. S. (
1990
).
The likelihood of parent-adult child coresidence: Effects of family structure and parental characteristics
.
Journal of Marriage and the Family
,
52
,
405
419
.
Aquilino, W. S. (
1996
).
The life course of children born to unmarried mothers: Childhood living arrangements and young adult outcomes
.
Journal of Marriage and the Family
,
58
,
293
310
.
Arias, E., & Xu, J. (
2019
).
United States life tables
,
2017
(National Vital Statistics Reports, No. 68/7).
Hyattsville, MD
:
National Center for Health Statistics
. Retrieved from https://www.cdc.gov/nchs/data/nvsr/nvsr68/nvsr68_07-508.pdf
Augustine, J. M., & Raley, R. K. (
2013
).
Multigenerational households and the school readiness of children born to unmarried mothers
.
Journal of Family Issues
,
34
,
431
459
.
Bailey, M. J., Guldi, M. E., & Hershbein, B. J. (
2013
).
Is there a case for a “second demographic transition”? Three distinctive features of post-1960 U.S. fertility decline
(NBER Working Paper No. w19599).
Cambridge, MA
:
National Bureau of Economic Research
.
Bell, F. C., & Miller, M. L. (
2005
).
Life tables for the United States Social Security Area 1900–2100
(Actuarial Study No. 120).
Baltimore, MD
:
Social Security Administration
. Retrieved from https://www.ssa.gov/oact/NOTES/pdf_studies/study120.pdf
Bengtson, V. L. (
2001
).
Beyond the nuclear family: The increasing importance of multigenerational bonds
.
Journal of Marriage and Family
,
63
,
1
16
.
Bureau of Labor Statistics
. (
2019
).
Labor force statistics from the Current Population Survey. Household data annual averages: 15. Employed persons in agriculture and nonagricultural industries by age, sex and class of worker 2018
[Data set].
Washington, DC
:
U.S. Bureau of Labor Statistics
. Retrieved from https://www.bls.gov/cps/cpsaat15.htm
Carlson, M. J., & Furstenberg, F. F.Jr. (
2006
).
The prevalence and correlates of multipartnered fertility among urban U.S. parents
.
Journal of Marriage and Family
,
68
,
718
732
.
Chase-Lansdale, P. L., Brooks-Gunn, J., & Zamsky, E. S. (
1994
).
Young African-American multigenerational families in poverty: Quality of mothering and grandmothering
.
Child Development
,
65
,
373
393
.
Cherlin, A. J. (
2010
).
The marriage-go-round: The state of marriage and the family in America today
.
New York, NY
:
First Vintage
.
Choi, N. G. (
2003
).
Coresidence between unmarried aging parents and their adult children: Who moved in with whom and why?
.
Research on Aging
,
25
,
384
404
.
Cross, C. (
2018
).
Extended family households among children in the United States: Differences by race/ethnicity and socio-economic status
.
Population Studies
,
72
,
235
251
.
DeLeire, T., & Kalil, A. (
2002
).
Good things come in threes: Single-parent multigenerational family structure and adolescent adjustment
.
Demography
,
39
,
393
413
.
Demo, D. H., & Cox, M. J. (
2000
).
Families with young children: A review of research in the 1990s
.
Journal of Marriage and the Family
,
62
,
876
895
.
Dunifon, R. (
2013
).
The influence of grandparents on the lives of children and adolescents
.
Child Development Perspectives
,
7
,
55
60
.
Dunifon, R., & Kowaleski-Jones, L. (
2007
).
The influence of grandparents in single-mother families
.
Journal of Marriage and Family
,
69
,
465
481
.
Dunifon, R. E. (
2018
).
You’ve always been there for me: Understanding the lives of grandchildren raised by grandparents
.
New Brunswick, NJ
:
Rutgers University Press
.
Dunifon, R. E., Ziol-Guest, K. M., & Kopko, K. (
2014
).
Grandparent coresidence and family well-being implications for research and policy
.
Annals of the American Academy of Political and Social Science
,
654
,
110
126
.
Edin, K., & Kefalas, M. (
2005
).
Promises I can keep: Why poor women put motherhood before marriage
.
Berkeley
:
University of California Press
.
Ellis, R. R., & Simmons, T. (
2014
).
Coresident grandparents and their grandchildren: 2012
(Population Characteristics No. P20–576).
Washington, DC
:
U.S. Census Bureau
.
Engelhardt, G. V., Gruber, J., & Perry, C. D. (
2005
).
Social Security and elderly living arrangements: Evidence from the Social Security notch
.
Journal of Human Resources
,
40
,
354
372
.
Furstenberg, F. F. (
2014
).
Fifty years of family change: From consensus to complexity
.
Annals of the American Academy of Political and Social Science
,
654
,
12
30
.
Gibson-Davis, C. (
2011
).
Mothers but not wives: The increasing lag between nonmarital births and marriage
.
Journal of Marriage and Family
,
73
,
264
278
.
Gibson-Davis, C. M. (
2009
).
Money, marriage, and children: Testing the financial expectations and family formation theory
.
Journal of Marriage and Family
,
71
,
146
160
.
Greenwood, J., & Seshadri, A. (
2002
).
The U.S. demographic transition
.
American Economic Review: Papers & Proceedings
,
92
,
153
159
.
Haines, M. R. (
1994
).
Estimated life tables for the United States, 1850–1900
(NBER Historical Paper No. 59).
Cambridge, MA
:
National Bureau of Economic Research
. Retrieved from https://www.nber.org/papers/h0059.pdf
Hamilton, B. E., Martin, J. A., Osterman, M. J. K., & Rossen, L. M. (
2019
).
Births: Provisional data for 2018
(National Vital Statistics Reports, No. 007).
Hyattsville, MD
:
National Center for Health Statistics
. Retrieved from https://www.cdc.gov/nchs/data/vsrr/vsrr-007-508.pdf
Hao, L., & Brinton, M. C. (
1997
).
Productive activities and support systems of single mothers
.
American Journal of Sociology
,
102
,
1305
1344
.
Heuser, R. L. (
1976
).
Fertility tables for birth cohorts by color: United States, 1917–73
(DHEW Publication No. (HRA) 76-1152).
Rockville, MD
:
U.S. Department of Health, Education, and Welfare
. Retrieved from https://www.cdc.gov/nchs/data/misc/fertiltbacc.pdf
Hummer, R. A., & Hernandez, E. M. (
2013
).
The effect of educational attainment on adult mortality in the United States
(Population Bulletin Vol.
68
, No.
1
).
Washington, DC
:
Population Reference Bureau
.
Kamo, Y. (
2000
).
Racial and ethnic differences in extended family households
.
Sociological Perspectives
,
43
,
211
229
.
Katz, M. S. (
1976
).
A history of compulsory education laws
.
Bloomington, IN
:
Phi Delta Kappa Educational Foundation
. Retrieved from https://files.eric.ed.gov/fulltext/ED119389.pdf
Kreider, R. M., & Ellis, R. (
2011
).
Living arrangements of children: 2009
(Current Population Reports No. P70-126).
Washington, DC
:
U.S. Census Bureau
.
Leopold, T., & Skopek, J. (
2015
).
The demography of grandparenthood: An international profile
.
Social Forces
,
94
,
801
832
.
Lesthaeghe, R. (
2011
).
The “second demographic transition”: A conceptual map for the understanding of late modern demographic developments in fertility and family formation
.
Historical Social Research
,
36
(
2
),
179
218
.
Margolis, R., & Verdery, A. M. (
2019
).
A cohort perspective on the demography of grandparenthood: Past, present, and future changes in race and sex disparities in the United States
.
Demography
,
56
,
1495
1518
.
Margolis, R., & Wright, L. (
2017
).
Healthy grandparenthood: How long is it, and how has it changed?
.
Demography
,
54
,
2073
2099
.
Martin, J. A., Hamilton, B. E., & Osterman, M. J. K. (
2019
).
Births in the United States
,
2018
(National Center for Health Statistics Data Brief No. 346).
Hyattsville, MD
:
National Center for Health Statistics
. Retrieved from https://www.cdc.gov/nchs/data/databriefs/db346-h.pdf
Mathews, M. S., & Hamilton, B. E. (
2019
).
Total fertility rates by state and race and Hispanic origin: United States, 2017
(National Vital Statistics Reports, No. 68/1).
Hyattsville, MD
:
National Center for Health Statistics
.
Mathews, T. J., & Hamilton, B. (
2002
).
Mean age of mother, 1970–2000
(National Vital Statistics Reports, No. 51/1).
Hyattsville, MD
:
National Center for Health Statistics
.
McLanahan, S. (
2004
).
Diverging destinies: How children are faring under the second demographic transition
.
Demography
,
41
,
607
627
.
Migration Policy Institute
. (n.d.).
Tabulation of data from U.S. Census Bureau, 2010–2017 American Community Surveys (ACS), and 1970, 1990, and 2000 decennial census
. Retrieved from https://www.migrationpolicy.org/programs/data-hub/us-immigration-trends?gclid=CjwKCAiA3abwBRBqEiwAKwICA0QHVB2sJiT4Xtj-tRAnZx5j81p7t_QIvn0tP4O1YHOF1BUrooq-2BoCMN8QAvD_BwE#history
Mollborn, S., Fomby, P., & Dennis, J. A. (
2012
).
Extended household transitions, race/ethnicity, and early childhood cognitive outcomes
.
Social Science Research
,
41
,
1152
1165
.
National Center for Health Statistics (NCHS)
. (
2018
).
Health, United States, 2017: With special feature on mortality
.
Hyattsville, MD
:
NCHS
. Retrieved from https://www.ncbi.nlm.nih.gov/books/NBK532685/pdf/Bookshelf_NBK532685.pdf
Penn Wharton Budget Model (PWBM)
. (
2016
).
Mortality in the United States, past, present and future
. Retrieved from https://budgetmodel.wharton.upenn.edu/issues/2016/1/25/mortality-in-the-united-states-past-present-and-future
Pilkauskas, N. V. (
2012
).
Multigenerational family households: Differences by family structure at birth
.
Journal of Marriage and Family
,
5
,
931
943
.
Pilkauskas, N. V. (
2014
).
Breastfeeding initiation and duration in coresident grandparent, mother and infant households
.
Maternal and Child Health Journal
,
18
,
1955
1963
.
Pilkauskas, N. V. (
2014
).
Living with a grandparent and a parent in early childhood: Associations with school readiness and differences by demographic characteristics
.
Developmental Psychology
,
50
,
2587
2599
.
Pilkauskas, N. V., & Cross, C. (
2018
).
Beyond the nuclear family: Trends in children living in shared households
.
Demography
,
55
,
2283
2297
.
Pilkauskas, N. V., & Dunifon, R. E. (
2016
).
Understanding grandfamilies: Characteristics of grandparents, nonresident parents, and children
.
Journal of Marriage and Family
,
78
,
623
633
.
Pilkauskas, N. V., & Martinson, M. L. (
2014
).
Multigenerational family households in early childhood: Comparisons between the United States, the United Kingdom, and Australia
.
Demographic Research
,
30
,
1639
1652
. 10.4054/DemRes.2014.30.60
Pilkauskas, N. V., & Michelmore, K. (
2019
).
The effect of the Earned Income Tax Credit on housing and living arrangements
.
Demography
,
56
,
1303
1326
.
Raley, R. K., Sweeney, M. M., & Wondra, D. (
2015
).
The growing racial and ethnic divide in U.S. marriage patterns
.
Future of Children
,
25
(
2
),
89
109
.
Reyes, A. M. (
2018
).
The economic organization of extended family households by race or ethnicity and socioeconomic status
.
Journal of Marriage and Family
,
80
,
119
133
.
Ruggles, S. (
1996
).
Living arrangements of the elderly in America: 1880–1980
. In T. K. Hareven (Ed.),
Aging and generational relations over the life course: A historical and cross-cultural perspective
(pp.
254
271
).
Berlin, Germany
:
de Gruyter
.
Ruggles, S. (
2003
).
Multigenerational families in nineteenth-century America
.
Continuity and Change
,
18
,
139
165
.
Ruggles, S. (
2007
).
The decline of intergenerational coresidence in the United States, 1850 to 2000
.
American Sociological Review
,
72
,
964
989
.
Ruggles, S. (
2011
).
Intergenerational coresidence and family transitions in the United States, 1850–1880
.
Journal of Marriage and Family
,
73
,
136
148
.
Ruggles, S. (
2015
).
Patriarchy, power, and pay: The transformation of American families, 1800–2015
.
Demography
,
52
,
1797
1823
.
Ruggles, S., Flood, S., Goeken, R., Grover, J., Meyer, E., Pacas, J., & Sobek, M. (
2019
).
Integrated Public Use Microdata Series: Version 9.0
[Dataset].
Minneapolis
:
University of Minnesota
. Retrieved from https://doi.org/10.18128/D010.V9.0
Song, X., & Mare, R. D. (
2019
).
Shared lifetimes, multigenerational exposure, and educational mobility
.
Demography
,
56
,
891
916
.
Stack, C. B. (
1974
).
All our kin: Strategies for survival in a Black community
.
New York, NY
:
Basic Books
.
Stykes, B., Manning, W. D., & Brown, S. L. (
2014
).
Grandparenthood in the U.S.: Residence status of grandparents
(NCFMR Family Profiles No. FP-14-12).
Bowling Green, OH
:
Bowling Green State University, National Center for Family & Marriage Research
.
Swartz, T. T. (
2009
).
Intergenerational family relations in adulthood: Patterns, variations, and implications in the contemporary United States
.
Annual Review of Sociology
,
35
,
191
212
.
Sweeney, M. M., & Raley, R. K. (
2014
).
Race, ethnicity, and the changing context of childbearing in the United States
.
Annual Review of Sociology
,
40
,
539
558
.
U.S. Department of Labor
. (n.d.).
Fast facts over time: Women in the labor force
.
Washington, DC
:
U.S. Department of Labor
. Retrieved from https://www.dol.gov/agencies/wb/data/facts-over-time/women-in-the-labor-force#labor-force-participation-rate-of-mothers-by-age-of-youngest-child
Wildsmith, E., Manlove, J., & Cook, E. (
2018
).
Dramatic increase in the proportion of births outside of marriage in the Unites States from 1990 to 2016
.
Bethesda, MD
:
Child Trends
. Retrieved from https://www.childtrends.org/publications/dramatic-increase-in-percentage-of-births-outside-marriage-among-whites-hispanics-and-women-with-higher-education-levels
World Bank
. (
2020
).
Fertility rate, total for the United States
(SPDYNTFRTINUSA) [Data set].
St. Louis, MO
:
FRED, Federal Reserve Bank of St. Louis
. Retrieved from https://fred.stlouisfed.org/series/SPDYNTFRTINUSA
Yun, M. S. (
2005
).
Hypothesis tests when decomposing differences in the first moment
.
Journal of Economic and Social Measurement
,
30
,
295
304
.

Publisher’s Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary data