Doubling Up When Times Are Tough: A Study of Obligations to Share a Home in Response to Economic Hardship

Abstract

This study uses a factorial vignette design embedded in an Internet survey to investigate attitudes toward an adult child and parent living together in response to economic hardship. Over half of Americans said the desirability of intergenerational co-residence depends on particularistic aspects of the family, notably the quality of family relationships. Support for co-residence is greatest when the adult child is single rather than partnered. Support is weaker if the adult child is cohabiting rather than married to the partner, although groups with greater exposure to cohabitation make less of a distinction between cohabitation and marriage. Presence of a grandchild does not affect views about co-residence. There is more support for sharing a home when a mother needs a place to live than when the adult child does. Responses to open-ended questions show that individuals invoke both universalistic family obligations and particularistic qualities of family relationships to explain their attitudes.

1. Introduction

Families can provide an important safety net for their members in times of economic hardship. The recession at the beginning of the 21^st century, with high levels of housing foreclosures and widespread job loss, created new pressures for intergenerational co-residence in response to economic crises. For example, twenty percent of adults in 2009 said they had trouble paying the rent or mortgage in the past year, an increase from 14 percent in 2008 (Pew Research Center, 2009). The increase in “doubled-up” households during the recession, documented by the Census Bureau (Mykyta and Macartney, 2011), suggests that co-residence with kin is one way that families have responded to financial difficulties.

State laws and public opinion view parents as responsible for providing economic support and housing for minor children (Edelstone, 2002; Rossi and Rossi, 1990), but responsibilities of parents and children to each other become more ambiguous once children reach adulthood. Relatively little is known about parents’ views on their obligation to share their home with adult children, or reciprocally, adult children’s views about their responsibility to take in an older parent. The primary source of public opinion data on whether or not parents and adult children should live together comes from a single question asked in the General Social Survey (GSS): “As you know, many older people share a home with their grown children. Do you think this is generally a good idea or a bad idea?” (National Opinion Research Center, 2009). Young people and those in recent cohorts are more likely to view co-residence favorably (Fischer and Hout, 2006). Responses to this question, however, provide limited information about Americans’ views about when it is appropriate for parents and adult children to live together or the reasons for viewing co-residence as desirable. Responses to single-statement attitude questions like those in the GSS provide little guidance about when intergenerational obligations are conditional or how individuals evaluate conflicting norms. Surveys rarely provide an explicit “it depends” response option or seek information about the contingencies respondents think are important. The GSS question also is ambiguous because it omits information about the family context, such as whether the adult child or the parent needs housing help.

In this paper, we report findings from a new study of attitudes about the desirability of parents and adult children living together in response to financial difficulties, with a specific focus on variation in adult children’s family statuses because these have changed rapidly in recent decades. Children delay transitions to adulthood, they cohabit before marrying and they are increasingly likely to have their own children before marriage. Many of these adult children go on to divorce and remarry or cohabit again, much as their parents did. These changes in the demography of U.S. families may be altering the boundaries of family membership and weakening the socio-emotional connections between kin that reinforce the family safety net.

We study how the adult child’s marital and parental status, the child’s gender, which generation needs help, and the anticipated duration of co-residence affect attitudes about intergenerational obligations. Building on past research that uses vignettes to study family obligations (Coleman and Ganong, 2008; Ganong and Coleman, 1999; Nock et al., 2008; Rossi and Rossi, 1990), we use a vignette design embedded in an Internet survey to investigate attitudes about co-residence of parents and adult children. We extend past research by including “it depends” as an explicit response option to measure the extent to which attitudes about the desirability of co-residence are contingent on other factors. We also include an open-ended follow-up question asking respondents to explain their answers. Responses to this question allow us to identify the motivations underlying attitudes about intergenerational co-residence.

2. Intergenerational norms, family change, and the safety net

The notion that families provide a safety net for their members in troubled times is rooted in a belief that individuals share an understanding about their obligations to each other. These shared understandings, or norms, about family members’ responsibilities to each other are not specific to an individual or a particular family, but instead are part of the fabric of the broader culture and reinforced by informal sanctions as well as by laws (Rossi and Rossi, 1990). Evidence that norms about intergenerational obligations influence how much parents and adult children help each other comes from correlational studies (Ganong and Coleman, 2005) as well as longitudinal analyses that show that attitudes about obligations predict individuals’ assistance to parents later in life (Silverstein et al., 2006).

The family safety net for parents and adult children depends on consensus about norms of parent-child obligations, a consensus that may unravel during periods of rapid family change. An effective family safety net also requires a common understanding of who is in the family. Changes in the organization of family life in the second half of the 20^th century may have undermined this shared understanding. High rates of divorce and childbearing outside of marriage have increased the percentage of children who grow up in single-parent households and have loose ties to a biological parent, usually their father (Bumpass and Lu, 2000; Stewart, 2010). The rise in remarriage has increased the numbers of parents and children who have both biological and stepkin ties (Wachter, 1997). Parents are more likely to have adult children who divorce and remarry, increasing the likelihood that members of the older generation become “grandparents” to children other than the biological offspring of their own children.

Obligations between stepparents and stepchildren are weaker and characterized by greater variability than those between biological parents and children (Ganong and Coleman, 1999; Rossi and Rossi, 1990). Indeed, U.S. adults believe that stepfathers’ financial obligations to their stepchildren end if the mother and stepfather divorce (Ganong et al., 1995). Attitudes about an adult stepchild’s obligations to a stepfather also depend on whether he remains married to the child’s mother (Ganong and Coleman, 1999; Ganong et al., 2009).

Perhaps the most profound change in family living arrangements during the past 40 years has been the rapid rise in cohabitation (Bumpass and Lu, 2000). Little is known about how cohabitation affects intergenerational relationships (Silverstein and Giarrusso, 2010), but it is likely that cohabitation reduces parents’ and children’s willingness to co-reside in response to economic difficulties. Older cohorts hold more negative attitudes about cohabitation than do younger cohorts (Gubernskaya, 2008). Parents who disapprove of non-marital cohabitation may be uncomfortable sharing a home with a cohabiting child, even if the parent is experiencing economic hardship and needs a place to live. Children who cohabit may be unwilling to expose their new relationship to the strain of living with a (disapproving) parent. In addition, cohabiting relationships, like those in a stepfamily, have ambiguously defined rights and responsibilities which may weaken consensus about whether it is appropriate for a parent and a cohabiting adult child to live together (Cherlin, 1978; Nock, 1995).

Compared to married couples, cohabiting couples spend less time with both sets of parents, are less emotionally close to their parents, and are less likely to rely on parents for help (Aquilino, 1997; Eggebeen, 2005; Hogerbrugge and Dykstra, 2009). The short duration of cohabiting relationships (Bumpass and Lu, 2000) may disqualify cohabiting partners as full-fledged family members (Powell et al., 2010), reducing parents’ willingness to support the cohabiting partner by extending an offer of co-residence.

Given the substantial changes in the family, particularly the rise in cohabitation, and the increased financial pressures on families to “double up” in today’s poor economy, an assessment of how young adults’ family configurations affect normative support for intergenerational co-residence is overdue. We use a factorial vignette design with an open-ended follow-up question to measure attitudes about whether and when parents and adult children should live together. This experimental design is ideal for eliciting attitudes about aspects of a hypothetical situation that alter individuals’ judgments about normative obligations (Nock et al., 2008). We discuss our hypotheses in the next section and then describe features of the study design in Section 4.

3. Hypotheses

The goals of this study are to assess how the family roles of adult children affect attitudes about co-residence in times of economic need and to explore respondents’ rationales for their attitudes. 1) We hypothesize that sharing a residence is viewed most favorably when the adult child is single, less so when the child is married, and least favorably when the child is cohabiting with a partner. Single adult children are likely to be viewed as not fully launched and therefore still eligible for parents’ assistance in periods of financial difficulty. They are much more likely than married adult children to live with parents (Furstenberg et al., 2005; White, 1994).

If the adult child is either married or cohabiting the prospect of co-residence entails sharing a residence not only with an adult child but also with the child’s partner with whom a parent may not be as comfortable. Once children marry, there also is a strong expectation that the couple should live independently and support themselves (Amato et al., 2007). In situations that involve an unmarried partner, there may be especially strong reluctance to endorse co-residence because of uncertainty about the child’s relationship or disapproval of cohabitation.

2) We expect the presence of a grandchild to increase support for co-residence with adult children of all union statuses, regardless of whether it is the parent or the adult child who needs help. The grandchild may benefit either by having a place to live, if it is the grandchild’s parents who need housing, or by potential child care if it is the grandparent who needs housing. Cohabiting couples who have a child together, a signal that the relationship is a longer-term commitment, are more likely to be considered a family than cohabiting couples who are childless (Powell et al., 2010). Thus, parents may be more willing to live with a cohabiting couple who has children together than with a childless cohabiting couple.

3) Members of groups with greater exposure to cohabitation – cohabitors themselves, younger individuals (Bumpass and Sweet, 1995), Hispanics (Castro Martin, 2002; Oropesa, 1996), and those with less education (Kennedy and Bumpass, 2008) – will be less likely to distinguish between the needs of a cohabiting versus a married adult child. Social groups differ in their understanding of intergenerational obligations (Jackson et al., 2008). When cohabitation is more common, individuals view these non-marital unions more favorably and are likely to have clearer ideas about the responsibilities of cohabiting couples (Manning et al., 2011; Nazio and Blossfeld, 2003). Therefore we expect greater exposure to cohabitation to result in more favorable views toward parents and cohabiting adult children living together.

In addition to the adult child’s family status, we consider other dimensions that may affect attitudes toward co-residence. 4) Norms of equal treatment of children lead us to expect no or limited differences in attitudes about co-residence with an adult son or daughter. 5) Evidence that parents more often support adult children than vice versa (Bengtson, 2001; Furstenberg et al., 2005; Hogan et al., 1993) leads us to expect that support for co-residence will be greater when the parent would provide housing than when the adult child would do so. 6) Finally, respondents should be more approving of co-residence when the arrangement is time limited than when the commitment is for an indefinite duration, regardless of who needs housing.

4. Data and methods

4.1 Sample

This paper uses data obtained through a collaboration between the Center for Family and Marriage Research at Bowling Green State University and Knowledge Networks (KN), which maintains a panel of potential respondents for Internet surveys. KN uses a national probability sample that combines address list sampling with random digit dialing sampling to ensure more complete coverage of the U.S. population than would be possible using a single sampling frame. KN provides Internet access and a laptop to individuals in households that do not already have Internet access. Those who already have Internet access receive points redeemable for cash as incentives for their participation. Individuals who agree to participate in the KN panel complete an initial demographic profile that determines eligibility for inclusion in specific studies. KN recruits panel members by email to participate in studies, which are about 10–15 minutes long. Respondents typically complete one survey a week.

We proposed questions that were included in the KN panel in August 2009. Our sample includes 3,129 respondents age 18 and older.¹ The cooperation rate for our study was 69.9%. The household recruitment rate was 21.2% and the profile completion rate was 57.8%, yielding a cumulative response rate of 8.6%. This response rate is low but because KN provides Internet access to those who do not have it, KN recruits a more representative sample than other Internet panels that recruit from non-probability samples (Chang and Krosnick, 2009). In our analyses we use a study-specific post-stratification weight to adjust the data to the demographic distributions provided by the Current Population Survey.

The Internet survey is ideal for this exploration of how the adult child’s family situation affects attitudes about intergenerational co-residence. Compared to telephone survey respondents, those who answer Internet surveys of probability samples, as ours is, are less likely to provide socially desirable responses (Chang and Krosnick, 2009, 2010; Kreuter et al., 2008). Web surveys also may be better than telephone surveys for cognitively demanding tasks (Fricker et al., 2005). Finally, accumulating evidence suggests that the web mode provides higher quality responses to open-ended questions than other self-administered modes (Smyth et al., 2009).

4.2 Vignette Manipulation

We use the General Social Survey (GSS) question about sharing a home as the framework for our vignette design because it is the most commonly used closed-ended question about intergenerational coresidence, and it has been in the GSS since 1973. We presented each respondent with a short vignette about co-residence in response to economic need. The vignette varies five dimensions of the context: the adult child’s union status (married, cohabiting, single), whether the child is a parent of a young child, adult child’s gender, who needs housing (adult child or parent), and the anticipated duration of co-residence (three months or for an indefinite period of time). An example of a vignette for a cohabiting adult child is:

John and his family are having financial trouble and they have lost their home. John is living with his girlfriend and their young child and they need a place to live for the next three months. John’s older mother lives nearby.

Each respondent was shown one vignette in which the five dimensions of the context were randomly combined. After the vignette, we asked respondents whether it was generally a good idea or bad idea for the child to move in with the mother (or the mother to move in with the adult child). The response options on the screen included “good idea,” “bad idea,” “it depends,” and “don’t know.” “It depends” and “don’t know” are explicit response choices because we are interested in when obligations are conditional or unclear and the reasons that prompt a conditional response. Previous research also suggests that some obligations are conditional on other factors (Coleman et al., 2005; Nock et al., 2008). The GSS, in contrast, allows contingent or “don’t know” responses only when respondents volunteer them. After their initial response about the desirability of co-residence, respondents were asked an open-ended question about why it was a good (bad) idea. If respondents said “it depends” or “don’t know,” they were asked on what their opinion depended, or what made them unsure. Respondents typed their explanations in a text box.

Table 1 shows the vignette wording and the unweighted distributions for the dimensions we varied across vignettes.² There are nearly equal percentages of vignettes with each characteristic of the situation. Five dimensions imply 48 random combinations. The factorial design provides a simple random sample of different hypothetical family conditions. The vignette characteristics are uncorrelated because their values are randomly assigned.

Table 1.

Vignette Design

Vignette
[John/ Mary] [and his/her family] [are /is] having financial trouble and [he/ she/ they] [has/ have] lost [his/ her /their] home. [John/ Mary] is living with [his wife/ her husband/ his girlfriend/ her boyfriend] [and their young child] and [he/ she/ they] [needs/ need] a place to live [for the next three months/ indefinitely]. [John/ Mary]’s older mother lives nearby.
Do you think it is generally a good idea or a bad idea for [John / Mary] [and his wife/ her husband/ and his wife and their young child/ her husband and their young child/ and his girlfriend/ her boyfriend/ and his girlfriend and their young child/ her boyfriend and their young child/ and his/her young child / no fill if single, no child] to move in with [John/ Mary]’s mother?
Vignette Characteristics	Percent (unweighted)
Characteristics of the adult child
Union Status
Married	34.0
Cohabiting	34.2
Single	31.7
Total	100.0
Parental Status
No young child	51.1
Has young child	48.9
Total	100.0
Gender
Male	49.7
Female	50.3
Total	100.0
Circumstances of co-residence
Who needs help
Parent	50.4
Adult child	49.6
Total	100.0
Length of Stay
3 months	49.9
Indefinite	50.1
Total	100.0
Unweighted N	3,129

Notes: The vignette also included combinations in which the adult child was a single parent. For example: [John/Mary] is having financial trouble and [he/she] needs a place to live. [John/Mary] is a single [father/mother]. [He/She] and [his/her] young child need a place to live for ….” Variables are described in the text. Percentages may not sum to 100 due to rounding.

The sample size of 3,129 cases required us to be selective about the number of dimensions to vary in the vignette to ensure enough cases per cell (unique combination of characteristics) for multivariate analyses. The five dimensions result in approximately 65 cases for each cell. We hold constant that the parent is an older, unmarried mother. We vary the relationship status of the adult child rather than of the parent because prior research has used vignettes in mail and telephone surveys to vary the biological relatedness of older parents and characteristics of adult children (Coleman and Ganong, 2008; Coleman et al., 1997; Ganong and Coleman, 1998). Our study complements this research and addresses the lack of knowledge about how cohabitation influences intergenerational obligations (Coleman and Ganong, 2011).

We asked about unmarried mothers instead of fathers because most single parents are mothers. Single women also are more economically vulnerable than single men, making support for older women potentially more crucial than for older men. We expect obligations to mothers to be freer of unmeasured variation than obligations to fathers because custody patterns after divorce disrupt mothers’ childrearing less frequently than fathers’. Sample size limits prohibited us from adding a vignette dimension to distinguish custodial from noncustodial fathers.

Recent and current cohabitors generally prefer to introduce each other as “boyfriend” and “girlfriend” instead of using the terms “partner” or “unmarried partner,” although there is ambiguity about the appropriate term to describe a cohabiting partner (Manning and Smock, 2005). Qualitative evidence also indicates that young adults use “boyfriend” and “girlfriend” to signify that their relationship is reasonably permanent (England et al., 2008), a concept we wanted to communicate in the vignette. Therefore we use these terms to refer to cohabiting partners. The open-ended responses indicate that respondents understood these terms to refer to a cohabiting partner. However, there may be age differences in how respondents describe cohabiting unions. Younger adults may use “boyfriend/girlfriend” to indicate a serious relationship, but older adults may think the vignette describes a more casual union than we intended. We return to the issue of terminology in the discussion in Section 6.

4.3 Covariates and quantitative analysis

We report descriptive statistics and then estimates from a multinomial logistic regression of vignette responses about the desirability of co-residence on the five dimensions of the vignette and the respondent’s characteristics. Table 2 shows respondents’ characteristics and the unweighted numbers of cases. Respondents’ characteristics include: age, gender, race-ethnicity (non-Hispanic white, non-Hispanic black, Hispanic, non-Hispanic other or multi-racial), highest level of schooling completed (less than high school, high school, some college, college or more), employment status (employed, unemployed, not in the labor force), lower-income household defined as income in the two lowest quintiles of the distribution of a KN-supplied categorical variable, union status (married, cohabiting, widowed, divorced or separated, single), whether the respondent lives with children under age 18, residence in a metropolitan statistical area, region (Northeast, Midwest, South, West), and whether the household had Internet access before entering the KN panel.

Table 2.

Sample Characteristics (weighted percentages, unweighted Ns)

	Percent	N		Percent	N
Age			Income in lowest 2 quintiles
18–24	9	214	No	57	1,896
25–34	19	505	Yes	43	1,233
35–44	20	596	Union status
45–54	18	549	Married	44	1,517
55–64	18	617	Cohabiting	8	215
65–74	10	423	Widowed	5	167
75 and older	5	225	Divorced or separated	15	450
Gender			Single	28	780
Female	51	1,552	Children under 18 in HH
Male	49	1,577	No children	69	2,226
Race and Hispanic ethnicity			Children present	31	903
Non-Hispanic White	69	2,439	Urban residence
Non-Hispanic Black	11	253	Does not live in MSA	16	542
Hispanic	13	264	Lives in MSA	84	2,587
Non-Hispanic Other	6	173	Region
Education			Northeast	18	564
Less than high school	12	338	Midwest	22	787
High school graduate	32	927	South	37	1,080
Some college	28	955	West	23	698
College degree or higher	28	909	Internet access
Employment status			Had internet before	62	2,020
Employed	57	1,697	Did not have internet	38	1,109
Unemployed	10	257
Not in labor force	33	1,175

Notes: Variables are described in the text. Percentages for each variable may not sum to 100 due to rounding.

4.4 Coding of open-ended responses

We use the open-ended responses in which respondents explain why co-residence was (un)desirable or why they said “it depends” to shed light on the quantitative results. Nearly all respondents, 95%, provided explanations in the text box. Mean response length among those who responded is 17 words. The median length is 13, and the maximum is 176 words.

We combined deductive and inductive approaches to develop codes for the open-ended responses. First we identified motivations for helping family members, drawing on sociological and economic theory as well as Coleman and Ganong’s (2008) findings about obligations to help parents. Then we added codes from themes that emerged as we reviewed the open-ended data. We repeated this process three times, coding the first reason that respondents provided. When we agreed that no new themes were emerging, two researchers independently coded the remaining cases (n = 2,166) that had not been used to develop the coding scheme. The inter-rater reliability was high (kappa = 0.83). When the two coders disagreed, a third coder independently assigned a code. In 85% of these cases, the third coder served as tie breaker. For the remainder, in which all three coders disagreed, we assigned the “other” code. We recoded all of the cases used to develop the coding scheme (n = 803) following the same procedure.

We identified 12 reasons and a residual “other” category. The substantive reasons are: (1) relationship quality – co-residence would strain existing relationships, or the desirability depends on how well the parent and child got along; (2) family obligation – duties of family members, or adult children should take care of parents; (3) resource constraints – whether the person who would provide housing could afford it, or whether the person needing housing had no alternative and would be homeless without co-residence; (4) short-term exchange – the parent or child could repay help with co-residence by doing housework or providing child or elder care; (5) responsibility for needing help – whether the person bore responsibility for the economic trouble or was trying to resolve the economic problem; (6) (in)dependence – adult children and parents should not live together, or children should be independent, or parents have their own lives; (7) cohabitation differs from marriage – parent does not owe anything to an adult child’s boy/girlfriend, or it would be different if the couple were married; (8) mother’s health – whether the mother needs care that co-residence would facilitate; (9) housing constraints – whether the person who would provide housing has enough space or bedrooms; (10) saves money – whether co-residence would enable the person to get back on his or her feet, or because it makes good economic sense; (11) duration of co-residence – whether the arrangement would be temporary and time-limited; (12) if everyone agrees – whether the parent and adult child agree or whether the spouse or partner of the child agrees.

5. Results

5.1 Attitudes toward co-residence and respondents’ qualitative explanations

Table 3 shows the distribution of attitudes toward intergenerational co-residence. Thirty percent of respondents say it is a good idea for an older mother and adult child to live together to alleviate economic hardship. The vignette elicited negative evaluations from 12% of respondents. Over half, however, report that the desirability of co-residence depends on other factors than those specified in the vignette. Very few say that they do not know if it is a good or bad idea.

Table 3.

Attitudes Toward Co-Residence and Respondents’ Explanations (percentages)

Vignette Responses
	Good idea	Bad idea	It depends	Don’t know	Total
All Respondents	30	12	54	4	100
Explanations from Open-Ended Follow-up
	Good idea	Bad idea	It depends	Don’t know	Total
Relationship quality/conflict	0.7	22.7	52.6	35.4	33.1
Family obligation	48.7	0	0.2	0	14.7
Resource constraints	6.2	2.2	7.4	1.9	6.3
Cohabitation	0.1	32.1	2.0	10.0	5.2
Short-term exchange	11.7	0	2.8	0	5.1
Housing	0.1	0.1	8.2	0	4.6
Duration	7.7	1.3	3.1	3.3	4.3
(In)dependence	0	27.7	1.0	6.0	4.0
Responsibility	0.3	1.4	5.7	1.7	3.5
If agree	0.4	0.3	4.6	4.2	2.8
Mother's health	0.2	0.3	4.8	0	2.8
Saves money	8.4	0	0.4	0	2.7
Other	15.5	11.9	7.5	37.5	11.2
Total (%)	100	100	100	100	100

Notes: Explanations are described in the text. Percentages may not sum to 100 due to rounding. Data are weighted. Unweighted N = 3,129 for vignette responses; 2,969 for explanations.

The second panel of Table 3 shows the distribution of reasons respondents gave to explain their vignette answers. The distribution for the full sample is in the far right column. There are significant differences in the reasons by vignette responses. Among those who say co-residence is a good idea, almost half explain their views by saying that sharing a home is a family obligation. Another 12% think that family members can help each other by trading housing for help of another type in a short-term exchange. For example, one respondent viewed co-residence as desirable because the older mother who needed housing could “help with her grandchild” (ID 2762).

The fact that 32% of those who said co-residence is undesirable explained their negative evaluation by referring to the adult child’s cohabiting relationship suggests that cohabiting unions may diminish the effectiveness of the family safety net. This percentage understates the negative view of cohabitation because only one third of respondents had vignettes with a cohabiting adult child (see Table 1). In vignettes with a cohabiting adult child, almost half (47%) explained their negative reaction to co-residence by referencing the cohabiting union (not shown). Those who said co-residence is a bad idea because the adult child is cohabiting gave responses like: “because they are not married” (ID 2759); “shows her approval of their living arrangement” (ID 709); “Because Mary & her boyfriend could break up @ any givin time, they are not permanint [sic]” (ID 1580).

Other respondents were opposed to co-residence because it would violate the norm of nuclear family independence (28%). One respondent put it this way: “With very few exceptions, it is never a good idea for either parent(s) from either side to live in the same house. You’re [sic] are never comfortable even in your own home when you have someone other than your immediate marriage family living with you” (ID 736). Co-residence was also seen as undesirable because it would create conflict (23%), for example, “family will always hurt you when you live together” (ID 1064) or it “will strain relationships” (ID 1165). None of those who said co-residence is a bad idea explained their responses by referring to family obligations or exchanges. No respondent, for example, said anything like “just because he’s her son, doesn’t mean she should take him in.”

Over half of those who gave a contingent response to the vignette, saying that “it depends,” referred to the quality of the family relationships. Respondents said that co-residence depended on the relationship between the parent and the adult child (“depends on if the mother and daughter get along,” ID 2755), between the parent and the adult child’s partner (“How well Mary’s boyfriend gets along with Mary’s mother,” ID 2841), or among all the parties (“It depends on the character of the individuals and the nature of the relationships,” ID 1029). Other respondents who gave contingent responses said it depended on the availability of space in the home (8%), resource constraints (7%), and who bore responsibility for the economic hardship (6%). Relationship quality was also a common explanation for those who gave a “don’t know” response to the vignette, but over a third of “don’t know” responses gave explanations that fell into the residual category, other.

5.2 Multivariate analyses of attitudes about intergenerational co-residence

Table 4 reports the results of the multinomial logistic regression of the desirability of co-residence on the vignette dimensions and respondent characteristics. The reference category is “bad idea.” We show odds ratios and Z-statistics.

Table 4.

Odds Ratios (OR) from a Multinomial Logistic Regression Predicting Attitudes Toward Co-Residence (“Bad idea” is reference category)

	Good idea vs. Bad idea		It depends vs. Bad idea		Don’t know vs. Bad idea		p value for joint Wald test
	OR	Z	OR	Z	OR	Z
Vignette Characteristics
Adult child union status (married)							< .01
Cohabiting	0.18	−8.55	0.34	−6.22	0.50	−2.10
Single	3.34	4.77	2.22	3.25	1.32	0.66
Adult child has young child	1.11	0.65	0.96	−0.27	1.24	0.79	.46
Adult child is man (woman)	1.38	1.96	1.31	1.86	0.91	−0.31	.13
Adult child needs help (parent)	0.60	−3.18	0.85	−1.14	0.67	−1.39	< .01
Indefinite stay (3 months)	0.49	−4.21	0.84	−1.20	1.09	0.30	< .01
Respondent Characteristics
Age, in years	0.97	−4.87	0.98	−3.56	0.96	−3.37	< .01
Male (female)	0.98	−0.14	0.89	−0.76	1.12	0.39	.67
Race and Hispanic ethnicity (Non-Hispanic White)							.16
Non-Hispanic Black	1.08	0.28	0.81	−0.81	1.21	0.40
Hispanic	1.94	1.99	1.36	0.99	1.63	1.08
Non-Hispanic Other	2.01	1.73	1.25	0.55	2.19	1.13
Education (college +)							.02
Less than high school	0.66	−1.33	0.56	−2.12	2.66	2.12
High school graduate	0.73	−1.50	0.73	−1.62	1.79	1.50
Some college	0.82	−0.86	0.95	−0.25	1.79	1.46
Employment status (employed)							.62
Unemployed	0.86	−0.46	0.97	−0.10	0.37	−1.76
Not in labor force	1.08	0.39	1.01	0.05	0.76	−0.70
Income in lowest 2 quintiles	0.96	−0.20	1.24	1.30	1.79	1.66	.08
Union status (married)							.77
Cohabiting	1.03	0.07	1.15	0.41	1.28	0.41
Widowed	1.05	0.16	1.47	1.35	3.00	1.74
Divorced or separated	1.34	1.10	1.46	1.65	1.48	0.93
Single	1.01	0.03	1.10	0.45	0.80	−0.58
Children live in household	0.95	−0.21	0.95	−0.25	0.67	−0.93	.83
Controls
Lives in MSA	1.27	1.12	1.02	0.12	2.29	2.19	.07
Region (Midwest)							.06
Northeast	0.73	−1.31	0.76	−1.31	0.74	−0.71
South	0.77	−1.28	0.69	−2.07	0.53	−1.89
West	1.36	1.20	1.46	1.67	0.91	−0.24
Did not have internet before	0.87	−0.74	0.93	−0.47	2.36	2.58	.02

Notes: Variables are described in the text. Reference category in parentheses. Data are weighted. Unweighted N = 3,129.

5.2.1. Vignette characteristics

A child’s marital status matters a great deal for whether intergenerational co-residence is seen as a good idea. Respondents viewed co-residence with a single adult child the most favorably. Compared to “bad idea” responses, the odds of “good idea” are 3.3 times greater when the child is single than for a married child. Respondents also are much more likely to say “it depends” (vs. bad idea) when the decision involves an adult child who is single. In contrast, when an adult child lives with a boyfriend or girlfriend, respondents are considerably less likely to say that co-residence would be a “good idea” (vs. bad idea). The odds of responses of “good idea” are only about 20% as high when the adult child is cohabiting as when the child is married. Respondents also are much less likely to say “it depends” than “bad idea” when the adult child is cohabiting compared to when the child is married.

Contrary to our expectations, whether the adult child in the vignette has a young child of his or her own is not associated with attitudes about co-residence. Even when it is the adult child who needs help, having a grandchild does not increase the likelihood of a response that co-residence would be a “good idea” (vs. bad idea). That is, the interaction of adult child needs help by adult child is a parent is not statistically significant (Wald test p = .42; not shown). Open-ended responses shed some light on why having a grandchild who needs help does not affect the desirability of co-residence. The slightly higher rates of “other” responses for “good” and “bad idea” in vignettes with a grandchild hint that respondents are more ambivalent when a grandchild also needs help. They may worry about conflict over who will have childrearing authority but recognize that children benefit from being close to their grandmother.

The adult child’s gender does not have a statistically significant association with attitudes about co-residence. However, respondents do hold more favorable attitudes toward having an older mother move into an adult child’s home than toward having an adult child move into a mother’s home. The odds that respondents view co-residence as a “good idea” (vs. bad idea) are only 60% as great if the adult child needs housing help versus the mother needs help.

Respondents view an indefinite stay as much less desirable than a three-month stay. The odds of “good idea” (vs. bad idea) for an indefinite stay are only about half the odds of “good idea” for a short-term stay. In the open-ended responses to vignettes with the three-month need for housing, 12% of respondents who said living together would be a “good idea” highlighted the short, time-limited duration of co-residence. In contrast, only 2% of those who said an indefinite stay is a “good idea” referred to the duration of co-residence in their explanations.

5.2.2 Demographic variation in attitudes about co-residence

Few characteristics of respondents are associated with attitudes about intergenerational co-residence. Age is associated with lower odds of thinking that co-residence is a “good idea” versus “bad idea,” as well as lower odds of “it depends” versus “bad idea.” Men and women do not differ in their attitudes about the desirability of intergenerational co-residence. Nor are there statistically significant race-ethnic differences in attitudes about the desirability of co-residence. Compared to their college-educated peers, respondents who did not complete high school are less likely to say “it depends” (vs. bad idea). There are no education differences, however, in the contrast between “good idea” and “bad idea.” The other variables measuring socioeconomic characteristics, employment and income, also are not associated with attitudes about co-residence.³

We expected respondents’ own family circumstances to be associated with their attitudes about co-residence. However, neither a respondent’s union status nor living with minor children is significantly associated with views about the desirability of intergenerational co-residence.

5.3 Co-residence in response to mother’s versus child’s need

Respondents were more likely to view co-residence favorably when it was the mother rather than the adult child who needed assistance, contrary to our hypothesis. Respondents’ explanations for their attitudes provide insight into the motivations for their more favorable view of helping a mother with co-residence than helping an adult child. Table 5 shows that respondents are much more likely to explain their favorable attitudes toward co-residence with an older mother who is experiencing economic need by referencing family obligations (59%) than they are to cite obligations when the adult child needs help (35%). When an adult child needs a place to live, exchanging help or making it possible for an adult child to save money are much more common motivations than when the older mother needs help.

Table 5.

Open-Ended Responses, by Who Needs Help and the Desirability of Co-Residence (weighted percentages)

Explanation	Parent Needs Help				Adult Child Needs Help
	Good idea	Bad idea	It depends	Total	Good idea	Bad idea	It depends	Total
Relationship quality/conflict	0.3	34.8	58.6	36.0	1.3	12.7	47.1	30.2
Family obligation	59.4	0	0.2	20.1	34.9	0	0.1	9.2
Resource constraints	5.7	2.5	7.6	6.2	6.8	1.9	7.1	6.3
Cohabitation	0	25.3	1.6	3.9	0.3	37.7	2.3	6.5
Short-term exchange	8.7	0	2.0	4.0	15.7	0	3.5	6.1
Housing	0.1	0.3	7.5	4.0	0	0	8.8	5.2
Duration	7.9	2.4	2.3	4.1	7.6	0.4	3.8	4.5
(In)dependence	0	22.7	0.6	2.9	0	31.8	1.4	5.1
Responsibility	0	0.7	1.5	0.9	0.7	2.0	9.5	6.0
If agree	0.1	0	4.0	2.2	0.9	0.6	5.1	3.4
Mother's health	0.3	0.7	5.7	3.2	0	0	4.0	2.4
Saves money	3.2	0	0.7	1.4	15.2	0	0.1	4.0
Other	14.5	10.7	7.8	11.0	16.7	13.0	7.2	11.3
Total (%)	100.0	100.0	100.0	100.0	100.0	100.0	100.0	100.0
N (unweighted)	485	178	803	1,502	387	196	852	1,467

Notes: Explanations are described in the text. Percentages may not sum to 100 due to rounding. “Don’t know” responses not shown due to small sample sizes.

The distribution of reasons that co-residence would be a “bad idea” also depends on who needs help. Respondents are almost three times more likely to worry about how co-residence would affect the quality of family relationships when the mother needs help than when the child does (35% vs. 13%, respectively). The open-ended data suggest that this is due to concerns about the effects on relationship quality of the role transition from having authority over the household, typically the parent’s right, to living in a home where the child holds more authority. Examples of these responses include: “It usually does not work once the young adults have lived out on their own. The parents still want to parent” (ID 255); “Mothers never completely lose the ‘mothering’ instinct, and, with John being an adult, they are sure to clash” (ID 1894).

5.4 What clarifies obligations if an adult child is cohabiting?

We investigated several aspects of the vignette situation and respondents’ characteristics that we thought would clarify uncertainty about the obligations associated with cohabiting relationships. To do this we included a series of interaction terms in models, building from the multinomial logistic regression summarized in Table 4. We entered each interaction term or set of interaction terms by itself. We first examined aspects of the cohabiting child’s characteristics in the vignette. We found that the negative association between cohabitation and the desirability of co-residence does not vary by whether the child or mother needs help (Wald test p = .23, not shown). There also was no evidence that attitudes about co-residence differed by whether or not the adult child and his girlfriend/her boyfriend had a young child (Wald test p = .60, not shown).

We next examined whether respondents’ own characteristics (cohabiting status, age, Hispanic ethnicity, and education) fostered more favorable attitudes toward intergenerational co-residence when the adult child is living with a boyfriend or girlfriend. There were statistically significant interactions between the union status of the adult child in the vignette and three respondent characteristics: age, race-ethnicity, and education. Table 6 shows the predicted probabilities of each response category for selected contrasts, evaluating other variables at the sample means.

Table 6.

Predicted Probabilities of Attitudes toward Intergenerational Co-Residence, by Adult Child’s Union Status and Respondent’s Characteristics

Adult Child Union Status	Respondent Characteristics	Good idea	Bad idea	It depends	Don’t know
Cohabiting	30 years old	.23	.14	.58	.05
Married	30 years old	.34	.06	.57	.03
Cohabiting	70 years old	.08	.37	.53	.02
Married	70 years old	.27	.11	.61	.02
Cohabiting	Non-Hispanic White	.14	.24	.60	.03
Married	Non-Hispanic White	.29	.09	.60	.03
Cohabiting	Hispanic	.31	.18	.46	.05
Married	Hispanic	.31	.05	.62	.02
Cohabiting	Less than HS	.21	.27	.39	.13
Married	Less than HS	.25	.14	.54	.07
Cohabiting	HS or more	.15	.23	.60	.03
Married	HS or more	.31	.08	.60	.02

Note: Probabilities are generated using three separate models based on Table 4. Each model adds an interaction between the adult child’s union status and one of the three respondent characteristics: age, race-ethnicity, education. The probabilities are evaluated at the means of the other variables in the model.

The results in Table 6 show that both older and younger respondents view co-residence with a married adult child more favorably than with a cohabiting child. However, the married versus cohabiting difference in favorable responses is much greater for older adults than for younger adults. For 70 year olds the probability of saying co-residence is a “good idea” is .27 if the adult child is married, more than three times the probability of “good idea” (.08) if the child is in a cohabiting relationship. For 30 year olds the probability of “good idea” when the adult child is married is only 1.5 times the probability for when the adult child is cohabiting (.34/.23).

We find that Hispanics, a group with high rates of cohabitation, think intergenerational co-residence is equally desirable for married and cohabiting adult children (.31). In contrast, Whites are significantly more likely to consider co-residence favorably for a married versus cohabiting adult child (.29 vs. .14). Less educated respondents, another group with high cohabitation rates, do not differentiate married and cohabiting adult children. Among respondents who did not complete high school, the predicted probabilities of saying that co-residence is a “good idea” are about the same whether the adult child in the vignette is married or cohabiting (.25 vs. .21). For better educated respondents, however, the probability of evaluating co-residence positively is twice as high for a married child as for a cohabiting child (.31 vs. .15).

There is no evidence that cohabitors have clearer ideas about the kin obligations of a cohabiting adult child than do those who are married. The test of the interaction of respondent’s own cohabitation status by whether or not the adult child in the vignette was in a cohabiting relationship indicates that cohabiting and married respondents hold similar attitudes about the desirability of co-residence with a cohabiting adult child (Wald test p = .24; not shown).

6. Discussion and conclusion

In this study, we used an experimental factorial vignette to study how family characteristics of adult children affect attitudes about the desirability of intergenerational co-residence in response to financial hardship. We also allowed respondents to express ambivalence about what was the “best” or “right” thing to do by offering an “it depends” category in the closed-ended response to the vignette and asking respondents to explain their answer in an open-ended follow-up question. We find that over half of respondents say “it depends” when presented with a scenario where parents and adult children may need to share a home due to economic hardship. In a society like the United States where independence and self-reliance are valued, but family ties are also important, the majority of respondents seem to understand intergenerational co-residence as neither fully “good” nor “bad.” Instead, they see it as contingent on other factors, particularly the quality of the relationships among parents, their adult children, and their children’s partners. Respondents also frequently express concern about the amount of family conflict the co-residence is likely to engender.

In the General Social Survey (GSS) time series that partly motivated our vignette study, there has been a steady increase in the proportion stating that intergenerational co-residence is a “good idea,” with over half choosing “good idea” in a question that only offers two opposing choices (authors’ calculations). We find Americans to be much more ambivalent than answers to the GSS question suggest. In response to the vignettes, only 30% of Americans state that co-residence is a good idea and only about half of these respondents base that assessment on the notion that families are obligated to provide housing assistance. Our findings suggest that normative support for intergenerational assistance through co-residence is relatively weak in the United States. At a time when an increasing share of American families faces housing foreclosures and unemployment, doubling up in households to reduce these hardships is not viewed in very favorable terms.

Demographic changes in the family have increased behaviors, such as cohabitation, that make many respondents reluctant to view co-residence positively. As we hypothesized, respondents express more favorable attitudes toward helping a single adult child than a child who has already taken on the adult role of spouse or cohabiting partner, perhaps because respondents are guided by the norm that parents are responsible for launching children into these adult roles. Respondents are especially negative in their appraisal of situations that involve cohabiting children. Close to half of respondents whose vignettes included a cohabiting adult child said explicitly that they thought co-residence between a parent and a cohabiting adult child was a bad idea because the couple was unmarried.

As expected, we find that respondents from groups with more exposure to cohabitation, those who may have clearer, more favorable ideas about the role of cohabiting partners in the family network, are less likely to distinguish between cohabiting and married adult children. This suggests that changes are underway that will increase the acceptance of cohabitation and reduce the ambivalence about helping children who are not married to their partner. Younger respondents are less likely to distinguish between married and cohabiting adult children compared to older respondents, but co-residence with a married adult child is viewed more favorably regardless of the respondent’s age. The age difference in the extent to which respondents distinguish between married and cohabiting adult children might reflect cohort differences in exposure to cohabitation. Alternatively it might be due to age differences in the interpretation of “boyfriend” or “girlfriend,” the terms we used to describe cohabiting partners. Compared to older respondents, younger respondents may consider “boyfriend” and “girlfriend” to be terms describing more committed, stable relationships and thus distinguish less between cohabiting and married adult children. We also find that Hispanics and respondents with less than a high school education report equally favorable attitudes about co-residence regardless of whether the adult child in the vignette is cohabiting or married. Still, the opposition to, or at least ambivalence about, cohabitation is a striking finding in our study.

There is a strong consensus in the literature that affective solidarity is stronger down than up the generational ladder (Giarrusso et al., 2004), but some of our findings suggest that the same is not true for intergenerational obligations. We expected that the vignette would elicit more favorable evaluations of co-residence when the adult child had a young child, but this was not the case. The lack of attention to a young grandchild’s needs puzzled us in light of the importance of grandparents when grandchildren experience family crises (Cherlin and Furstenberg, 1986). In the open-ended data, some respondents were worried that the grandmother and adult child might have disagreements about childrearing. Helping grandchildren without interfering with parents’ authority may be more difficult when help is in the form of co-residence instead of financial contributions or occasional babysitting.

We also expected more support for co-residence in vignettes where it was the adult child who needed help, given the finding in many studies that parents are more likely to help children than vice versa. We found exactly the opposite: Respondents were more likely to view living together as a good idea when the mother needs help than when the adult child does because, to paraphrase respondents’ words, this is what families are supposed to do. Responses to the open-ended question suggest that respondents may be ambivalent about transfers to adult children because of the two conflicting norms of always providing for family members in need, on one hand, and raising children to be independent and stand on their own two feet, on the other hand. For example, when an adult child needs help, respondents are less likely to explain their positive views by invoking universalistic family obligations. Instead they are more likely to say it is a good idea if the adult child compensates the mother by helping around the house as part of a short-term exchange or if sharing a home allows the adult child to save money (presumably so that he/she could establish independence in the future). This ambivalence is consistent with findings reported by Pillemer et al. (2007) that mothers express mixed feelings toward an adult child when the mother gives more help than she receives from the child.

It is possible that attitudes about the desirability of helping mothers are more favorable than attitudes about helping fathers. Thus our results may overstate parents’ relative advantage in this type of intergenerational assistance. High percentages of single-mother households due to divorce and non-marital childbearing disrupt father-child relationships early in life in ways that may diminish both fathers’ and adult children’s willingness to share a home later in life. Disengaged fathers’ and their adult children’s views about their connections to each other and their feelings about whether or not they are part of the same family are increasingly significant research questions (Coleman and Ganong, 2011).

In this study, we chose the language of “good” and “bad” idea because it is used in the question on attitudes about co-residence available in the General Social Survey, the only survey with trend data on this attitude. However, this language may not tap what individuals feel they should do or what they would do if they themselves faced these circumstances. Our finding that family obligation is only one of several motivations for co-residence suggests that the distinction between what is desirable and what is the right way to behave may be important. Future research should investigate how attitudes about the desirability of co-residence compare to attitudes about what individuals should do. Beliefs about whether sharing a household is a good idea may differ from beliefs that family members are obliged to share a household when one member needs help. People do things because they feel they should even when they recognize that fulfilling an obligation may come at a cost and therefore be a bad idea in some ways.

Anticipating how individuals are likely to respond to economic crises like the current recession was one motivation for our study. Our underlying model assumes that attitudes about what is desirable predict behavior, in this case, whether parents and children will share a home to alleviate economic hardship. Although there is some evidence that individuals’ attitudes about family obligations affect their behavior (Ganong and Coleman, 2005; Silverstein et al., 2006), vignettes may be better at eliciting attitudes about what people should do than how they will actually behave (Collett and Childs, 2011). Untangling how general attitudes, personal feelings of responsibility, and helping behavior are related are important challenges for understanding the durability of the family safety net. Ascertaining how the generations view their obligations toward each other and under what conditions they act in accord with these views, should remain a high priority for future research. Families have undergone rapid change in recent decades. If we expect families to be the “fall back” when members experience hardship, we must understand when family change erodes feelings of obligation as it appears that cohabitation does. Although studying the normative aspects of the family safety net does not directly address policy concerns, it is important to understand the context that sets limits on what individuals consider to be their responsibility toward others and that defines which “others” they consider worthy of support. This study is a critical first step toward understanding the conditions that may affect family members’ willingness to assist each other in times of economic downturn and uncertainty.

Highlights.

• We use a factorial vignette to study attitudes on intergenerational co-residence.

• There is more support for co-residence with a single vs. partnered adult child.

• There is less support for co-residence if the adult child is cohabiting vs. married.

• Co-residence attitudes depend on relationship quality.

• Universalistic obligations and particularistic aspects of families explain attitudes.