How to do this Annotated Bibliography on Multi-generational Families

Sociological Perspectives, Volume 45, Number 1, pages 1–20. Copyright © 2002 by Paci�c Sociological Association. All rights reserved. Send requests for permission to reprint to: Rights and Permissions, University of California Press, Journals Division, 2000 Center St., Ste. 303, Berkeley, CA 94704-1223. ISSN: 0731-1214; online ISSN: 1533-8673

IN WHOSE HOME? MULTIGENERATIONAL FAMILIES IN THE UNITED STATES, 1998–2000

PHILIP N. COHEN* University of California, Irvine

LYNNE M. CASPER National Institute of Child Health and Human Development

ABSTRACT: This article examines multigenerational living arrange- ments of white, black, and Latino individuals using data from the Current Population Surveys. We describe people in multigenerational households as “hosts” or “guests.” In terms of resources, guests have no home of their own, whereas hosts maintain an important source of independence. By age, the proportion of adults living as guests peaks in the late twenties, then declines until the late seventies. In contrast, hosting rates peak in the �fties. Men have higher guest rates and women have higher host rates at almost all ages. While blacks and Latinos are more likely than whites to live in multigenerational households, those with higher incomes are less likely to live in multigenerational households and if they are living in multigen- erational households are less likely to be guests, regardless of race-ethnicity. We interpret this as consistent with the assumption that residential inde- pendence is generally preferred.

Complex household structures, their determinants and consequences, are impor- tant for understanding a wide variety of family-related research questions, includ- ing inequality and well-being within and across families, caregiving arrangements, intergenerational transfers of wealth, and the effects of family-related policy. This article looks at multigenerational living arrangements across the life course for white, black, and Latino individuals at the turn of this century. Its contribution is primarily conceptual and descriptive. Descriptive work in this area is important, as Burr and Mutchler (1993:S55) explain: “Understanding the household status of any population is critical because households serve as a platform from which other elements related to individual well-being and the maintenance of life chances are channeled.” Conceptually, standard practices for identifying multigenera- tional living arrangements and their implications remain elusive. In this article, we develop a method for identifying one type of multigenerational household—

* Direct all correspondence to: Philip N. Cohen, Department of Sociology, University of California, Irvine, Irvine, CA 92697-5100; e-mail: [email protected].

2 SO CIOLO GICAL PERSPECTIVES Volume 45, Number 1, 2002

parents and adult children living together—and examine how multigenerational living changes over the life course and across racial-ethnic groups. The method we use here is applicable not only to the widely available Current Population Survey but also to other U.S. government data sets and data sets with sim ilar �le structures.

Family structure is related to several aspects of inequality, and causation runs in both directions. Family structure is a purposeful response to hardship (Billingsley 1992), as has been shown in studies of extended households, especially for black and Latino families (Angel and Tienda 1982; Baca Zinn 1982–83; Blank 1998; Hogan, Hao, and Parish 1990). Multigenerational households can pool money, labor, and other resources and extend personal networks and support systems (Raley 1995; Tienda and Angel 1982). Marriage and divorce are also in�uenced by economic conditions and the �nancial situation of each member of the couple (Albrecht et al. 1997; Brines and Joyner 1999). Inequalities in the job market, incar- ceration rates, health status, and residential segregation differentially affect mar- riage rates and household structures by limiting options for some groups of women (Geronimus, Bound, and Waidmann 1999; Lichter et al. 1992; Wilson 1987). On the other hand, family structure can also be a cause of income inequal- ity across families because it connotes the number of potential earners and depen- dents in the family, as well as their gender and age, and these characteristics, in turn, affect economic outcomes (Bryson and Casper 1999).

Family structure and inequality issues intersect in the arena of welfare policy. The role of the extended family has received attention in the media and policy arenas as welfare reform takes hold (DeParle 1999; Harris 1999). For black and Latina mothers, relying on extended support networks to raise their children is “a traditional cultural remedy for a very modern structural situation” (Roschelle 1999:325). And, in an era of reduced welfare support, “we can assume that the kin and nonkin support network will become more crucial than ever to the survival of single-parent families” (p. 333). However, the bene�ts of household extension will be conditioned by the economic situation of members of the extended household (Hofferth 1984), as poor families �nd themselves drawing on the resources of poor network members (Roschelle 1999; Trent and Harlan 1994).

Earlier research on multigenerational households focused on the needs of the aged and their ability to coreside with adult children in times of need. Lower fer- tility and mortality rates in the twentieth century meant a smaller number of adult children on which the aged could rely (Treas 1977). However, by 1990, 90 percent of men and 84 percent of women age sixty-�ve or older were living in their own households, re�ecting a steep increase from the middle of the century and greatly improved conditions for the elderly (Treas and Torrecilha 1995:69). In the 1980s the great majority of coresidence between parents and adult children took place in the households of the parents (Aquilino 1990). The practice of host- ing multigenerational living arrangements has led to dramatic increases in the rates at which grandparents bring grandchildren into their homes, even without the presence of the middle generation of parents (Bryson and Casper 1999; Casper and Bryson 1998).

Thus, rather than a shortage of adult children with whom to live, it appears that

Multigenerational Families in the United States 3

many in today’s older population may face the opposite problem: too many younger relatives living in their homes. Today’s older Americans are the parents of baby boomers, so their chances of having a living adult child are relatively high (Crimmins and Ingegneri 1990). As the baby boomers reach retirement age they will have fewer adult children—either to lean on or to support—than previous generations did.

However, these trends are not independent of race-ethnicity (Mutchler 1992). The trend toward living in the homes of older parents was driven by whites and blacks, as younger Asians and Latinos were much more likely than comparable whites to bring older parents into their homes (Kamo 2000). Treas and Torrecilha (1995:70) report that whites alone account for the increase in independent living in the 1980s. Similarly, Casper and Bryson (1998:Table 2) report that only 19 per- cent of children living with their grandmothers and without parents are non- Hispanic whites.

CONCEPTUAL ISSUES

Researchers have suggested that the preference for independent living has increased over the past century (e.g., Ruggles 1996). Household independence is generally preferred to extended household structures. A widespread preference for privacy and independence has been linked to minimalist living arrangements (Wister and Burch 1987). Most researchers assume people will use their resources to obtain such independence if they can (e.g., Burr and Mutchler 1993). In fact, studies have consistently found that income is one of the most important deter- mining factors of independent living; older Americans with more money are more likely to live independently (Crimmins and Ingegneri 1990; Mutchler 1992). This is also consistent with research showing higher rates of complex or multigen- erational households among economically disadvantaged groups, such as blacks, Latinos, and Asians (Angel and Tienda 1982; Mutchler 1992; Kamo 2000; Speare and Avery 1993), although cultural in�uences contribute as well (Kamo 2000), and attitudes toward multigenerational arrangements have become more accepting in recent decades (Alwin 1996).

An important milestone event in the transition to adulthood is the ability to establish and maintain an independent residence, either alone or with a spouse or roommates. Young people start their careers with lower wages and may not be able to afford to live on their own right away, but as they age, they usually acquire more education, skills, experience, and pay and are more able to establish inde- pendent households. Working adults in midlife typically increase their incomes until the time they retire. On retirement, incomes drop and people eventually begin to develop chronic age-related health problems that may make it impossible for them to maintain independent households. Thus the ability to sustain an inde- pendent residence as an adult is greatest in midlife and least in the early and late years of adulthood. Indeed, Callis (1997) reports that home ownership rates are highest among those ages �fty-�ve to sixty-four, which are the prime years for hosting adult children (as we show below).

By the latter part of the twentieth century, adult children were much more

4 SO CIOLO GICAL PERSPECTIVES Volume 45, Number 1, 2002

likely to live with their parents and older adults were much more likely to live alone than they had been at midcentury (Bianchi and Casper 2000; Casper and Bianchi 2001; Fields and Casper 2001). Children were slower to move out of their childhood households and more likely to move back in as adults, in large part because of factors that reduced their relative independence: lower marriage rates, increased housing costs, reduced �nancial support for college attendance, lower real wages for some groups, and even the repeal of the military draft (Gold- scheider and Goldscheider 1994). At the same time, government policies have helped to improve the �nancial security of older Americans (Treas and Torrecilha 1995) and hence their ability to maintain households. The GI Bill sent many of today’s older Americans to college. The federal government helped many—espe- cially whites (Oliver and Shapiro 1995)—to buy homes and allowed them to take a deduction on their interest payments; it also protected their private pension plans and—in addition to providing Social Security—used tax laws to protect their incomes when they reached age sixty-�ve. Patterns of intergenerational sup- port, apparently responding to this generational balance of economic forces, shifted so that adult children came to rely more on their older parents.

Nevertheless, Aquilino (1990) points out that researchers still commonly assume that the coresidence of parents and adult children results from parents’ need for assistance. When Burr and Mutchler (1993) look only at unmarried older women it is in part because these women are more likely to need assistance of some kind. Data on the actual transfer of resources within households, needed to resolve these questions, frequently is unavailable. Given the preference for privacy and living in one’s own household, householder status suggests the direction of �ow for one important resource, the home itself. The multigenerational “guest” generally has no home of his or her own, whereas the “host” maintains at least one important source of independence, which she or he shares with extended family members. Thus, even though both hosts and guests give up privacy in the arrangement—and even though guests may provide essential assistance in the form of child care, rent, or other contributions—hosts may be expected to have greater resources, as evi- denced by their ability to independently maintain a home.1

Most studies have not posed the question in terms of whose home is hosting multigenerational arrangements. For example, Crimmins and Ingegneri (1990) look at whether older Americans live with an adult child but not at whose house they live in. Speare and Avery (1993:S72) conclude that “when children live with parents under age 75 they are likely to be the primary bene�ciaries of the relation- ship” but do not report on whose home is involved in these coresidences. The atti- tude trend data analyzed by Alwin (1996) show that adults living with their par- ents is a living arrangement that is looked on more favorably by younger people than by older people. However, the survey question did not ask in whose home the hypothetical coresidence takes place.

Aquilino (1990) does examine the question of householder status. He �nds that in the late 1980s more than 90 percent of parents living with their adult children lived in the parents’ household. In Kamo’s (2000) study of census data, however, more than half of Asian families with adult children were in the home of the chil- dren, as were about one-third of such Latino families. The arrangement of adult

Multigenerational Families in the United States 5

children living in their parents’ home does not preclude the possibility that the children are assisting the parents, of course, but the parents are contributing at least in the provision of the household. And beyond the question of contributing resources, those who host extended families maintain a crucial aspect of their independence by keeping their own homes.

The observed relationship between resources and multigenerational living is complicated, however. For example, although extended households may repre- sent ef�cient income pooling strategies, Kamo (2000) �nds that per capita income is lower in extended households across all racial-ethnic groups. Although better- off people may be more able to provide support for other family members, those same better-off hosts are presumably less likely to have relatives who need assis- tance. Thus Aquilino (1990) �nds that parents’ education reduces the likelihood that their adult children come to live with them, but this may be because parents with more education have children with more education, who are more �nan- cially independent. So hosting may be an indicator of having high levels of resources relative to others in a particular family network but also an indicator of a resource-poor network overall. And within a multigenerational network guests are expected to be those in the least advantageous position.

This article focuses on a distinction between “hosts” and “guests,” which identi�es multigenerational household members by their householder status. In the absence of such a distinction, it makes more sense for studies to focus on either the old (Burr and Mutchler 1992, 1993) or the young (Goldscheider 1997), because old people tend to “host” and young people tend to “guest.” But distin- guishing between host and guest roles enables us to describe more speci�cally the patterns of multigenerational living arrangements across the life course. Concep- tually, this approach is similar to that offered by Kamo (2000), who identi�es extended households as “upward,” “downward,” or “horizontally” extended, depending on the relationship of non-nuclear members to the household head. At the individual level, Kamo’s measure of “dependent members” parallels our “guest” identi�er. The host/guest model we use here sacri�ces some detail on the type of extension among hosts but gains simplicity as well as an individual-based measure for hosts as well as guests. This simplicity permits analysis of multigen- erational status for individuals across the life course, allowing us to examine the gender of hosts, for example. Other methodological issues are addressed below.

DATA AND MEASURES

Data for this article are from the 1998–2000 March Current Population Surveys (CPSs), which include about �fty thousand households per year. We pool three years of data to increase reliability for the relatively small subgroups under exam- ination here. However, because households are usually interviewed in two con- secutive March CPSs, we include only those in the outgoing rotation from 1998, the full sample in 1999, and in the incoming rotation in 2000. So about half of our sample is from 1999, with a quarter each coming from the 1998 and 2000 surveys.2

These data include a lot of information about household composition and per- sonal characteristics but do not include important contextual information such as

6 SO CIOLO GICAL PERSPECTIVES Volume 45, Number 1, 2002

whether individuals have any living children outside the home (Aquilino 1990), the presence of nearby relatives (Logan and Spitze 1994), children ever born to women, or economic transfers between household members. Thus, although the CPS provides the most recent nationally representative data and is very useful for demographic pro�les, it is not as useful as some richer data sets for developing causal models of living arrangements and relationships.

Multigenerational households—households with two or more generations of adults—may be maintained (hosted) by the parents of adult children or by the adult child of older parents. All adults in these households are identi�ed as “hosts” or “guests.” Speci�cally, we de�ne multigenerational households as those that include an adult child of the householder3 or spouse (including cohabiting part- ner) of the householder and/or a parent of the householder or spouse (including cohabiting partner) of the householder. Adult children are de�ned in either of two ways: (1) any child age eighteen or older who has ever been married or has a child of his or her own in the household or (2) any child age twenty-�ve or older. After multigenerational households are identi�ed, each person is categorized as either a “host” (the householder or spouse or cohabiting partner of the householder) or “guest” (everyone else in the household).4 This is the �rst analysis of which we are aware that treats cohabiting partners as spouses for purposes of examining familial household extension, an option available since 1995 with CPS data.

In the absence of data on assistance or exchange in households, measures of com- plex or extended households broadly de�ned (Angel and Tienda 1982; Tienda and Glass 1985) seem less desirable than a measure based on known generational rela- tions, which may more reasonably be assumed to include resource- and labor-sharing arrangements. Burr and Mutchler (1993) use a similar construct to identify complex households. They exclude women from their complex household measure if they live only with children under age 18 or with children 18 to 25 who are still in school. Their de�nition of “adulthood” for those age 18 to 25 is school attendance rather than subfamily construction. We do not treat 18- to 25-year-old children as adult children, unless they have children of their own or have been married, because most of them probably have never established an independent household. Our de�nition of adult children is similar to Kamo’s (2000), except that we use age 25 instead of age 30 as the cutoff point. Although any age cutoff is admittedly arbi- trary, age 25 is the usual labor force conception of independent adulthood.

We include only non-Latino white, non-Latino black, and Latino adults. Sample sizes in the CPS are too small to look at Asian Americans, especially given their cultural diversity and family diversity later in life (Burr and Mutchler 1993), and Native Americans. Although we include all Latinos in one group regardless of national origin, we would have preferred to examine the larger Latino subgroups separately, as Mexicans and Puerto Ricans, for example, have different patterns of family structure. Thus this analysis is only general, and �ndings for the Latino population should be regarded as preliminary. The descriptive statistics are all weighted with the March CPS person weight.

Two caveats are in order. First, the measure here does not include the approxi- mately one million households in which grandchildren and grandparents live with no parents present (Casper and Bryson 1998). On the other hand, households

Multigenerational Families in the United States 7

may be identi�ed as multigenerational even if there are no children present. Thus “multigenerational” as used here refers to more than one generation of adults. Second, the analysis is cross-sectional. Research that uses life histories will iden- tify much higher rates of multigenerational living. Goldscheider and Gold- scheider (1994), for example, report that about 40 percent of adults who left their parents’ homes in the 1970s and 1980s at some point returned to their parents’ home. The data employed here �nds a much lower proportion of young adults living with their parents.

RESULTS

Descriptive Analysis

To give a sense of what can and cannot be learned about households in the CPS, three four-generation households from the data are diagrammed in Figure 1, with �ctional names. The most simple (panel B), shows a four-generation black family. The householder (Sophie) is a 47-year-old widow with a high school diploma. She worked full time for part of last year as a nursing aide and reported earning $16,000. In her home live her mother—a 76-year-old widow who did not go to high school and is not employed—and her daughter, a 24-year-old single mother with two young children of her own. The daughter also �nished high school only and also worked full time for part of last year as a nursing aide, earning $15,000. In a household such as Sophie’s it appears plausible that both the older and younger generations are bene�ting from access to Sophie’s home, whatever other exchanges are taking place.

A more complex, white household is shown in panel A. This home is main- tained by a married couple (Bob and Mary). Both have elementary education and worked in blue-collar jobs for $20,000 in the previous year. They are joined in their home by Mary’s widowed mother, Sylvia, and their daughter Gretchen’s family. Gretchen and her husband, Rich, both have some college education and earnings of $23,000; he is a sewing machine operator and she is an of�ce supervisor. They have a two-year-old daughter. In this household, again, it may be that the older and younger generations are bene�ting from access to the home of Mary and Bob. Maybe Sylvia helps to care for the young daughter. Gretchen and Rich may be saving up to move out on their own.

The most complicated household (panel C) is a Latino household maintained by a 43-year-old divorced woman (Angela), who has three children, ages 10, 13, and 19. She is a naturalized citizen with an elementary education who worked as a butcher for $19,000 last year. She is joined in her home by her mother, a 79-year- old separated woman who has not become a citizen and is not employed. The old- est child is Mike (born in the United States, he is identi�ed as white non- Hispanic), a prison guard who did not �nish high school and earned $9,000 in 1998. He and his 17-year-old immigrant wife (still in school and a waitress) live in Angela’s home with their one-year-old son. Finally, the family is joined by a 21- year-old, foreign-born relative, who may work with Angela in the meat industry, earning $21,000. Although one may imagine support running in many directions in such a household, it is unlikely that Angela needs this many other relatives to

8 SO CIOLO GICAL PERSPECTIVES Volume 45, Number 1, 2002

F ig

u re

1

E xt

en d

ed H

o u

se h

o ld

R el

at io

n sh

ip s

in t

h e

C u

rr en

t P

o p

u la

ti o

n S

u rv

ey

Multigenerational Families in the United States 9

help her care for her two young children. More likely, at least some of these rela- tives are unable or not ready to establish a household of their own, for economic or other reasons.

These are not typical multigenerational households; instead we have chosen unusually complex households to show the reach of the data. In the terms speci�ed above, a person who is both an adult child and a parent of adult children hosts each of these households, since the householder (or couple) has adult chil- dren as well as an elderly parent at home. The dark boxes indicate the hosts; everyone else is a guest. As these partial portraits show, the CPS offers a probing snapshot of households with a great deal of detail but little in the way of explana- tion in terms of their history, motivations, networks of exchange and support, atti- tudes, or beliefs. These portraits help us to see an important feature of the guest and host roles across the life course. In each household, there are young adult guests, middle-aged hosts, and older guests.

For a broader picture, we offer Figure 2, which shows the percent of all adults who live in multigenerational households as hosts and guests, by race-ethnicity and age. The �gure con�rms that blacks and Latinos are more likely than whites to live in multigenerational homes at all ages, with the highest guest rates appar- ent for young blacks and old Latinos. The proportion of adults living as guests peaks for each group in the late twenties.5 It then declines until the seventies,

Figure 2

Multigenerational Host and Guest Rates by Age and Race-Ethnicity, 1998–2000

10 SO CIOLO GICAL PERSPECTIVES Volume 45, Number 1, 2002

except for Latinos, among whom guest rates rise starting in the �fties. Hosting rates show the opposite pattern, with the highest rates reached in the �fties and sixties. In strictly demographic terms, middle-aged people are at greatest risk of living in multigenerational homes, because they are most likely to have living adult children, older parents, or both. In fact, however, the great majority of middle- aged multigenerational residents are hosts, providing homes for their adult chil- dren rather than their older parents.6

The host/guest distribution differs by gender. Women are much more likely than men to live with and care for young children (England 2000), and this discrep- ancy continues for care of adults as well. Figure 3 shows the percentage of men and women who live in multigenerational households as hosts and guests by age. For both groups, the life course pattern is similar, but men have higher guest rates and women have higher host rates at almost all ages. The exceptions are at the young- est ages (under 25), when women with small children may be living with their par- ents, and at the oldest ages, when women have higher rates of widowhood.

Although marriage is no longer the dominant reason for young adults to leave

Figure 3

Multigenerational Host and Guest Rates by Age and Gender, 1998–2000

Multigenerational Families in the United States 11

home (Goldscheider and Goldscheider 1994), men’s older age of �rst marriage— estimated to be 26.7 in 1998, compared with 25.0 for women (Lugaila 1998)—may account for some of the young men’s higher guest rates (especially given our age- 25 cutoff). For each racial-ethnic group, the guest rates are highest for never-married men, reaching 27 percent for never-married black men. Welfare programs could also play a role in women’s lower guesting rates, if the presence of young children helps women qualify for public assistance, including housing assistance.

Multivariate Analysis

We argue that the process of landing in any multigenerational household dif- fers from the subsequent process that differentiates hosts and guests within multi- generational arrangements. This can be considered several ways. On the one hand, if multigenerational arrangements are made out of necessity, the risk of multigenerational living is increased by weakness or hardship among individuals or their kinship networks. Among those at risk, then, members will live with those best situated to host an extended family. On the other hand, to the extent that multigenerational arrangements follow from cultural preferences, age structures, or the structural conditions faced by racial-ethnic groups (such as local housing costs [Kamo 2000]), a multigenerational population emerges, and network mem- bers live with those best suited for hosting.7 With these data we are able to look at the likelihood of multigenerational living—and the processes of hosting and guesting separately—but we cannot differentiate between these two explanations.

To model this as a two-stage process, we compare the determinants of the two statuses in nested logistic regression models. In the �rst model, we investigate the odds of living in a multigenerational household for the entire adult sample. In the second, we use the same variables to predict guest versus host role status among those in multigenerational households. This differs from the approach used by Kamo (2000), which models the odds of being a (similarly de�ned) guest versus the combined categories of hosting an extended family, living alone, or living in a nuclear family. It also differs from a multinomial approach, which would simulta- neously model the odds of being hosts, guests, or not living in multigenerational households.8 In our analysis, because we consider these as conceptually sequen- tial processes, people are only differentiated as hosts or guests once they are in the multigenerational population.

Variables in the model include six age categories (to capture the nonlinearity apparent in Figures 1 and 2); a set of dummy variables indicating combined mari- tal status/gender/own-child-under-6 status;9 the log of personal income;10 an indicator for whether each person received any public assistance or welfare cash payments in the previous year; an indicator for foreign-born status; an indicator for full-time full-year employment; years of education completed; and an indica- tor for work-limiting disabilities (the only measure of disability available in the data). Based on previous research (e.g., Kamo 2000) and a preliminary check of racial-ethnic interactions, we model the outcomes separately for white, black, and Latino adults, indicating signi�cant differences where they occur.

Table 1 presents the percentage of white, black, and Latino adults living in mul-

12 SO CIOLO GICAL PERSPECTIVES Volume 45, Number 1, 2002

TABLE 1

Multigenerational Household Status, by Race-Ethnicity and Personal Characteristics, 1998–2000 (Percent)

White Multi-

generational

Black Multi-

generational

Latino Multi-

generational

Characteristics Guests Guests Guests

Total 11.4 52.2 22.5 63.1 21.4 60.3

Age 15–24 9.1 87.4 19.2 89.0 18.4 86.6 25–29 16.3 91.8 29.3 91.9 21.5 82.9 30–39 8.3 81.3 18.1 86.2 13.8 66.9 40–49 9.4 52.4 20.8 54.9 18.9 44.4 50–59 14.6 20.3 26.7 28.5 34.3 34.8 60–69 14.6 17.6 28.7 22.3 36.5 32.6 701 13.3 34.0 27.9 34.1 36.7 56.4

Family status Married man, child ,6 2.8 48.5 4.6 63.5 7.9 40.8 Married man, no child ,6 8.6 11.6 16.0 10.2 19.9 22.2 Married woman, child ,6 2.8 47.6 4.9 66.5 8.1 39.8 Married woman, no child ,6 8.6 11.9 16.1 10.3 19.3 21.0 Formerly married man 16.6 72.2 26.5 85.1 30.3 79.8 Formerly married woman, child ,6 23.4 87.9 27.4 86.9 33.0 87.5 Formerly married woman, no child ,6 18.0 45.2 29.8 34.8 33.7 57.9 Never-married man 17.5 91.0 28.7 93.3 23.9 88.9 Never-married woman, child ,6 21.9 92.0 24.4 89.6 27.0 86.6 Never-married woman, no child ,6 12 13.1 90.0 22.6 79.9 23.3 81.5

Income None 17.3 84.6 29.4 88.1 26.3 87.3 $1–10,000 15.3 66.4 25.6 66.7 22.8 63.8 $10,001–20,000 13.3 52.0 23.0 56.0 21.0 54.7 $20,001–40,000 10.3 44.0 18.7 55.2 18.9 47.9 More than $40,000 7.1 30.3 14.9 40.6 15.8 39.1

Welfare income previous year 11.3 63.8 14.9 67.3 15.2 56.8 Full-time, full-year employed 10.3 52.3 20.2 59.2 20.0 55.9

Education Less than high school 13.6 55.7 23.7 60.9 21.8 58.0 High school only complete 13.9 51.3 25.0 64.9 23.3 62.9 Some college 10.6 53.0 20.7 65.9 19.4 64.7 College degree only complete 8.0 54.1 17.6 63.7 17.0 54.5 Higher degree 6.4 36.1 13.9 28.5 16.3 49.2

Work-limiting disability 15.2 51.3 27.4 59.1 27.2 53.3 Foreign born 15.9 46.5 20.8 58.7 22.2 56.9 N 144,388 18,338 26,830

Note: Percent living in multigenerational households and percent of those who are guests. White and black are non- Latino. Income is own income or one-half of married-couple income. See text for de�nitions.

Multigenerational Families in the United States 13

tigenerational households and of those, the percentage who are guests, in total and in categories for each of the independent variables.11 Overall, 11 percent of whites are in multigenerational households, compared with 23 percent of blacks and 21 percent of Latinos. Within multigenerational households, guests are more common among blacks (63%) and Latinos (60%) than whites (52%), re�ecting the larger household size among black and Latino multigenerational households.

Table 2 shows the logistic regression coef�cients for the odds of living in a mul- tigenerational household among the total adult sample. The bivariate age pattern for multigenerational living has two peaks—early adulthood and late middle age. However, in the multivariate model, which controls some predictors of need, the

TABLE 2

Logistic Regression Coef�cients for Living in a Multigenerational Household, 1998–2000

White Black Latino

Intercept 2.766*** 2.770*** 21.397***a

Age 15–24 21.336*** 2.823***a 2.562***a

25–29 .330*** .347*** .103a

30–39 2.151*** 2.171** 2.341***a

40–49 — — — 50–59 .521*** .330***a .762***a

60–69 .400*** .350*** .825***a

701 .011 .179*a .755***a

Family status Married man, no child ,6 — — — Married man, child ,6 21.042*** 21.455***a 2.838*** Married woman, child ,6 21.106*** 21.507***a 2.750***a

Married woman, no child ,6 2.218*** 2.253*** 2.197*** Formerly married man .657*** .454***a .508*** Formerly married woman, child ,6 1.399*** .747***a .927***a

Formerly married woman, no child ,6 .691*** .599*** .548*** Never-married man 1.267*** .878***a .699***a

Never-married woman, child ,6 1.797*** .971***a 1.034***a

Never-married woman, no child ,6 .997*** .594***a .637***a

Income (ln) 2.061*** 2.054*** 2.026***a

Welfare income previous year 2.547*** 2.763*** 2.564*** Full-time, full-year employed 2.080*** 2.035 .018 Education (years) 2.078*** 2.025**a 2.001a

Work-limiting disability 2.098*** 2.031 2.126* Foreign born .364*** 2.043a .159***a

Likelihood ratio Chi-square 7,429*** 1,114*** 1,903*** N 144,388 18,338 26,830

1 p , .10; * p , .05; ** p , .01; *** p , .001 (two-tailed tests). a different from white coef�cient at p , .05 (two-tailed test). Note: Adults 151. White and black are non-Latino. Income is own income or one-half of married-couple income.

14 SO CIOLO GICAL PERSPECTIVES Volume 45, Number 1, 2002

odds across the life course re�ect instead something closer to the crude demo- graphic risk, with those in their �fties and sixties having the highest odds of liv- ing with multiple generations. Although the tests for black-white and Latino- white differences are signi�cant in a number of age categories, the differences are in degree, not direction of the effects. The exception is the oldest age group, in which Latinos and to a lesser extent blacks have higher odds of multigenerational living than do whites.

The effects of gender, marital status, and small children are also quite similar across racial-ethnic groups despite differences in the strength of the effects. For all three groups, married men and women have the lowest rates of multigenerational living, and never-married women with young children have the highest rates. For each group except never-married and formerly married women, having young children reduces the odds of multigenerational living. In almost every category, the signi�cant racial-ethnic interactions represent smaller effects for black and Latino adults, except in the case of married blacks.

More income, more education, welfare receipt, and full-time full-year employ- ment are associated with lower rates of multigenerational living (or have non- signi�cant effects). Among whites and Latinos, the foreign born are signi�cantly more likely to live in multigenerational households. Surprisingly, whites and Latinos with work-limiting disabilities are less likely to live in multigenerational households. Note that the variable identi�es disabilities only for people whose employment is affected by their disabilities, and these people are of the ages at which people are most likely to host.

Turning to the guest versus host distinction—the model shown in Table 3— several different dynamics emerge. Age has a very different effect on the guest versus host distinction. For each group, the odds of being a guest decline from the late twenties to age seventy. This is consistent with the argument that the resource balance favors those who are older, with the exception of those over seventy, who through health problems and widowhood start to lose relative independence. Because these effects persist net of some controls for economic condition, it appears additional resources—such as home ownership and other assets—play an important role in this balance. Despite signi�cant differences in degree, this pattern is consistent across racial-ethnic groups.

Married people are most likely to live independent of any multigenerational arrangement. Among those in multigenerational households, however, gender and marital status play a further role in determining host versus guest roles. In each group married adults are the least likely to be guests. Thus marriage is con- sistently associated with greater odds of independent living. By contrast, the effect of small children is not consistent across the two processes, as those with small children are more likely to be guests than those without. Given the relative economic well-being of divorced men compared to women (Bianchi, Subaiya, and Kahn 1999), it is perhaps surprising that formerly married men are not hosting more multigenerational families compared to formerly married women (except those with small children, who may be seeking out extended family guests to help care for their children).

The effects of income, welfare receipt, full employment, and education are gen-

Multigenerational Families in the United States 15

erally consistent with the assumed preference for independence. Although the welfare effects may be counterintuitive, remember that the models control for income. So, at a given level of income, those who receive welfare are more likely to live independently or in multigenerational households of their own. The income, employment, and education effects show that among those in multigen- erational households, those with lower income, employment, and education lev- els live in the homes of those with more of these assets. This may be interpreted as those with greater resources exercising their choice to maintain their own house-

TABLE 3

Logistic Regression Coef�cients for Guest Status in Multigenerational Households, 1998–2000

White Black Latino

Intercept .367* 2.448a .015

Age 15–24 .583*** .729*** 1.384***a

25–29 1.638*** 1.569*** 1.493*** 30–39 .950*** 1.252*** .646*** 40–49 — — — 50–59 21.163*** 2.696***a 2.444***a

60–69 21.452*** 21.311*** 2.720***a

701 21.046*** 2.984*** .147a

Family status Married man, no child ,6 — — — Married man, child ,6 .3041 1.230**a 2.142 Married woman, child ,6 .065 1.264**a 2.391* Married woman, no child ,6 2.317*** 2.316 2.330* Formerly married man 2.556*** 3.579***a 2.380*** Formerly married woman, child ,6 3.309*** 3.203*** 2.642*** Formerly married woman, no child ,6 1.702*** 1.523*** 1.341***a

Never married man 2.886*** 3.433***a 2.159***a

Never married woman, child ,6 2.935*** 3.084*** 1.674***a

Never married woman, no child ,6 2.841*** 2.477*** 1.564***a

Income (ln) 2.094*** 2.065*** 2.108*** Welfare income previous year 2.980*** 2.933*** 2.770** Full-time, full-year employed 2.194** 2.2501 2.109 Education (years) 2.049*** 2.046* 2.012a

Work-limiting disability .008 .094 .027 Foreign born .286** .194 .088

Likelihood ratio Chi-square 9,475*** 2,217*** 2,273*** N 15,750 3,944 5,414

1 p , .10; * p , .05; ** p , .01; *** p , .001 (two-tailed tests). a different from white coef�cient at p , .05 (two-tailed test). Note: Adults 151 living in multigenerational households. White and black are non-Latino. Income is own income or one-half of married-couple income.

16 SO CIOLO GICAL PERSPECTIVES Volume 45, Number 1, 2002

holds in the face of family networks in need of help or as those with greater resources being called on to help those with fewer resources—or both.12

In both models, black-white and Latino-white differences for employment, edu- cation, and income, where they are signi�cant, show greater effects for whites. It is possible that for whites these variables are serving as proxies for greater resources including other assets. In the �rst model, the stronger effects of education and income for whites may imply that those whites with higher education and income are less likely to have family network members in need of multigenerational liv- ing arrangements.

Consistent with most previous research, blacks and Latinos are more likely to live in multigenerational households, even when personal characteristics are con- trolled.13 These results are consistent with culture differences described by others (Kamo 2000), but there is insuf�cient evidence here to conclude that cultural fac- tors are decisive, because the condition of members of their family networks is not controlled. At equal levels of income, for example, black mothers might be more likely than white mothers to have poor older parents. Even controlling for addi- tional factors, such as community context variables (Kamo 2000), without measur- ing the well-being of family network members, interpretation of racial-ethnic dif- ferences should be made cautiously.

CONCLUSION

We �nd that the proportion of adults living as guests peaks in the late twenties and then declines until the late seventies, whereas hosting rates peak in the �fties. Women are less likely to be guests and more likely to be hosts than men are. Con- sistent with previous research, blacks and Latinos are more likely than whites to live in multigenerational households, even when other factors are controlled. Those with higher incomes and other resources are less likely to live in multigen- erational households altogether and less likely to be guests if they are living in multigenerational households.

The structure of multigenerational households informs us about intergenera- tional relations, issues of privacy and independence over the life course, and strat- egies for coping with poverty and hardship. These results present a picture of younger adults leaning on the resources of their older relatives, who in turn sacri�ce some of their privacy—but maintain their independence—when they open their homes. The guest/host distinction we offer contributes to previous research to help clarify the pattern of these relationships. The results here suggest several areas that might bene�t from more sustained attention.

Newman (1999:194) argues in her study of the working poor that “[a]f�uence loosens the ties that remain tight, even oppressive at times, in poor communities.” Facing a dearth of economic capital, she argues that the working poor “preserv[e] a form of social capital that has all but disappeared in many an American sub- urb.” Clearly in the area of poverty and welfare, and racial-ethnic inequality research, household structure has �gured prominently. Some who argue for the return of the nuclear family hold patriarchal assumptions about gender relations (Coontz 2000), but expecting extended household structure to resolve postwelfare

Multigenerational Families in the United States 17

hardships runs the risk of idealizing what may actually be the desperate measures of the poor (Roschelle 1999). Nevertheless, the dynamics of household extension help to explain how people respond to and compensate for the hardships and inequities they face (Jarrett 1994; Trent and Harlan 1994).

These results offer an interesting set of �ndings regarding gender and its inter- action with race-ethnicity. First, we have shown that women across most of the life course have higher rates of hosting and lower rates of guesting than do men. Perhaps it is not surprising that women are more likely to be on what we have identi�ed as the giving end of intergenerational family support systems. But the extent of this fact is especially striking given men’s greater economic resources. Single men very rarely host multigenerational family members, despite their apparent advantages in terms of resources. To illustrate this, we present Table 4. This shows that half of white and Latino multigenerational households and less than one-third of black multigenerational households are hosted by married or cohabiting couples. Women alone host about one-third of white and Latino multi- generational households and more than half of black multigenerational house- holds. Thus, in only 11 percent (black) to 15 percent (Latino) of multigenerational households does the primary family not include a woman. The extent of gender inequality in caring for young children has received much more attention than the imbalance in the provision of households to adult children and older parents, which is clearly also dominated by women, especially black women.

Finally, this analysis has not included consideration of the characteristics of multigenerational household members in relation to each other. For example, black women hosting multigenerational families may get more or less �nancial or other help from their guests than do white or Latino women. Future research should consider these dynamics, which require attention to causal ordering and more complex relationship measures than are possible with these data.

NOTES

1. In some cases, the householder may only be able to maintain the household because of the contribution of guests. Nevertheless, we think it likely that the resource balance generally favors the host over the guests.

2. For details on the survey sample and rotation scheme, see U.S. Census Bureau (2000).

TABLE 4

Gender and Hosting Multigenerational Households, 1998–2000

White Black Latino

Percent of multigenerational households hosted by Married or cohabiting couples 50.9 30.0 51.4 Single women 35.6 59.3 33.4 Single men 13.5 10.7 15.3

Percent of single hosts who are women 72.5 84.7 68.6

Note: For de�nitions, see text.

18 SO CIOLO GICAL PERSPECTIVES Volume 45, Number 1, 2002

3. The householder is the person in whose name the house or apartment is owned or rented.

4. Note that hosts may be parents of adult children and adult children themselves, if there are three generations of adults present.

5. The speci�c location of the peak is partly an artifact of the coding scheme, which identi�es children as “adults” when they reach age 25, unless they have been married at an earlier age or have children.

6. The majority of white, black, and Latino householder-hosts are hosting adult children, not older parents (not shown). About three-fourths of hosts in their �fties—the point at which they are most likely to have living older parents and older children—are hosting adult children, not elderly parents. Thus, although guest rates do climb again at advanced ages, hosting is predominantly the behavior of parents rather than children.

7. Note that given the choice between analyzing individuals and households as the unit of analysis, we chose individuals. Clearly, decisions regarding multigenerational living arrangements are not made individually. So our models are not necessarily decision- making models but rather re�ect the likelihood of the two outcomes given individual characteristics.

8. Our second model, differentiating hosts from guests among those in multigenerational households, is equivalent to the second model obtained in the multinomial logistic. We ran multinomial logistic models but do not present those here because they do not model the odds of being in a multigenerational household—that is, the odds of host or guest versus neither. Results from these models are available from the authors on request.

9. Because of the small number of single men with small children, formerly married and never-married men are not coded separately by the presence of children.

10. The variable represents the annual individual income for the previous year for singles and one-half of the couple’s combined earnings for those who are married with spouse present.

11. For ease of interpretation, the descriptive table is set up to parallel the multivariate analysis, but it includes the information needed to derive the percentage of the total population living as either hosts or guests. For example, the table shows that the pro- portion of all white adults living in multigenerational households is .114, and of those .522 are guests. Thus .114 3 .522 5 .060 of whites are guests and .114 3 (1 2 .522) 5 .054 are hosts.

12. The lack of signi�cant effects for employment, education, and nativity for Latinos may re�ect the greater diversity within the Latino population and the need for �ner-grained measures, such as length of stay and English �uency (Kamo 2000).

13. This is based on the regression result for a multigenerational living model with black and Latino dummy variables and no interaction terms (not shown).

REFERENCES

Albrecht, Carol M., Mark A. Fossett, Cynthia M. Cready, and K. Jill Kiecolt. 1997. “Mate Availability, Women’s Marriage Prevalence, and Husbands’ Education.” Journal of Family Issues 18(4):429–52.

Alwin, Duane F. 1996. “Coresidence Beliefs in American Society—1973 to 1991.” Journal of Marriage and the Family 58(2):393–403.

Angel, Ronald and Marta Tienda. 1982. “Determinants of Extended Household Structure— Cultural Pattern or Economic Need.” American Journal of Sociology 87(6):1360–83.

Multigenerational Families in the United States 19

Aquilino, William S. 1990. “The Likelihood of Parent Adult Child Coresidence: Effects of Family-Structure and Parental Characteristics.” Journal of Marriage and the Family 52(2):405–19.

Baca Zinn, Maxine. 1982–83. “Familism among Chicanos: A Theoretical Review.” Humboldt Journal of Social Relations 10(1):224–38.

Bianchi, Suzanne M. and Lynne M. Casper. 2000. “American Families.” Population Bulletin 55(4):3–43.

Bianchi, Suzanne M., Lekha Subaiya, and Joan R. Kahn. 1999. “The Gender Gap in the Eco- nomic Well-Being of Nonresident Fathers and Custodial Mothers.” Demography 36(2):195–203.

Billingsley, Andrew. 1992. Climbing Jacob’s Ladder the Enduring Legacy of African-American Families. New York: Simon & Schuster.

Blank, Susan. 1998. “Hearth and Home: The Living Arrangements of Mexican Immigrants and U.S.-born Mexican Americans.” Sociological Forum 13(1):35–59.

Brines, Julie and Kara Joyner. 1999. “The Ties That Bind: Principles of Cohesion in Cohabi- tation and Marriage.” American Sociological Review 64(3):333–55.

Bryson, Kenneth R. and Lynne M. Casper. 1999. “Co-Resident Grandparents and Their Grandchildren.” U.S. Census Bureau Current Population Reports P23(198).

Burr, Jeffrey A. and Jan E. Mutchler. 1992. “The Living Arrangements of Unmarried Elderly Hispanic Females.” Demography 29(1):93–112.

———. 1993. “Nativity, Acculturation, and Economic Status: Explanations of Asian-American Living Arrangements in Later Life.” Journals of Gerontology 48(2):S55–S63.

Callis, Robert R. 1997. “Moving to America: Moving to Homeownership.” U.S. Census Bureau, Current Housing Reports H121(97-2).

Casper, Lynne M. and Suzanne M. Bianchi. 2001. Continuity and Change in the American Family. Thousand Oaks, CA: Sage.

Casper, Lynne M. and Kenneth R. Bryson. 1998. “Co-Resident Grandparents and Their Grandchildren: Grandparent Maintained Families.” Population Division Working Paper No. 26. Washington, DC: U.S. Census Bureau.

Coontz, Stephanie. 2000. “Historical Perspectives on Family Studies.” Journal of Marriage and the Family 62(2):283–97.

Crimmins, Eileen M. and Dominque G. Ingegneri. 1990. “Interaction and Living Arrange- ments of Older Parents and Their Children: Past Trends, Present Determinants, Future Implications.” Research on Aging 12(1):3–35.

DeParle, Jason. 1999. “Life after Welfare: The Grandmothers; As Welfare Rolls Shrink, Load on Relatives Grows.” New York Times, February 21, p. 1.

England, Paula. 2000. “Marriage, the Costs of Children, and Gender Inequality.” Pp. 320–42 in The Ties That Bind: Perspectives on Marriage and Cohabitation, edited by L. J. Waite. New York: Aldine de Gruyter.

Fields, Jason and Lynne M. Casper. 2001. “America’s Families and Living Arrangements.” Washington, DC: U.S. Census Bureau.

Geronimus, Arline T., John Bound, and Timothy A. Waidmann. 1999. “Health Inequality and Population Variation in Fertility Timing.” Social Science & Medicine 49(12):1623– 36.

Goldscheider, Frances. 1997. “Recent Changes in U.S. Young Adult Living Arrangements in Comparative Perspective.” Journal of Family Issues 18(6):708–24.

Goldscheider, Frances and Calvin Goldscheider. 1994. “Leaving and Returning Home in 20th-Century America.” Population Bulletin 48(4):2–35.

Harris, Craig. 1999. “State’s Welfare Windfall to Go to Child Care.” Courier-Journal (Louis- ville, KY), August 31, p. A1.

20 SO CIOLO GICAL PERSPECTIVES Volume 45, Number 1, 2002

Hofferth, Sandra L. 1984. “Kin Networks, Race, and Family Structure.” Journal of Marriage and the Family 46(4):791–806.

Hogan, Dennis P., Ling-Xin Hao, and William L. Parish. 1990. “Race, Kin Networks, and Assistance to Mother-headed Families.” Social Forces 68(3):797–812.

Jarrett, Robin L. 1994. “Living Poor: Family Life among Single-Parent, African-American Women.” Social Problems 41(1):30–49.

Kamo, Yoshinori. 2000. “Racial and Ethnic Differences in Extended Family Households.” Sociological Perspectives 43(2):211–29.

Lichter, Daniel T., George Kephart, Diane K. McLaughlin, and David J. Landry. 1992. “Race and the Retreat from Marriage—A Shortage of Marriageable Men?” American Socio- logical Review 57(6):781–99.

Logan, John R. and Glenna D. Spitze. 1994. “Family Neighbors.” American Journal of Sociol- ogy 100(2):453–76.

Lugaila, Terry A. 1998. “Marital Status and Living Arrangements: March 1998 (Update).” U.S. Census Bureau Current Population Reports P20-514.

Mutchler, Jan E. 1992. “Living Arrangements and Household Transitions among the Unmarried in Later Life.” Social Science Quarterly 73(3):565–80.

Newman, Katherine S. 1999. No Shame in My Game: The Working Poor in the Inner City. New York: Knopf and the Russell Sage Foundation.

Oliver, Melvin L. and Thomas M. Shapiro. 1995. Black Wealth/white Wealth: A New Perspec- tive on Racial Inequality. New York: Routledge.

Raley, R. K. 1995. “Black-white Differences in Kin Contact and Exchange among Never Married Adults.” Journal of Family Issues 16(1):77–103.

Roschelle, Anne R. 1999. “Gender, Family Structure, and Social Structure: Racial Ethnic Families in the United States.” Pp. 311–40 in Revisioning Gender, edited by M. M. Fer- ree, J. Lorber, and B. B. Hess. Thousand Oaks, CA: Sage.

Ruggles, Steven. 1996. “Living Arrangements of the Elderly in America: 1880–1990.” Pp. 254–63 in Aging and Generational Relations: Historical and Cross-Cultural Perspectives, edited by T. Harevan. New York: Aldine de Gruyter.

Speare, Alden and R. Avery. 1993. “Who Helps Whom in Older Parent-Child Families.” Journals of Gerontology 48(2):S64–S73.

Tienda, Marta and Ronald Angel. 1982. “Headship and Household Composition among blacks, Hispanics, and Other whites.” Social Forces 61(2):508–31.

Tienda, Marta and Jennifer Glass. 1985. “Household Structure and Labor-Force Participa- tion of black, Hispanic, and white Mothers.” Demography 22(3):381–94.

Treas, Judith. 1977. “Family Support Systems for the Aged: Some Social and Demographic Considerations.” Gerontologist 17(6):486–91.

Treas, Judith and Ramon Torrecilha. 1995. “The Older Population.” Pp. 47–92 in State of the Union: America in the 1990s, Vol. 2: Social Trends, edited by R. Farley. New York: Rus- sell Sage Foundation.

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(2):309– 37.

U.S. Census Bureau. 2000. Current Population Survey Technical Paper 62: Design and Methodol- ogy. Washington, DC.

Wilson, William J. 1987. The Truly Disadvantaged: The Inner City, the Underclass, and Public Policy. Chicago: University of Chicago Press.

Wister, Andrew V. and Thomas K. Burch. 1987. “Values, Perceptions, and Choice in Living Arrangements of the Elderly.” Pp. 180–98 in Critical Issues in Aging Policy, edited by E. F. Borgatta and R. J. V. Montgomery. Newbury Park, CA: Sage.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.