ArticlePDF Available

Extended family households among children in the United States: Differences by race/ethnicity and socio-economic status

Authors:

Abstract and Figures

This study uses nationally representative longitudinal data from the Panel Study of Income Dynamics, to examine the prevalence and predictors of extended family households among children in the United States and to explore variation by race/ethnicity and socio-economic status (SES). Findings suggest that extended family households are a common living arrangement for children, with 35 per cent of youth experiencing this family structure before age 18. Racial/ethnic and SES differences are substantial: 57 per cent of Black and 35 per cent of Hispanic children ever live in an extended family, compared with 20 per cent of White children. Further, 47 per cent of children whose parents did not finish high school spend time in an extended family, relative to 17 per cent of children whose parents earned a bachelor's degree or higher. Models of predictors show that transitions into extended families are largely a response to social and economic needs.
Content may be subject to copyright.
1
Extended Family Households among Children in the United States: Differences by
Race/Ethnicity and Socioeconomic Status
Christina J. Cross
Department of Sociology
Gerald R. Ford School of Public Policy
University of Michigan
Ann Arbor, MI 48109
crosscj@umich.edu
ABSTRACT
This study uses nationally representative data from the Panel Study of Income Dynamics
(N=4,484), to longitudinally examine the prevalence and predictors of extended family
households among U.S. children and to explore variation by race/ethnicity and socioeconomic
status (SES). Overall, 35% of youth experience this family structure before age 18. Racial/ethnic
and SES differences are substantial: Fifty-seven percent of Black children and 35% of Hispanic
children live with an extended relative, compared to 20% of White children. Further, 47% of
children whose parents did not finish high school spend time in an extended family, relative to
17% of children whose parents earned a Bachelor’s degree or higher. Economic capacities and
family needs are key predictors of extended family coresidence. Findings suggest that extended
family households are a common living arrangement for children and that the transition into an
extended family is largely a response to social and economic need.
Material is based upon work supported by the National Science Foundation Graduate Research
Fellowship under Grant No. DGE 1256260. The author thanks Jennifer Barber, Paula Fomby, and three
anonymous reviewers for their thoughtful feedback on drafts of this article; the author also thanks Barbara
Anderson, Karyn Lacy, Fabian Pfeffer, Natasha Pilkauskas and the Inequality, Demography, and Family
workshop at the University of Michigan for very helpful comments and suggestions.
2
Keywords: Extended Family, Family Structure, Coresidence, Children, Race/Ethnicity, Socioeconomic
Status
Introduction
Most research on trends in children’s living arrangements focuses on the presence or
absence of a child’s biological parents in a household and parents’ marital or cohabitation status
(Brown, 2010; Manning, Brown, & Stykes, 2014). However, children often live in households
whose members extend beyond the nuclear family; they may also live with grandparents, aunts,
uncles, and other relatives, such as cousins. Numerous studies have documented the central role
of the extended family in the lives of children, particularly those from minority and/or
economically disadvantaged backgrounds (Stack, 1974; Giordano, Cernkovich, & DeMaris,
1993; Trent & Harlan, 1994; Bengston, 2001; Hirsch, Mickus, & Boerger, 2002; Pernice-Duca,
2010; Garrett-Peters & Burton, 2016). Coresident extended family members often contribute to
or constrain household finances through the exchange of resources such as money, food, and
transportation (Stack, 1974; Edin & Schaefer, 2015; Garrett-Peters & Burton, 2016). They may
also nurture children, provide childcare assistance, act as a co-parent, or even raise a child in the
absence of their parents (Stack, 1974; Burton, 1992; Hunter, 1997; Pittman, 2007).
Despite the well-established importance of the extended family structure, little is known
about the full extent to which children live with extended relatives, which groups are most likely
to experience this living arrangement, and which factors determine whether a child will live in an
extended household. Using data from the 1988 to 2013 waves of the Panel Study of Income
Dynamics (PSID), this study examines the prevalence of extended family coresidence across
childhood (from birth to age 17), investigates differences by race/ethnicity and socioeconomic
status (SES), and identifies predictors of this living arrangement. Documenting the pervasiveness
and predictors of extended family households is particularly important at a time when research
3
has begun to find strong associations (both positive and negative) between extended family
coresidence and children’s cognitive, behavioral, and educational outcomes (Dunifon &
Kowaleski-Jones, 2007; Foster & Kalil, 2007; Mollborn, Fomby, & Dennis, 2012; Pilkauskas,
2014; Kang & Cohen, 2017). Results from this study will allow us to better understand the
potential breadth of influence of extended family households on child outcomes and under what
conditions children are most likely to be impacted by coresidence, which can inform policy
related to family structure and child wellbeing.
I build upon previous research on family structure in several ways. First, whereas most
studies focus on a particular type of extended family, namely grandparent families (e.g., Hill,
Yeung, & Duncan 2001; Ellis & Simmons, 2014), I examine coresidence with a broader set of
extended relatives (e.g., aunts, uncles, and other relatives). Second, I use a nationally
representative sample of children. Much of the research on extended family coresidence has
focused on children from low-income and/or minority families living in urban areas, and thus
could not shed light on the overall commonness of this living arrangement. Further, it could not
compare across groups, for example, comparing minority versus non-minority or low-income
versus higher-income children’s experiences. Lastly, I use longitudinal data to document
prevalence over time; other studies have typically used point-in-time measures that do not fully
capture children’s lifetime experiences (e.g. Kreider & Ellis, 2011), and it is unclear how
dramatically this approach underestimates the prevalence of extended family coresidence.
Background
The Prevalence of Extended Family Households
A few studies examine trends in extended family households, mainly multigenerational
householdsthat is families including a child, at least one grandparent, and/or at least one parent
4
(for exceptions see Beck & Beck 1984, 1989; Mollborn, Fomby, & Dennis, 2011). One major
reason that prior research focuses on these multigenerational households is that coresidence with
a grandparent is the most common type of extended family. Most recent published cross-
sectional estimates indicate that 16% of children live with extended family, and 10.5% of
children coreside with a grandparent (Kreider & Ellis, 2011). However, an exclusive focus on
grandparents does not provide a full picture of children’s experiences in extended families.
Approximately 5% of children also live with an aunt or uncle and 7% live with other relatives
(these categories are not mutually exclusive; author’s calculation using data from the Survey of
Income Program Participation; U.S. Census Bureau, 2011). Thus, more research is needed on the
experience of living with an extended family member, more broadly defined (e.g., living with
uncles or cousins).
Further, the prevalence of extended families has slowly increased in recent decades. In
1996, 13% of children lived with an extended relative, and this figure rose to 17% by 2014
(author’s calculation using data from the Survey of Income Program Participation; U.S. Census
Bureau, 2017). Cross-sectional single-year estimates, however, underestimate the prevalence of
ever living in an extended family. Indeed, longitudinal studies confirm that incidence over time
is substantially higher than single-year estimates. Although coresidence among children was not
the focus of this study, Beck and Beck (1984) found that 24% of White women lived in an
extended family household during a 15-year time period, compared to approximately 6% in a
single-year. Pilkauskas (2012), focusing on the prevalence of three generation households among
children, found that the number of children living in three-generation families was approximately
four times higher in longitudinal data than in a cross-section. However, due to data limitations,
this study was unable to track this family structure across all of childhood, include all types of
5
extended relatives, and draw from nationally representative data. Here, I provide nationally
representative estimates of the prevalence of all types of extended family households throughout
childhood, with the expectation that including all types of extended families and calculating
estimates longitudinally will result in substantially higher prevalence than previously estimated.
There is significant racial/ethnic and class variation in the extended family structure, with
minority and/or low-income children more likely to coreside with extended relatives. Recent
estimates show that 10% of White children live with an extended relative, compared to 25% of
Black children, 24% of Hispanic children, and 20% of Asian children (Kreider & Ellis, 2011).
Additionally, of youth living with extended family members, 71% live in households receiving
public assistance, compared to 46% of children overall (Kreider & Ellis, 2011). Thus, when we
broaden the prevalence estimate to include multiple types of extended family households and to
include all of the childhood years, we expect these racial/ethnic and income differentials to
persist.
Predictors of Living in an Extended Family Household
Although no large scale quantitative studies have identified childhood experiences that
predict when children will subsequently live in extended family households, several useful
studies have described the characteristics of extended family households (Angel & Tienda, 1982;
Burr & Mutchler, 1993; Kamo & Zhou, 1994; Kamo, 2000; Cohen & Casper, 2002; Choi, 2003;
Pilkauskas, 2012). This literature suggests three types of experiences that may be related to the
formation of an extended family household: economic capacities, family needs, and cultural
norms and preferences.
Economic factors such as household income, education, housing tenure, and employment
status are associated with extended family coresidence. Families with less economic capacity
6
(e.g., less education, lower income and job loss) may form extended family households in order
to pool and more effectively use limited economic resources (Stack, 1974; Angel & Tienda,
1982; Mutchler & Krivo, 1989; Kamo, 2000; Cohen & Casper, 2002; Pilkauskas, 2012). In this
way, coresidence operates as a survival strategy to redistribute resources and minimize economic
risks. Thus, I expect that variables indicating lower economic capacity will be positively
predictive of extended family coresidence.
The needs of family members are also correlated with living in an extended family. In
particular, mother’s age at child’s birth, the age of a child, whether a child lives with both, one,
or neither parent and the health status of parents and other household members may influence the
decision to coreside. Young parents, especially single parents of young children, may be more
likely to live in extended families so that they can get additional help with childrearing (Hogan,
Hao, & Parish, 1990; Trent & Harlan, 1994; Cohen & Casper, 2002; Pilkauskas, 2012).
Similarly, parents or extended family members in poor health may also choose to coreside, either
because they need extra assistance, or because they themselves help with childcare and/or
provide aid to other family members (Burr & Mutchler, 1992; Choi, 2003). Hence, I expect that
factors indicating greater family need will be positively predictive of the tendency to live in an
extended family household.
Cultural norms and preferences are also likely to be related to the decision to live in an
extended family. Families that place greater emphasis on familism the needs of the family
take precedence over individual needs are more likely to coreside (Baca Zinn & Wells, 2000).
This cultural ideal may valorize coresidence in a way that makes it a functional and attractive
strategy for organizing household living arrangements and promoting family connectedness.
Prior research suggests that religious preference, language spoken at home, and immigrant status
7
are useful indicators of familism (Angel & Tienda, 1982; Burr & Mutchler, 1993; Kamo, 2000;
Oropesa & Landale, 2004; Sarkisian, Gerena, & Gerstel, 2006). While my nationally
representative data allow me to include rich measures of economic capacity and family needs,
they do not permit the inclusion of a robust set of indicators of cultural norms and preferences.
Thus, in this study, I focus on assessing what I refer to as resource-driven motivations for
coresidence, that is the extent to which the decision to coreside is shaped by economic capacities
and family needs. In earlier analysis, I found that being Catholic is positively related to extended
family coresidence, but because I am unable to incorporate other indicators of culture, I include
religious preference as a control variable in this analysis.
Scholars have debated whether the factors shaping the decision to live in an extended
family differ by race/ethnicity. Prior research has suggested that economic capacities and family
needs may be more predictive of extended family coresidence for minority families than White
families. The idea here is that racial/ethnic groups may differ in the strategies they employ to
cope with hardships such as financial or health crises. In one group, it may be more customary to
rely on formal support from public institutions to address a given crisis, whereas in another
group, that crisis may be addressed by informal support from the extended family in the form of
coresidence (Stack, 1974; Neighbors et al., 2007; Woodward, 2010). On the one hand, Whites,
who benefit from historical and contemporary structural advantages, may perceive institutions
such as governmental agencies and the employment structure as more welcoming and
supportive, and may draw more heavily on these entities in times of need. On the other hand,
Blacks and Hispanics, who face ongoing racial discrimination, may perceive these same
institutions as hostile and exploitative, and may depend more on support from extended family
members via coresidence (Hays & Mindel, 1973; Mutran, 1985; Musa et al., 2009). If this is the
8
case, then economic capacities and family needs will be stronger predictors of coresidence for
minority families than White families. To date, it is unclear whether resource-driven motivations
for coresidence differ by group membership. To test the extent to which economic capacities and
family needs differentially predict the likelihood of living in an extended family, I run my
multivariate models separately for White, Black, and Hispanic children and test for significant
differences. If predictors do differ by race/ethnicity, I would expect to find statistically
significant differences in the magnitude and/or direction of the coefficients by group.
Method
Data I use data from The Panel Study of Income Dynamics (PSID), from 1988-2013. I focus
on this period in order to follow a recent cohort of children through their childhood years and
better capture the extended family experiences of contemporary youth. PSID began in 1968 as a
nationally-representative sample of approximately 5,000 households. Original respondents and
their descendants were followed annually until 1997 and have been followed biennially since
then. To maintain population representativeness, a sample refresher in 1997 added approximately
500 households headed by immigrants who had entered the United States since 1968. At each
wave, the household head or the spouse or cohabiting partner of the head reports on the
household roster, employment, income, education, housing characteristics, expenditures, and
health/health care for him/herself and all other family members since the previous interview. In
2013, the interviewed sample included information on almost 25,000 adults in nearly 9,000
households.
Measures
Dependent Variable. The dependent variable is whether a child lives with an extended
family member by the observation period. This variable is dichotomous, with children who do
9
not live with extended relatives during a given wave being assigned a value of 0 and those who
do live in an extended family household being assigned a value of 1 (estimates are used to
predict only the first observed transition into an extended family). Coresidence with extended
relatives was determined using the PSID’s Family Identification Mapping System (FIMS) and
household roster information. FIMS provides unique identifiers for each focal person’s parents,
grandparents, and siblings. From this information, I identified each child’s grandparents and
aunts and uncles (siblings of parents). If a child shared the same household with at least one of
these extended relatives in a given wave, he or she is identified as living in an extended family. I
established whether a child lived with an “other relative” using the household roster, which
identifies each household member’s relationship to the head of household, spouse of head, or
head’s cohabiting partner. A child is coded as living with an other relative if he or she is the
child of the head, spouse, or cohabiting partner, and another individual in the household is the
cousin, niece, nephew, brother-, sister-, mother-, or father-in-law of the child’s parent.
Additionally, in rare instances in which neither of the child’s parents are present in a given wave,
a child is identified as living with an other relative if his or her own value for relationship to
head of household, spouse of head, or cohabiting partner is coded as “other relative”, which
indicates that the child is related to this individual by birth, marriage, or adoption, but their
relationship is not included in any other category. Thus, this measure includes children who live
with an extended relative but not with a parent; it does not include children who coreside with
nonrelatives such as boarders or friends. This study does not focus on coresidence with
nonrelatives due to the high level of volatility and limited ability to accurately capture these
households, as well as the fact that the reasons for coresidence with nonrelatives may be
qualitatively different from those related to coresidence with biological relatives (Richards et al.,
10
1987; Kalil, Ryan, and Chor, 2014). Household relationships were measured once per year until
1997, when the PSID changed to a biannual survey. After that time, a child was counted as living
in an extended household during a noninterview year if he or she lived in that arrangement in
both the surrounding interview years. If a child was not living with an extended relative in one
year, and did so in the other year, that child is coded as not living in an extended household in
the middle (noninterview) year.
Independent Variables. I use two categories of variables: economic capacities and family
needs to assess predictors of extended family coresidence. Indicators of economic capacity
include: family income, parents’ education, home ownership, and parents’ employment status.
Family income is coded into five categories (1) at or below poverty threshold; (2) 101%-200% of
poverty threshold; (3) 201-300% of poverty threshold; (4) 301-400% of poverty threshold; and
(5) greater than 400% of poverty threshold (reference). These categories are constructed by
dividing reported household income for the calendar year by the poverty threshold adjusted for
family size in that year. In 2012, 100% of the poverty threshold was $23,050 for a family of four;
400% of the poverty threshold was $92,200 for a family of this size (Department of Health and
Human Services, 2012). Parents’ education is specified as the highest level of education
completed by either parent: less than high school, high school, some college, and Bachelor’s
degree or higher (reference). Home ownership indicates whether the child’s household is owned
(reference), rented, or neither owned nor rented by the head of household (an individual may fall
into the latter category if he or she lives in non-profit housing or receives government subsidies
for housing). Parents’ employment status is determined by whether both parents are employed
(reference), at least one parent is unemployed, or at least one parent is out of the labor force.
11
Measures of family needs include mother’s age at birth, the number of a child’s parents
present in the household, the number of children in the household, the age of the focal child, and
the health status of household members. Mother’s age at birth is a categorical variable: 19 and
under (reference), 20-29, 30-39, and 40+. Number of parents is coded as both (reference), one, or
neither. The variables indicating the number of children in the home and the age of the child are
both continuous measures (at one point in the analysis, I also included a binary variable
specifying whether there is a child under age five in the home to capture whether parents need
more help with pre-school aged children, but I later excluded it because it did not improve model
specification). The health status of all household members is reported by the household head or
the spouse or cohabiting partner of the head. It consists of two dichotomous variables indicating
whether either parent is not in good health (i.e., in fair or poor health) and whether any other
household member is not in good health.
In addition to these covariates, I control for several demographic characteristics of the
child that have been correlated with extended family coresidence: race/ethnicity, sex, region
where child lives, and religious preference (Mollborn et al., 2012; Pilkauskas, 2014).
Race/ethnicity is coded into four categories: (1) White (reference); (2) Black; (3) Hispanic; and
(4) Other race. Child sex is measured as male or female (reference), and region is coded as South
(reference) versus non-South. Religious preference is divided into four categories: (1) Catholic
(reference), (2) Protestant, (3) other denomination, and (4) no religious preference. When the
religious preference of parents differs, the preference of the parent designated as the head of the
household is used. While I would have liked to also control for parents’ work schedule, hours
worked per week, and family wealth, they were not included in the analysis due to their
inconsistent availability during the observation period. Work schedule is only available for
12
children who participated in the Child Development Supplement (CDS) in 1997, 2002, and 2007
and weekly hours is available between 1988 and 1993 and biannually between 2003 and 2013.
Family wealth is available in 1989, 1994, and biannually between 1999 and 2013. In a separate
sub-analysis (results not shown), for each year that it was available, I included the inverse
hyperbolic sine of wealth excluding home equity, which adjusts for the highly skewed
distribution of wealth in the sample (and in the larger U.S. population). This factor was not
significantly related to coresidence.
Taking advantage of the longitudinal design of the PSID, multivariate analyses include
both time-invariant and time-varying variables. The time-invariant variables are: child’s race,
mother’s age at birth, parents’ religious preference, and parents’ education. While my preference
would be to use a time-varying measure of educational attainment, this variable was treated as
time-invariant for household heads and their spouse/partner until 2009, when updated
information was collected. All other covariates: income, employment status, number of parents
and children present in the household, age of focal child, and the health status of household
members are time-varying. To adjust for biennial interviewing starting in 1997, I assign the
previous year’s reported values (adjusting income for inflation) as the missing year’s values for
the time-varying covariates during noninterview (i.e., even) years in the 1998-2012 period. All
time-varying covariates are lagged one year prior to the observation of extended family
coresidence. Following the example of Carlson, VanOrman, and Pilkauskas (2013), I use
multiple imputation with chained equations in Stata 14 to restore missing time-constant
independent variables and to improve the generalizability of my findings. The proportion of
missing cases ranges from .02 on parents’ education to.06 on parents’ religious preference.
Analytic strategy
13
Most research using observational data uses measures of current household
characteristics to predict the outcome of interest (extended family coresidence), and both are
measured cross-sectionally and refer to the same period. Thus, previous studies cannot
determine whether household characteristics pre-dated extended family coresidence, or whether
these characteristics are a consequence of extended family coresidence. To identify predictors of
subsequent extended family coresidence, I use discrete-time event history models, implemented
with logistic regression to explore how economic capacities and family needs, experienced
prior to coresidence, are associated with the transition into living with extended family.
Discrete-time event history models model the duration until the occurrence of an event of
interest (in this case, the first time a child is observed living in an extended family household)
and estimate the effects of explanatory variables on the timing of the event. These models can
incorporate time-varying covariates, which is important, given that children’s household
characteristics can change over time, and they account for right-censoring. However, right-
censoring is not a major concern in this study, as the data includes measures for the entire time
period of interest (from age 0 up to but not including age 18) for 95% of the sample.
Equation 1 depicts the discrete-time logit model:
󰇡 
󰇢  (1)
where is the outcome of interest for child living with an extended relative,  is
the probability of an event occurring during interval t, (given that it has not occurred prior to
interval t) is a vector of functions of the cumulative duration by interval t with coefficients,
and  is a vector of the aforementioned demographic, economic, cultural, and family needs
variables with coefficients. All analyses use sample weights to account for the complex
14
multistage clustered design of the PSID sample, unequal probabilities of selection, nonresponse,
and poststratifcation to calculate weighted, nationally representative population estimates and
standard errors.
Sample
To examine the transition into an extended family household, I created person-year (by
age) files in which I specified the risk period for first observed onset of coresidence starting at
age 0 (the first full year of life) and followed children until the first time they were observed
living with an extended relative, or until the end of the observation period (up to but not
including age 18). Because the data follows children from birth, no respondents are excluded due
to left censoring. I began with a sample which included children between the ages of 18 and 25
in the most recent (2013) wave of the PSID, and who were present in at least 50% of the waves
in which they could have been observed PSID (N=4,926). This first analytic sample consists of
approximately 75% of sample children born between 1988 and 1995 and it is used to evaluate the
(unadjusted) baseline risk of first coresidence by age. I then imputed missing data on time-
invariant covariates. After excluding cases with missing data, my final analytic sample included
4,484 children (1,731 of which experience extended family coresidence), representing 65,907
person-years. Weighted data are representative of young adults born between 1988 and 1995.
It should be noted that this measurement strategy may lead to a more conservative
estimate of extended family coresidence. Children excluded from the study due to high levels of
missing reports are more likely to be Hispanic, come from lower income families, have an
unemployed parent, be born to a teenage mother, and be Catholic, all of which may be positively
related to coresidence. Additionally, among children with spells of missing reports, children
typically had complete information up to middle childhood and a prolonged spell of missing
15
information before age 18, which may downwardly bias the estimate of extended family
coresidence during adolescence. Thus, while this paper provides useful insight into children’s
lifetime experience of extended family coresidence, the prevalence of this household type will
likely be underestimated, and differentially so by factors of interest (e.g., SES).
Results
Sample Characteristics
Table 1 reports time-constant and time-varying (averaged across all person-years) sample
characteristics. About half (49%) of respondents are female and most lived in a region other than
the South (57%). White children make up the largest group in the sample (48%), Hispanic
children constitute 13%, and Black children account for 33%. There is a modest upward
distortion in the proportion of Black children in this sample. This distortion is related to the
nearly 600 Black families with young children who were identified to be dropped as part of a
larger sample size reduction in 1997, but were retained so that they could be members of the
PSID’s original Child Development Supplement (Freedman & Schoeni, 2016). To adjust for this,
all analyses use weights post-stratified to the Current Population Survey for Black children.
Eighteen percent of children had parents who did not finish high school, 36% had parents with a
high diploma, 26% had parents with some college experience, and 20% had parents with a
college degree. Most children were born to mothers between the ages of 20 to 29 (55%) or 30 to
39 (34%). Fifty-five percent of children spent the majority of their childhood years living with
both parents, 42% lived with one biological parent, and a small fraction (3.5%) lived with neither
of their parents. Most respondents spent the better part of their youth living with parents and
other household members who were in good health (though children do experience considerable
year-to-year variation in family members’ health status).
16
Prevalence of Extended Family Households
Table 2 shows the overall percentage of children who lived in an extended family, the
share of children who lived in various types of extended families, and differences by
race/ethnicity and parents’ education. For cross-validation purposes, I compared single-year
estimates from the PSID to those that I calculated from the Survey of Income Program
Participation (SIPP), one of the main sources for single-year estimates of extended family
coresidence among children (see Figure 1). For comparable years (1996, 2001, and 2009), my
PSID estimates are within approximately one percentage point of SIPP estimates, and both sets
of estimates have overlapping confidence intervals. This provides evidence that any higher
prevalence rates that I may observe when I look across the entire span of childhood are not
simply due to a difference in sampling frames or a result of peculiarities in the PSID data.
My estimates show that living in an extended family is fairly common--over one-third
(35.1%) of children lived with an extended relative at some point during childhood. This
longitudinal estimate of extended family coresidence is more than two times higher than a recent
single-year estimate of 16% (Kreider and Ellis, 2011), indicating that a substantially greater
proportion of children experience this living arrangement than previously shown by cross-
sectional data. Making use of the 25-year span of data, in results not shown, I examined whether
there is a general pattern of change in coresidence over time and across cohorts. I find that
although children in later cohorts are no more likely than those in earlier ones to experience
coresidence, there has been a statistically significant increase in coresidence over my analysis
period. While seemingly contradictory, this trend appears quite plausible when we consider the
potential influence of compositional changes in the population on prevalence rates during these
years. The overall prevalence of extended family coresidence could have increased, even if the
17
likelihood of coresidence remained constant, provided that groups for whom coresidence is more
common (e.g., non-Whites) grew as a relative share of the population. Given that racial/ethnic
minorities continue to comprise a larger segment of U.S. children (Child Trends, 2016), we
might expect to observe this trend, and over time, prevalence rates may continue to rise.
Taking into account various types of extended families, we see that living with a
grandparent, the focus of most prior research on extended family coresidence, is only slightly
more common than living with an aunt and/or uncle, and similar in prevalence to living with an
other relative. Approximately 24% of respondents lived with a grandparent, 18% lived with an
aunt or uncle, and 24% of children lived with other relatives. We also see that coresidence with
more than one type of extended family member is often occurring simultaneously or at various
points throughout childhood. This is evidenced by the fact that only 6% of children only ever
lived with a grandparent, 1% only ever lived with an aunt or uncle, and 7% only lived with an
other relative.
There are dramatic differences by race/ethnicity and SES in the percentage of children
who lived in an extended family. Approximately 58% of Black children and 35% of Hispanic
children spent time in an extended family, compared to 20% of White children. This pattern
holds true when looking at each specific type of extended family coresidence, as well Black
children are the most likely to live in extended families of all types, followed by Hispanic
children. Whites are the least likely to live in extended families of all types. A higher percentage
of children from low-SES families lived with extended relatives, relative to those from higher-
SES families. Forty-seven percent of youth whose parents did not graduate high school lived in
an extended family, compared to 39% and 35% of youth whose parents had a high school
diploma or some college, respectively. The percentage of children with college-educated parents
18
who experienced extended family coresidence is nearly three times lower (17%) than those
whose parents did not finish high school. Similarly, a much higher percentage of children whose
parents had less education lived in extended families of all types, compared to peers whose
parents had more education.
What is most striking are the differences in coresidence when we consider the
intersection of race and SES in Figure 2. The percentage of Black children with college-educated
parents who lived with an extended relative (39.1%) is higher than the percentage of White
children whose parents who did not graduate high school (37.9%). In a similar vein, 26% of
Hispanic children whose parents have a Bachelor’s degree or more lived in an extended family,
relative to 22.8% of White children whose parents earned a high school diploma. These gaps
persist when we consider racial/ethnic and education differences in coresidence among various
types of extended families, such as grandparents, aunts/uncles, and other relatives (results not
shown). Generally speaking, a higher percentage of Black and Hispanic children whose parents’
education levels are at the top of the education distribution lived in an extended family, relative
to White children whose parents’ education is at the bottom of the education distribution.
This finding differs from research on racial/ethnic and SES differences in family
involvement (e.g., offering advice and giving money), which reports similar levels of
involvement among racial/ethnic groups from the same social class (Gerstel, 2011; Sarkisian and
Gerstel, 2012). Although related, coresidence is a qualitatively distinct type of support that may
be determined, at least in part, by different underlying mechanisms than family involvement.
This may be especially true for types of involvement that are not contingent on geographic
proximity to relatives. For example, technologies such as smart phones and wire transfers make
it possible for middle class Blacks and Whites to offer advice or transfer money to extended
19
relatives at similar rates, regardless of their proximity to extended relatives. Spatial distance,
however, does limit the probability of coresidence, and racial/ethnic minorities are more likely to
live closer to extended relatives than Whites (Cherlin and Furstenberg, 1986; Connidis, 2001).
Therefore, it is possible for middle class minorities to offer housing assistance to family
members at higher rates than middle class Whites, even if they engage in similar levels of other
types of support.
Descriptive Hazard Probabilities
Figure 3 presents the model-based predicted probability of first observed transition into
an extended family household. These estimates are based on unadjusted weighted hazard models
including only age as a covariate. These probabilities indicate the yearly probability that a child
will first experience the outcome (in this case, extended family coresidence) at a specific age,
given that he or she has not yet done so. Overall, children are most likely to experience first
extended family coresidence at younger ages. The risk of first moving into an extended family is
highest between birth and age 1 and it drops steadily until age 9. This finding is similar to studies
of three-generation coresidence, which find that multigenerational households are most common
when children are young (Bryson and Casper, 1999; Mutchler and Baker, 2004; Pilkauskas,
2012). The rate of risk, however, follows a slightly U-shaped pattern. During middle childhood
(ages 9 to 12), the hazard rate is relatively low and stagnant, and then it slightly increases during
adolescence (ages 13 to 17). Thus, a non-negligible percentage of children (27%) who lived in an
extended family began doing so at older ages (not shown in tables). Confidence intervals
constructed for these hazard rates (not shown in tables) indicate that this rise in the probability of
coresidence during adolescence is statistically significant and that this pattern of risk is
consistent across racial/ethnic groups. These findings support my assertion that looking at
20
extended family coresidence across childhood may reveal higher rates than cross-sectional
estimates or estimates that focus on early childhood as the key risk period. They may also
suggest periods of time when parents or extended family members most need help. While a full
examination of differences in predictors by developmental stage is beyond the scope of this
paper, research has suggested that coresidence during early childhood may be motivated by a
need for childcare, whereas coresidence during adolescence may be more related to the needs of
extended relatives (e.g., aging grandparents needing assistance from their adult children) (Cohen
& Casper, 2002; Pilkauskas, 2012), or it may represent another challenging period for parents,
who may rely on extended relatives for additional help with childrearing.
Predictors of Extended Family Households
Table 3 displays the odds ratios from discrete time hazard models predicting the
transition into an extended family. It summarizes results from a full model that includes all
covariates. To assess whether economic capacities and family needs were differentially
associated with extended family coresidence by race/ethnicity, I also ran the models separately
for Black, White, and Hispanic children and then conducted Chow tests on the fully interacted
model that compared each group (White vs. Black, White vs. Hispanic, Black vs. Hispanic).
Significant differences by race/ethnicity in the factors that predict the transition into an extended
family are indicated with footnotes.
Considering indicators of economic capacity, we see that children who experienced
greater disadvantage during childhood have a higher risk of living in an extended family (My use
of the term “risk” refers to the yearly conditional odds that a child will experience coresidence; it
does not reflect a preference for or against this living arrangement). In general, children raised in
households below 400% of the poverty threshold had higher rates of entry into extended families
21
than children who grew up in households above 400% of the poverty threshold. Odds ratios for
youth living in households at or below 100% of the poverty line indicate that they are not
statistically different from those from households above 400% of poverty. This may reflect a
lack of available extended relatives with stable and/or attractive housing for this group. As prior
research has shown, the availability of extended family support via coresidence is conditioned by
the economic situation of members of one’s extended family network (Trent & Harlan, 1994;
Roschelle, 1999; Cohen & Casper, 2002). The poorest members of society may not have the
opportunity to live in extended households because they are drawing on the resources of poor
network members, who are not in a position to offer this type of support. At every level of
education, the relative risk of living with extended family is higher for children whose parents
had less than a Bachelor’s degree, compared to those who had college-educated parents. Youth
whose parents were unemployed or out of the labor force were also at greater risk of entering
into an extended family household, relative to those with employed parents. Further, youth who
lived in household units that were rented were at lower risk of living with extended relatives,
relative to youth who lived in owned homes.
With respect to family needs, mother’s age at child’s birth, the child’s age, the number of
parents present in the home, and the health status of household members all increase the risk of
subsequent first-time extended family coresidence. Children born to teenage mothers have an
approximately 60%-70% higher risk of moving into an extended family, relative to those born to
mothers aged 20 or older. Younger children are also at higher risk of experiencing this family
structure. Compared to youth who live with both parents, youth who live with one or neither
parent have significantly higher rates of transition into this household type. Moreover, children
22
who have parents and other household members who are not in good health are at a substantially
higher risk of experiencing subsequent extended family coresidence.
In the full sample, I find that after accounting for both economic factors and family
needs, racial/ethnic minorities have significantly higher rates of living in extended families than
Whites. Black children, on average, are more than twice as likely to experience this family
structure, and Hispanic children are more than 1.5 times as likely to do so, relative to Whites.
When I examine group differences in predictors by race/ethnicity, I find that overall, there is no
pattern of statistically significant differences across groups. The exceptions are, region, housing
tenure, and one category of household income. Whereas region is inconsequential in the full
sample, White children raised in the South live in an extended family at higher rates than their
White peers raised outside of the South, but Black and Hispanic children in the South do not
have higher rates than their otherwise similar peers. In the full model, children living in a rented
home are at lower risk of living with extended family compared to peers living in owned homes;
however, this factor is only significant for non-White children. Finally, among all children, those
whose household income is 201%-300% of the poverty threshold are more likely to live with
extended relatives than those whose household income is 400% or above the poverty threshold;
this factor is not significant for Black children. These results demonstrate that while these factors
operate similarly across groups, level differences in economic capacities and family needs help
account for racial/ethnic differences in prevalence rates.
In results not shown, I also evaluated the extent to which predictors differ by type of
extended relative (grandparent vs. aunt, uncle, or other relative). This supplemental analysis was
motivated by the idea that reasons for coresidence may be qualitatively different by relative type.
I find few statistically significant differences in predictors. Children who are first observed living
23
with a grandparent are: (1) more likely to be living with one parent, (2) less likely to be born to a
mother aged 40 or older, and (3) less likely to have a parent whose religious preference is a
denomination other than Catholic or Protestant, compared to children who are first observed
living with an aunt, uncle, or other relative.
Discussion
This article examines the prevalence and predictors of extended family households
among a recent birth cohort of children and explores racial/ethnic and SES differences in this
living arrangement. It builds on prior literature that has largely used cross-sectional estimates and
has focused on multigenerational households, to consider extended families more broadly and
longitudinally. By including a broader set of extended relatives, and looking across childhood, I
find that extended family households are a fairly common living arrangement for children: over 1
in 3 youth spend some time in an extended family before age 18. Taking into account various
types of extended families, estimates show that children are almost equally likely to live with a
grandparent, aunt or uncle, or other relative, with 24% of children having lived with either a
grandparent or other relative, and nearly 20% of children having spent some time living with an
aunt or uncle. These estimates highlight the complexity of this household type and confirm that
using single-year data and focusing on grandparent coresidence does not provide a full picture of
children’s experience living in an extended family.
When the whole span of childhood is considered, the percentage of Black children who
live with an extended relative (58%) is nearly three times higher, and the percentage of Hispanic
children who live with an extended relative is approximately 1.5 times higher (35%) than the
percentage of White children (20%) who live in an extended family. This disparity is
substantially larger than what previous cross-sectional studies have observed (e.g., Kreider &
24
Ellis, 2011). Moreover, at every level of SES, a much greater proportion of Black and Hispanic
youth have lived with an extended relative, compared to their White peers. This finding may
reflect racial/ethnic differences in the socioeconomic composition of extended family networks.
Higher-SES minorities may have more low-SES extended relatives than their White
counterparts, which increases the chance that they will offer housing assistance to an extended
family member. Indeed, using the same data, Heflin and Patillo (2004) found that middle class
Blacks on average have more siblings than Whites, and they are more likely to have a poor
sibling, which increases the likelihood that middle class Blacks will have a relative turn to them
for assistance and subsequently live in an extended family.
Together, these findings on racial/ethnic and SES differences in extended family
coresidence shed light on the salience of extended family households for minority children
and/or those from disadvantaged backgrounds. They are particularly important, given that
nuclear family households have long been considered the normative and standard household type
in the U.S., and that the White middle class experience of family structure is typically treated as
the baseline experience to which all other groups are compared (Coontz, 1992; Kamo, 2000
Gerstel, 2011). Here, we see that extended family households are widespread, and that this
family type is only atypical for high-SES White families. Thus, a narrow focus on the nuclear
family structure overlooks the diverse ways in which families, particularly those from minority
and/or disadvantaged backgrounds constitute household living arrangements, and family life
more generally. Further, as racial/ethnic minorities continue to make up a larger share of the U.S.
population, extended family households will likely become increasingly widespread, and given
the potential positive and/or negative consequences associated with living in an extended family
(Dunifon, 2013; Garrett-Peters & Burton, 2016), this oversight may increasingly limit our
25
understanding of the effect of family structure on child wellbeing and the ways in which
extended relatives may help exacerbate or reduce racial/ethnic and class disparities in child
outcomes.
Turning to predictors of extended family households, I find strong evidence for the role
of economic capacities and family needs as key determinants of coresidence. Consistent with
prior literature on extended family living arrangements (e.g., Kamo, 2000; Cohen & Casper,
2002; Pilkauskas, 2012) economic factors such as parents’ educational attainment, household
income, and having employed parents are negatively related to the risk of living in an extending
family. Counter to expectation, children who lived in a rented household unit were less likely to
live with an extended relative than peers living in owned homes. By way of explanation,
individuals in need typically live with relatives best suited to host extended family members
(Cohen & Casper, 2002). Children living in owned homes may be more likely to live in an
extended family because their parents’ home ownership better positions them to provide stable
housing assistance to relatives in need than children of renters. In terms of family needs, being
born to a teenage mom, being a young child yourself, having at least one parent absent from your
home, and having a household member who is not in good health are strong, positive predictors
of subsequent extended family coresidence. These results appear consistent with a life course
pattern of the need for child care. Younger mothers with young children, especially those not
living with the child’s father may rely more heavily on extended relatives for childcare
assistance. Combined, these findings provide further support that the transition into an extended
family is largely a response to social and economic need. Finally, aside from SES, I found very
few significant differences in predictors across racial/ethnic groups. Thus, I find little evidence to
support the assertion that resource-driven motivations to coreside differ by race/ethnicity.
26
There are some limitations to this study that should be noted. First, extended family
households are often short-lived. In fact, in my sample, among children who were first observed
living in an extended family, approximately 30% were not coresiding by the next wave and 45%
were no longer doing so two waves later (results not shown). Because of the highly transient
nature of extended family households, short term residence may be missed between PSID waves.
(Mollborn et al., 2012; Pilkauskas, 2012). Second, this analytic sample is based on children born
between 1988 and 1995, many of whom were present in the PSID prior to its immigrant refresher
wave, (500 immigrant-headed households were added to the PSID in 1997 to account for post-
1965 immigration), and whose parents were born before numerical restrictions to U.S.
immigration were lifted in 1965. Thus, sample estimates may not be representative of the
experience of contemporary youth whose families entered the U.S. during the most recent
immigration waves, and who may be more likely to live in an extended family. Additionally,
children excluded from this analysis due to high levels of missing waves were more likely to be
members of demographic groups who have increased odds of living in an extended family (e.g.,
low-SES children) and/or to have long spells of missing reports during adolescence. Therefore,
the figures presented here may underestimate the percentage of children who have lived in an
extended family, and gaps by race/ethnicity and SES may be even wider. Third, while this study
makes an important step towards identifying predictors of coresidence, due to data limitations, it
does not distinguish between the movement of children into the household of extended family
members and vice versa. Previous studies indicate that when individuals host extended relatives,
they are less likely to be receiving assistance and more likely to be providing it, which may have
consequences for child wellbeing (Alquilino, 1990; Jayakody et al., 1993; Cohen & Casper,
2002; Grundy, 2005). If the child’s immediate family is in need, then we might expect that the
27
resources flowing to them via coresidence might improve child outcomes. However, if an
extended family member moves into the household of the child’s immediate family, this may
divert resources away from the child, potentially undermining his or her wellbeing. Future work
focused on the extended family structure and child wellbeing should consider differentiating
between these two circumstances. Finally, given the data available, I am unable to include all
variables that may be indicators of cultural norms and preferences. In particular, I would have
liked to include a measure of the primary language spoken at home and indicators of familial
attitudes. Future research with more robust cultural indicators should explore if and how these
cultural factors are related to the transition into an extended family.
Despite these limitations, this study is the first to use nationally representative data to
document the prevalence of extended family households across childhood and to identify factors
predicting this living arrangement. Additionally, this study is unique in examining racial/ethnic
differences in the predictors of coresidence. To the extent that extended relatives play a role in
child development and wellbeing, this is an important phenomenon that has implications for both
research and public policy.
28
References
Angel, R. and M. Tienda. 1982. Determinants of extended household structure: cultural pattern
or economic need? American Journal of Sociology 87: 1360-1383.
Baca Zinn, M. 1994. Adaptation and continuity in Mexican-origin families, in R. L. Taylor (ed),
Minority Families in the United States: A Multicultural Perspective. Englewood Cliffs, NJ:
Prentice Hall, pp. 6481.
Beck, R. W. and S.H. Beck. 1989. The incidence of extended households among middle-aged
black and white women: estimates from a 15-year panel study. Journal of Family Issues 10:
14768.
Beck, S. H. and R.W. Beck. 1984. The formation of extended households during middle age.
Journal of Marriage and Family 46: 27787.
Bengtson, V. L. 2001. Beyond the nuclear family: the increasing importance of multigenerational
bonds. Journal of Marriage and Family 63: 116.
Bras, H., J. Kok, and K. Mandemakers. 2010. Sibship size and status attainment across contexts:
evidence from the Netherlands, 1840-1925. Demographic Research 23: 73-103.
Brown, S. L. 2010. Marriage and child well-being: research and policy perspectives. Journal of
Marriage and Family 72: 105977.
Bryson, K. and L.M. Casper. 1999. Coresident grandparents and grandchildren. Current
Population Reports P23-198, U.S. Census, Bureau, Washington, DC.
Burr, J. A. and J.E. Mutchler. 1993. Ethnic living arrangements: cultural convergence or cultural
minifestation? Social Forces 72: 169-179.
Burton, L. M. 1992. Black grandparents rearing children of drug-addicted parents: stressors,
outcomes, and social service needs. Gerontologist 32: 74451.
Carlson, M.J., A. VanOrman, and N.V. Pilkauskas. 2013. Examining the antecedents of U.S.
nonmarital fatherhood. Demography 50: 1421-1447.
Cherlin, A. and F. F. Furstenberg. 1986. The New American Grandparent: A Place in the Family,
a Life Apart. New York, NY: Basic Books.
Child Trends. 2016. Racial and ethnic composition of the child population. Retrieved from:
https://www.childtrends.org/wp-content/uploads/2016/07/60_Racial-Composition.pdf
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.
Cohen, P. N. and L.M. Casper. 2002. In whose home? Multigenerational families in the United
States, 1998-2000. Sociological Perspectives 45:1-20.
Connidis, I. 2001. Family Ties and Aging. Thousand Oaks, CA: Sage
Coontz, Stephanie. 1992. The Way We Never Were: American Families and the Nostalgia Trap.
New York, NY: Basic Books.
Deleire, T. and A. Kalil. 2002. Good things come in threes: single-parent multigenerational
family structure and adolescent adjustment. Demography, 39: 393413.
Department of Health and Human Services 2012. Annual update of the HHS poverty guidelines.
Federal Register 77: 1-2.
Dunifon, R. 2013. The influence of grandparents on the lives of children and adolescents. Child
Development Perspectives 7:55-60.
Dunifon, R. and L. Kowaleski-Jones. 2007. The influence of grandparents in single-mother
families. Journal of Marriage and Family 69:465481.
Edin, K. and L. Schaefer 2015. 2.00 a Day: Living on Almost Nothing in America. Boston, MA:
29
Houghton Mifflin Harcourt.
Ellis, R.R., and T. Simmons. 2014. Coresident grandparents and their children: 2012. Current
Population Reports P20-576, U.S. Census Bureau, Washington, DC.
Foster, M.E. and A. Kalil. 2007. Living arrangements and children's development in low-income
white, black, and Latino families. Child Development 78:1657-1674.
Freedman, V.A. and R. Schoeni. 2016. Generating nationally representative estimates of multi-
generational families and related statistics for blacks in the PSID. PSID Technical Paper
Series, Paper #1601.
Garrett-Peters, R. and L.M. Burton. 2016. Tenuous ties: the nature and costs of kin support
among low-income rural black mothers. Women, Gender, and Families of Color 4: 435.
Gerstel, N. 2011. Rethinking families and community: the color, class, and centrality of extended
kin ties. Sociological Forum 26: 120.
Giordano, P. C., S.A. Cernkovich, and A. DeMaris. 1993. The family and peer relations of black
adolescents. Journal of Marriage and Family 55:277-287.
Glick, J.E. 2000. Nativity, duration of residence and the life course pattern of extended family
living in the USA. Population Research and Policy Review 19: 17998.
Goldscheider, F.K. and R. Bures. 2003. The racial crossover in family complexity in the United
States. Demography 40: 569-587.
Goldscheider, F. K. and C. Goldscheider. 1996. Leaving home before marriage: ethnicity,
familism, and generational Relationships. Madison, WI: University of Wisconsin Press.
Hays, W.C. and C.H. Mindel. 1973. Extended kinship relations in black and white families.
Journal of Marriage and Family 35: 51-57.
Heflin, C.M., and M. Patillo. 2004. Poverty in the family: race, siblings, and socioeconomic
heterogenity. Social Science Research 35: 804-822.
Hill, M.S., W.J. Yeung, and G.T. Duncan. 2001. Childhood family structure and young adult
behaviors. Journal of Population Economics 14: 271-299.
Hirsh, B. J., M. Mickus, and R. Boerger. 2002. Ties to influential adults among black and white
adolescents: culture, social class, and family networks. American Journal of Community
Psychology 30: 289-303.
Hogan, D. P., L. Hao, and W.L. Parish 1990. Race, kin networks, and assistance to mother-
headed families. Social Forces 68: 797 812.
Hunter, A. 1997. Counting on grandmothers: Black mothers’ and fathers’ reliance on
grandmothers for parenting support. Journal of Family Issues 18: 251-269.
Kalil, A., R. Ryan, and M. Carlson. 2014. Time investments in children across family structures.
The ANNALS of the American Academy of Political and Social Science 654: 150-168.
Kamo, Y. 2000. Racial and ethnic differences in extended family households. Sociological
Perspectives 43: 21129.
Kamo, Y. and M. Zhou. 1994. Living arrangements of elderly Chinese and Japanese in the
United States. Journal of Marriage and Family 56: 554-558.
Kang, J. and P. Cohen. Extended kin and children's behavioral functioning: family structure and
parental immigrant status. Social Science and Medicine 186: 61-69.
Kreider, R. M. and R. Ellis. 2011. Living arrangements of children : 2009. Current Population
Reports 70126.
Manning, W., S. Brown, and J.B. Stykes. 2014. Family complexity among children in the United
States. The Annals of the American Academy of Political and Social Science 654: 48-65.
Mollborn, S., P. Fomby, and J.A. Dennis. 2011. Who matters for children’s early development?
30
race/ethnicity and extended household structures in the United States. Child Indicators
Research 4: 389411.
Mollborn, S., P. Fomby, and J.A. Dennis. 2012. Extended household transitions, race/ethnicity,
and early childhood cognitive outcomes. Social Science Research 41: 1152-1165.
Mutchler, J. E. and L.A. Baker. 2004. A demographic examination of grandparent caregivers in
the Census 2000 Supplementary Survey. Population Research and Policy Review 23: 359-
377.
Musa, D., R. Schulz, R. Harris, M. Silverman, and S. Thomas. 2009. Trust in the health care
system and the use of preventive health services by older black and white adults. American
Journal of Public Health 99: 1293-1299.
Mutran, E. 1985. Intergenerational family support among blacks and whites: response to culture
or to socioeconomic differences. Journal of Gerontology 40: 382-389.
Neighbors, H.W., C. Caldwell, D.R. Williams, R. Nesse, R.J. Taylor, K.M. Bullard, M. Torres,
and J.S. Jackson. Race, ethnicity, and the use of services for mental disorders: results
from the National Survey of American Life. Archives of General Psychiatry 64: 485-
494.
Oropesa, R. S. and N.S. Landale. 2004. The future of marriage and Hispanics. Journal of
Marriage and Family 66: 901-920.
Pernice-Duca., F.M. 2010. An examination of family social support networks as a function of
ethnicity and gender: a descriptive study of youths from three ethnic reference
groups. Journal of Youth Studies 13: 391-402.
Pilkauskas, N. V. 2012. Three-generation family households: differences by family structure at
birth. Journal of Marriage and Family 74: 93143.
Pilkauskas, N. V. 2014. Living with a grandparent and parent in early childhood: associations
with school readiness and differences by demographic characteristics. Developmental
Psychology 50: 2587-2599.
Pittman, Laura D. 2007. Grandmothers’ involvement among young adolescents growing up in
poverty. Journal of Research on Adolescence 17: 89-116.
Richards, T., M.J. White, and A.O. Tsui. 1987. Changing living arrangements: a hazard model of
transitions among household types. Demography 24: 77-97.
Roschelle, A.R. 1999. Gender, family structure, and social structure: racial ethnic families in the
United States, in M. M. Ferree, J. Lorber, and B. B. Hess (eds), Revisioning Gender.
Thousand Oaks, CA: Sage, pp. 311-340.
Sarkisian, N., M. Gerena, and N. Gerstel. 2006. Extended family ties among Mexicans, Puerto
Ricans, and Whites: superintegration or disintegration? Family Relations 55: 331-344.
Sarkisian, N. and N. Gerstel. 2012. Nuclear Family Values, Extended Family Lives: The Power
of Race, Class, and Gender. New York, NY: Routledge.
Stack, C. B. 1974. All Our Kin: Strategies for Survival in a Black Community. New York, NY:
Harper and Row.
Stack, C.B. and L.M. Burton. 1993. Kinscripts. Journal of Comparative Family Studies 24: 157-
170.
Trent, K. and S.L. Harlan. 1994. Teenage mothers in nuclear and extended households:
differences by marital status and race/ethnicity. Journal of Family Issues 15: 309337.
U.S. Census Bureau. 2011. Survey of Program Participation 2008 Panel Topical Module 4.
Retrieved from: https://www.census.gov/programs-surveys/sipp/data.html.
31
U.S. Census Bureau. 2017. Survey of Program Participation 2014 Panel Wave 1. Retrieved
from: https://www.census.gov/programs-surveys/sipp/data/2014-panel/wave-1.html
Van Hook, J. and J. Glick. 2007. Immigration and living arrangements: moving beyond
economic need versus acculturation. Demography 44: 225-249.
Woodward, A., L.M. Chatters, R.J. Taylor, H. Neighbors, and J.S. Jackson. 2010. Differences in
professional and informal help seeking among older African Americans, Black
Caribbeans, and non-Hispanic whites. Journal of the Society for Social Work and
Research 1: 124-130.
32
Table 1. Sample characteristics of children born between 1988 and
1995, Panel Study of Income Dynamics 1988-2013 (N=4,484)
Variables
M or %
Demographic Factors
Female
48.94
Race/Ethnicity
White
47.87
Black
33.37
Hispanic
12.89
Other race
5.98
Region
South
43.04
Non-South
56.96
Parents' religious preference
Catholic
33.18
Protestant
53.49
Other religion
4.25
No preference
8.84
Economic Capacities
Income-to-needs ratio
At or below 100% of poverty threshold
9.43
101%-200% of poverty threshold
27.76
201%-300% of poverty threshold
23.18
301%-400% of poverty threshold
18.50
Above 400% of poverty threshold
21.13
Parents' education level
Less than high school
18.22
High school
36.10
Some college
25.81
Bachelor's degree or higher
19.77
Home ownership
Family owns home
49.76
Family rents home
48.29
Family neither owns nor rents home
1.95
Parent's employment status
Both parents employed
49.21
At least one parent unemployed
33.48
At least one parent out of labor force
17.32
Family needs
Mother's age at birth
19 and under
9.32
20-29
54.53
30-39
33.56
40+
2.44
Child's age (mean, range 1-17)
7.89
(SD)
(5.22)
No. of children in household (mean, range 0-11)
2.43
(SD)
(1.10)
No. of parents in household
Both parents
54.55
One parent
41.99
Neither parent
3.46
33
Table 1. Sample characteristics of children born between 1988 and
1995, Panel Study of Income Dynamics 1988-2013 (N=4,484) Cont.
Variables
M or %
Health of household members
Parents in good health
93.15
Other household members in good health
96.10
Observations
4484
Notes: Values are percentages unless otherwise noted. Analysis uses sample
weights to account for the complex multistage clustered design of the PSID
sample. Total for the time-varying characteristics is the person-year average.
No.=Number. SD=standard deviation.
34
Table 2. Percentage of children ever living in extended family households by race/ethnicity and parents'
education, Panel Study of Income Dynamics 1988-2013 (N=4,484)
Total
Race/ethnicity
White
Black
Hispanic
Other race
Parents' education level
Less than high school
High school
Some college
Bachelor's degree or
higher
Lived only with this type of
extended relative during childhood 6 1 7
Notes: Analysis uses sample weights to account for the complex multistage clustered design of the PSID sample.
35
Notes: Estimates are weighted to account for the complex multistage clustered design of
the PSID sample. All estimates contain overlapping confidence intervals for comparable
years.
13.01% 14.81% 16.09%
13.85% 13.80% 14.79%
0.0%
2.0%
4.0%
6.0%
8.0%
10.0%
12.0%
14.0%
16.0%
18.0%
1996 2001 2009
Figure 1: Comparison of Single-Year Estimates of
Children's Extended Family Coresidence,
PSID (1996-2009), SIPP (1996-2009)
SIPP PSID
36
38%
69%
40%
23%
61%
35%
19%
54%
35%
9%
39%
26%
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
White Black Hispanic
Fig. 2. Percentage of children living in an extended family
by both race/ethnicity and parents' education
Less than high school High school Some college Bachelor's degree or higher
37
0.13
0.05 0.02
0.02 0.01 0.01
0.02 0.03
0.00
0.02
0.04
0.06
0.08
0.10
0.12
0.14
0.16
0 3 6 9 12 15 18
Fig. 3. Risk of first time living in an extended family by age,
Panel Study of Income Dynamics (1988-2013)
Age
Hazard Rate
38
Table 3: Odds ratios from discrete time hazard models for the risk of first time living in an
extended family, Panel Study of Income Dynamics, 1988-2013
Full Sample
B
SE
OR
Demographic factors
Female
0.11*
0.06
1.12*
Race/Ethnicity (vs. White)
Black
0.86***
0.08
2.37***
Hispanic
0.51***
0.12
1.66***
Other
0.57***
0.14
1.77***
South (vs. Non-South)ac
0.08
0.07
1.08
Parents' religious preference (vs. Catholic)
Protestant
-0.20**
0.09
0.82**
Other religion
0.13
0.15
1.14
No religious preference
-0.07
0.13
0.94
Economic capacities
Poverty threshold (vs. at or above 400% of poverty threshold)
At or below 100% of poverty threshold
0.15
0.13
1.16
101%-200% of poverty threshold
0.35***
0.12
1.41***
201%-300% of poverty thresholdabc
0.23*
0.12
1.26*
301%-400% of poverty threshold
0.28**
0.13
1.33**
Parents' Education (vs. BA or higher)
Less than high school
0.47***
0.13
1.60***
High school
0.46***
0.11
1.59***
Some college
0.40***
0.11
1.50***
Home ownership (vs. own)
Rentac
-0.32***
0.07
0.73***
Neither own or rentac
-0.06
0.13
0.94
Parents' employment status (vs. both employed)
At least one parent unemployed
0.41***
0.09
1.51***
At least one parent out of labor force
0.21***
0.07
1.24***
Family needs
Mother's age at birth (vs. 19 and under)
20-29
-0.84***
0.10
0.43***
30-39
-1.13***
0.11
0.32***
40+
-1.14***
0.28
0.32***
Child's age
-0.22***
0.02
0.80***
No. of children in household
0.00
0.03
1.00
No. of parents in HH (vs. both parents)
Single parent
1.33***
0.07
3.78***
Neither parent
1.56**
0.62
4.74**
Parents in good health
-0.25***
0.09
0.78***
Other HH members in good health
-0.28**
0.13
0.76**
Number of person-years
65,097
*** p<0.01, ** p<0.05, * p<0.1
Notes: All models include a set of time-varying dummy variables for single years of age. Analysis uses sample weights to account for the
complex multistage clustered design of the PSID sample. No=Number. aWhite-Black difference in coefficient is significant at p <.05.
bBlack-Hispanic difference in coefficient is significat at p<.05. cHispanic-White difference in coefficient is significant at p<.05.
39
... Demographers and family sociologists have established that family instability, defined as parental relationship dissolution and repartnering, has negative consequences, on average, for child and adolescent behavior, cognitive scores, and educational attainment (Cavanagh and Fomby 2019). Approximately 35 percent of children live with an extended family member at some point during childhood (Cross 2018); shared households that include adults other than the head of household and their romantic partner experience frequent changes in composition (Pilkauskas 2012). Thus, beyond changes in parental relationships, a substantial share of children experiences changes in household composition involving extended family members and nonrelatives (Perkins 2017;Raley et al. 2019). ...
... Marriage is also less protective against poverty for Black mothers compared with White and Latinx mothers (Williams and Baker 2021). Household composition beyond the nuclear family, selection into changes in household composition, and the effects of these changes may differ by race (Cross 2018;Mollborn et al. 2012;Perkins 2017Perkins , 2019 and our predictions of such changes and estimates of their effects should account for different selection mechanisms. ...
... Most households in the PSID contain only one family unit, but for the approximately 10 percent of households that contain more than one family unit I must use the variable identifying the relationship between heads of different family units to infer relationships within households, but across family units. I use a four-category measure of exposure to household change involving parents, extended family, and nonrelatives based on observing children's households through age seventeen: first, experienced household changes involving only parents and stepparents; second, experienced household changes involving only nonparents (that is, adult siblings age twenty-five and older, and extended family and Cross (2018) finds that more than one-third of children lived with an extended relative at some point during childhood and that coresidence often occurs simultaneously with more than one type of extended family member, further justifying an approach that considers household members beyond parents. nonrelatives, including children and adults); third, experienced household changes involving both parents and nonparents, and, fourth, experienced no household changes. ...
... Black individuals are more likely to reside in households comprising members beyond the nuclear family model than White people. For example, Black children live with extended family members at a rate almost three times that of White children, with much of this coresidence occurring due to socioeconomic status (SES) constraints (Cross, 2018). ...
... Within these overall trends, there are large differences by race, potentially placing Black individuals at greater risk of exposure to household member deaths. For example, although Cross (2018) found that overall, 35 percent of children ever coresided with extended kin, these rates are much higher for Black (57%) than for White children (20%). Moreover, Black children (15%) are more than twice as likely to live with a grandparent than White children (7%); the disparity in the likelihood of a grandparent being the child's primary caregiver is even greater (8% vs. 3%), potentially resulting in increased risk of exposure to household member deaths given that most deaths occur at older ages (Pew Research Center, 2013). ...
... Research suggests there is a nonrandom selection of families into coresident households, with socioeconomic constraints as a primary driver of the formation of these households (Cross, 2018). Coresident households tend to have higher poverty rates and lower levels of education and income (Dunifon et al., 2014). ...
Article
Full-text available
As a result of Black–White inequities in life expectancy, recent research has indicated that Black individuals are disproportionately exposed to the deaths of multiple family members compared to White individuals. Black individuals are also more likely to live in coresident households—that is, households that extend beyond the nuclear family. However, it is unclear the degree to which this population may be disproportionately exposed to the loss of deaths marked by the geographic closeness of the household. In this study, I use data from the Panel Study of Income Dynamics to provide the first nationally representative estimates of Black-White disparities in exposure to household member deaths. I find that Black people are significantly more likely than White individuals to have experienced the death of a household member. Based on these findings, I argue the dual inequities of racial disparities in life expectancy and racial disparities in coresidence are an overlooked, salient source of racial disparities in exposure to death. By illuminating a broader range of network sources that contribute to racial inequities in exposure to death, this study sets forth a new conceptual unit of analysis—that of the household—to investigate the intergenerational reproduction of inequality in health and socioeconomic status due to exposure to death.
... To break free from an abusive and unsafe home environment, some participants opted to sever ties with their spouses, leading to the eventual solo parenting and acquisition of their unique role. Due to the additional responsibilities bestowed upon the solo mothers, limitation in resources compelled them to create extended family households to gain benefits from resource pooling (Cross, 2018). In addition, most of them often relied on the help of their elderly parents in child-rearing responsibilities, leading to the multigenerational living arrangements (Herbst-Debby, 2018). ...
Article
Full-text available
This study explored the experiences of solo parents who provide simultaneous support to their minor children and senior parents in selected resettlement areas in Cavite, Philippines. Specifically, it sought to determine the reasons for their role pursuance, unravel the adversities they encountered and the coping strategies they employed, and identify their motivations for continuously juggling their multigenerational responsibilities. Using narrative analysis, the stories of eight participants selected through non-proportional quota sampling were uncovered and interpreted. Findings revealed that the high regard they placed on cultural family values and traditions drove them to become caregivers for two generations. The demands of having young and old dependents led them to face financial constraints, unfulfilled aspirations, work interference, and health-compromising behaviors associated with their role as family mediators. Social support systems and multiple sources of livelihood played a significant part in easing these burdens. Despite the hardships entailed in their household responsibilities, they managed to carry on due to the long-term care needs of family members and the positive feelings they acquired from caregiving. This may further support the need for implementation or amendment of policy programs intended to provide recognition and workplace consideration for individuals in this unique familial position.
... Social networks are key sources of support when households cannot afford their housing or are forced to move due to various shocks such as neighborhood violence, housing quality failures, or changes in income or employment (Clampet-Lundquist 2003;Edin and Shaefer 2015;Mazelis 2017). Shared housing arrangements, or doubling up, is a common strategy (Cross 2018;Harvey, Dunifon, and Pilkauskas 2021). Although such arrangements have short-term benefits, including preventing households from living on the street, in cars, or in shelters, research increasingly shows that doubling up can create fraught dynamics between hosts and guests, strain hosts' space and finances, and potentially lead to adverse effects later in life (Harvey 2020a(Harvey , 2020b. ...
... Black families have many strengths; for instance, individuals who identify as Black often place a high value on family as a social support network and regularly receive and provide support to nuclear and extended family members . Indeed, much higher percentages of Black children (57%) spend some time in extended family households than do White (20%) or Hispanic (35%) children (Cross, 2018). Given that Black families have also historically been underserved by helping professions, including disability services (Hassiotis, 2020), it follows that Black families may take on greater responsibilities caring for family members with disabilities. ...
Article
Although natural supports benefit individuals with intellectual and developmental disabilities (IDD), little is known about natural support provided within specific life domains or how race/ethnicity or support from professionals impacts the extent of natural support one receives. In this study, 518 parents of adults with IDD responded to a national survey about natural supports, including who provides support, the number of supporters, and variables that predict natural supports. Family most often provided support, although professionals and family friends were frequent supporters in several domains. Natural support was most extensive in health, least extensive in employment and housing. Individuals with IDD who regularly participated in daytime activities and/or identified as Black had more extensive natural support. Implications are discussed.
Article
Although the challenges of bereavement among adolescents are well-documented, there is potential for loss to foster growth, development, and positive psychological changes during this developmental period. To examine this possibility, 408 adolescents (46.6% male; 53.4% female) in grades 6-12 from three urban schools were surveyed about their experiences with bereavement and posttraumatic growth (PTG), along with several hypothesized moderators: type of loss, relationship with the deceased, active and ruminative coping, and social support. Seventy-six percent of the participants reported signs of PTG. Hierarchical linear regression analyses revealed significant main effects for bereavement, active coping, and social support on PTG, but found no evidence of moderation for any variables. The findings are discussed in the context of systemic racism and cultural strengths, with directions for future research provided.
Article
Objective The current study asks how race, class, and the social pressure to care for children who are not one's own impact how childless Black women experience and make meaning of their parental status. Background While much of the existing qualitative research on childlessness has asked how white, middle‐class women experience social pressure to have children, this study shifts this focus to ask how Black women experience social pressure to care for children who are not their own. Method The study relied on virtual interviews with 40 class‐diverse childless Black women between 40 and 55. The recruitment strategy utilized a mixed‐method approach, including snowball sampling, online and physical advertising, and targeted outreach within online communities for Black women. Interviews were transcribed and iteratively analyzed to identify thematic codes and categories. Results The study identifies and terms “compulsory mothering” as a significant social pressure that leads childless Black women to assume caregiving roles within their kin networks, regardless of their parental status. This pressure is more pronounced among working‐class women, who engage more extensively in these roles compared to their middle‐class counterparts, highlighting how race and class intersect to shape their experiences of childlessness. Conclusion The study concludes that race and class influence women's experiences of childlessness. It also finds that childless Black women's relationships with their kin networks shape their experiences of childlessness and their reproductive preferences.
Article
Objective This study uses an intersectional approach to examine whether bonding and bridging family social capital change after adolescent delinquency and arrest. Background Family social capital (the resources and energy investments parents make in their children) has important implications for numerous youth outcomes. To date, little research has examined how stressful behaviors (like delinquency) and life events (such as arrest) strain or strengthen parent–child relationships, particularly across Black, White, and Hispanic families. Methods Drawing on data from the National Longitudinal Survey of Youth, 1997 cohort, the authors use fixed effects, dynamic panel, and correlated random effects models to analyze how delinquent behavior and arrest impact bonding and bridging forms of family social capital in adolescence. Stratified models by race/ethnicity and gender test whether the effects vary across groups. Results Results show that delinquency is negatively associated with bonding and bridging family social capital. Black girls experienced the sharpest reduction in family social capital resulting from delinquent behavior. Arrest was significantly associated with decreased bridging capital for Hispanic boys and increased bridging capital for Black girls. Conclusion Delinquency creates stress for parents and reduces investments in children, especially for Black girls. The effects of arrest vary by race and gender. Implications This study demonstrates the dynamism of family social capital and the impact of adolescent delinquency and arrest on parent–child ties, providing insights into the racialized and gendered development of family social capital amid heightened concern about youth deviance and incarceration.
Chapter
Past evidence has revealed the negative effects of children’s witnessing of family violence. With our increasingly aging society comes a higher risk of elder abuse, which creates new challenges in the form of the indirect and direct victimization of children in the home. However, research on the relationship between children’s witnessing of elder abuse and their victimization experiences is limited. This study examines the physical and mental health outcomes of witnessing elder abuse, as well as the relationship between children’s witnessing of elder abuse at home and child abuse victimization. A large-scale representative sample of 18,504 students aged 14–18 from six major cities in China was analyzed to investigate the associations. The results showed that children who had witnessed more than one type of elder abuse were more likely to have depression and poor health than those who had witnessed only one type. Children’s witnessing of elder abuse was also significantly associated with child abuse and bullying victimization. These findings offer implications for policymaking and service delivery in family-based child protection and interventions: Future interventions aimed at addressing indirect child victimization should be integrated with those focused on direct forms of victimization in order to effectively identify at-risk families as a whole.
Article
Although prior research suggests the importance of nonparental adults to adolescents, the ecological context of those relationships has received little attention. This study examined ties to influential adults among 122 adolescents who varied by race, family structure, and gender. The strongest effects were for race. Blacks reported stronger ties than Whites to the maternal grandmother as well as more supportive interactions with adult males. While race differences in grandparental ties were robust across social class (SES), ties to an influential adult male became nonsignificant upon controlling for SES. African American girls from divorced families consistently reported the strongest ties. Discussion considers the role of culture versus SES in explaining race differences. Implications for mentoring interventions are proposed, with special attention to the role of actualizing latent ties to already existing network members.
Article
Two competing hypotheses based on the relationship between living arrangements, social class, and cultural factors are developed. These hypotheses are examined for a sample of unmarried elderly Hispanic females, employing logistic regression techniques with data from the 1980 census. Support is found for the hypothesis that as economic status increases, the impact of cultural factors on living arrangement status diminishes. This result holds for Mexican American and Puerto Rican women but not for Cuban women.
Article
Using the Los Angeles Family and Neighborhood Survey (L.A. FANS), this paper examines the association between the presence of co-resident extended kin and children's internalizing and externalizing behaviors. The paper demonstrates the differential role of extended kin by family structure, as well as across parental immigrant status – specifically, nativity and documentation status. Children in the sample were found to be disadvantaged in extended family households, especially with regard to internalizing behaviors. This disadvantageous association was found mostly among married-parent extended family households, whereas there was no association between the presence of extended kin and behavior problems in children from single-parent families. This pattern emerged more clearly among children of documented immigrants, compared to those with native-born parents and those whose parents were unauthorized immigrants. These findings suggest a need to modify previous theories on extended family living arrangements; they also provide policy implications for immigrant families.
Article
Using data from the National Educational Longitudinal Study (NELS), we found that teenagers who live in nonmarried families are less likely to graduate from high school or to attend college, more likely to smoke or drink, and more likely to initiate sexual activity. Not all nonmarried families are alike, however. In particular, teenagers living with their single mothers and with at least one grandparent in multigenerational households have developmental outcomes that are at least as good and often better than the outcomes of teenagers in married families. These findings obtain when a wide array of economic resources, parenting behavior, and home and school characteristics are controlled for.
Article
Since Stack's (1974) landmark ethnography of kin support in a close-knit group of poor black mothers in the Midwest, there has been ample research on social support among low-income black families. While this body of work has largely painted a picture of the cohesive and supportive nature of families in black communities, recent research has highlighted the limited nature of kin support, especially the support available to low-income black mothers. Much of this work, however, has focused primarily on urban black mothers and paid less attention to the conditions that poor rural black mothers face when seeking and giving family support. Using longitudinal ethnographic data from a sample of 16 low-income black mothers in the rural South, we draw on social exchange, negotiated-order, and social capital perspectives to scrutinize the nature and costs of kin support in family networks marked by limited resources, instability, and chronic need. Our findings reveal the centrality of problematic resources and unpredictable family networks as conditions that diminish mothers' autonomy and compromise important "side bets" as mothers seek out, manage, and repay support. Implications of this study for theories of social support and social capital are also discussed.
Article
Nuclear Family Values, Extended Family Lives shows how the current emphasis on the nuclear family - with its exclusion of the extended family - is narrow, even deleterious, and misses much of family life. This omission is tied to gender, race, and class.
Article
This article compares time invested in children across family structures as a means to understand differences in children's development. Using data from the 1997 Panel Study of Income Dynamics' Child Development Supplement, we measure time investments from multiple caregivers and distinguish time children spend with a caregiver alone versus shared with another caregiver. We examine six family structures-married biological parents, cohabiting biological parents, mother and stepfather, mother and cohabiting boyfriend, single mother only, and multigenerational households. The total care-giving time that children receive in married biological parent families and cohabiting biological parent families is comparable to that for children living in stepfather families and multigenerational families. This is because children in stepfather families and multigenerational households receive substantial time investments from nonresident biological fathers and grandparents, respectively. In contrast, children receive little time investment from resident nonbiological father figures; and children in single mother, cohabiting boyfriend, and multigenerational households receive little time investment from their nonresident biological fathers. Finally, children who live with married biological parents receive the greatest share of caregiving time in the form of shared caregiving compared with children in all other family structures. Our findings suggest that having two resident biological caregivers predicts greater time investments in children and that shared parenting may be an important dimension of family structure. © 2014 by The American Academy of Political and Social Science.