Do Young Mothers and Fathers Differ in the Likelihood of Returning Home?

Abstract

Building on research examining “boomerang” adult children, I examine multigenerational living among young parents. Returning home likely differs between young mothers and fathers given variation in socioeconomic characteristics, health and risk-taking, their own children’s coresidence, and union stability. Using the National Longitudinal Survey of Youth 1997 (NLSY97), I find that more than 40% of young parents (N = 2,721) live with their own parents at first birth or subsequently. Mothers are generally less likely to move home than fathers but only when not controlling for child coresidence and union stability. Individuals who live with all their children are less likely to return home, and controlling for child coresidence reverses gender differences, though this association disappears in the full model. Young parents who are stably single and those who experience dissolution are highly likely to return home compared to the stably partnered, with the association significantly stronger for fathers than mothers.

In American society, it is widely accepted that the young adult years are a time of immense change and multiple transitions (Arnett, 2000; Settersten & Ray, 2010). As young adults – those under age 30 – try to figure out their long-term plans and goals, they move in and out of households, relationships, schooling, and jobs. Coresidence with parents – either through living with parents continuously or “boomeranging” back home – is common during this period (Fry, 2013; Newman, 2013). While most of the transitions are reversible – one can break up with a partner, go back to school, or change jobs – becoming a parent is not. In 2006–2010, about 55% of women and 42% of men aged 25–29 were parents (Martinez, Daniels, & Chandra, 2012).

Thus, the instability that typically accompanies young adulthood presents several challenges overall, but some may be unique to young parents. One of these challenges is logistical: where, and with whom, do young parents live? Though parents are less likely to move back home than non-parents (Stone, Berrington, & Falkingham, 2014), young parents’ unstable relationships and low incomes may impede residential independence. As multigenerational households have become increasingly common, with the modal category consisting of a householder (grandparent generation, or G1), an adult child (middle generation, or G2), and a grandchild (G3) (Lofquist, 2012), the growing body of research on the topic has largely focused on the grandparents (G1) (e.g., Keene & Batson, 2010; Stykes, Manning, & Brown, 2014) or the grandchildren (G3) (e.g., Dunifon, Ziol-Guest, & Kopko, 2014; Kreider & Ellis, 2011). The living arrangements of the middle generation (G2) remains an understudied topic beyond some work on teen parents (Trent & Harlan, 1994), minority parents (Cohen & Casper, 2002), and disadvantaged young mothers (Pilkauskas, 2012). Young fathers have received less attention, yet they, too, may return to their own parents’ home. Because the reasons young mothers and fathers return home may differ, this paper examines gender differences in the likelihood of returning to the parental home among young parents using several waves of the National Longitudinal Survey of Youth 1997 (NLSY97).

Background

From a life course perspective, leaving the parental home is a key step in the transition to adulthood, and the age at home-leaving has been gradually rising (Furstenberg, 2010). In 2012, more than a third of young adults aged 18–31 lived with their own parents (Fry, 2013). Along with social changes in the meaning and desirability of adulthood (Kimmel, 2008), long-term economic shifts have impeded young adults’ ability to achieve independence because moving – and staying – out of the parental home is partially dependent on the labor market (Card & Lemieux, 2000; Bell, Gurtless, Gornick, & Smeeding, 2007). Returning to home is so common, in fact, that there is a term for these young adults: “boomerang kids” (Newman, 2013).

Young mothers and fathers

Economic and structural impediments to the transition to independence (Newman, 2013) have dominated the research on boomerang kids, so little is known about young parents’ experiences of multigenerational living, and this is a major oversight. Since many young adults are parents, the return of young parents (G2) into their own parents’ (G1) household can mean their own children (G3) are potentially experiencing residential changes and multigenerational households, both of which are generally negatively linked to well-being among children (Adam, 2004; Black et al., 2002), though some studies find that multigenerational living is beneficial for children (DeLeire & Kalil, 2002). Some characteristics, such as the middle generation’s (G2) age at birth, race/ethnicity, and G1 family factors (i.e., structure, education, own age at birth) have been identified as correlates of multigenerational living among young parents (Caputo, 2001; Cohen & Casper, 2002; Trent & Harlan, 1994).

Less is understood about differences in young parents’ multigenerational living by gender, yet parenthood is a highly gendered experience. I outline several reasons that young parents’ gender is important for returns to the parental home. At the broadest level, work on intergenerational assistance shows that daughters receive (and give) more support than sons (Davey, Janke, & Savla, 2004), especially when daughters’ perceived needs are high (Suitor, Pillemer, & Sechrist, 2006). Similarly, adult children (G2) who have their own children (G3) receive more assistance from their own parents (G1) than the childless (Fingerman, Miller, Birditt, & Zarit, 2009). It is unclear whether this varies by child coresidence, but it seems reasonable that the G1 generation would be particularly invested in helping out their own grandchildren (G3); mothers’ higher rates of living with their own children (G3) compared to fathers might increase the odds that they (G2) return to the parental home. However, other work finds that sons are more likely to be coresident (Payne, 2012; Vespa, Lewis, & Kreider, 2012), perhaps men are less likely to have actually left the parental home (Copp, Giordano, Longmore, & Manning, forthcoming) since their ages at achieving various adult transitions is later than women’s (Oesterle, Hawkins, Hill, & Bailey, 2010). To the extent that returning to the parental home, as a form of intergenerational assistance, follows other work finding gendered variation in parents’ (G1) aid to adult children (G2), I derive the following hypothesis:

Hypothesis 1: Mothers are more likely to return to the parental home than fathers, net of basic demographic and family background variables (respondent’s age at birth, race-ethnicity, adolescent family structure, and maternal (G1) education and teen fertility).

Disadvantage among young mothers and fathers

Young mothers and fathers, though, differ in ways that may mediate the association between gender and returning to the parental home. I focus on two broad areas of variation: disadvantage and parenting context. Disadvantage is a multifaceted concept, but here I highlight two main aspects: (1) economic disadvantage, and (2) health and risk-taking. Looking first at economic disadvantage, it is well-established that young parents are socioeconomically disadvantaged compared to both their childless peers and those who delay parenthood (Edin & Tach, 2012), and socioeconomic disadvantage is associated with multigenerational living (Vespa, Lewis, & Kreider, 2012). Although young women have higher levels of education than young men (U.S. Department of Education, 2015), young mothers may face more financial challenges than young fathers. Parenthood more strongly inhibits young mothers’ labor force participation than fathers (Dey, 2014) and may also encourage mothers to pursue additional education (Wilsey, 2013), both of which may increase the need to live with one’s own parents to offset financial difficulties. Young mothers are more likely to receive various forms of government assistance (cash transfers, housing vouchers, food stamps) than fathers, but such assistance likely does little to offset the overall financial disadvantage young mothers face relative to their male counterparts. Drawing on the literature about disadvantage and multigenerational living, I posit the following hypothesis:

Hypothesis 2: Accounting for economic factors attenuates or reduces any baseline gender differences in returns to the parental home.

Another difference between young mothers and fathers may lie in their health and risk-taking behaviors. In general, the more problems adult children (G2) have, the more help they receive (Fingerman et al., 2009). For instance, poor health is associated with multigenerational living (Suitor et al., 2006), and women report worse health than men (Hosseinpoor et al., 2012). Risk-taking may be another type of need; this is also gendered but in the opposite direction, with men more likely to use drugs and alcohol and engage in crime (Cotto et al., 2010; Holmila & Raitosalo, 2005; Snyder, 2010). Young parents often view parenthood as a “turning point,” spurring them to improve their lives (Edin & Kefalas, 2005; Edin & Nelson, 2013). However, parenthood more strongly and negatively affects risk-taking for mothers than fathers (Laub & Sampson, 2001; Kreager, Matsueda, & Erosheva, 2010), though not all work finds evidence of desistance or gender differences (Giordano et al., 2002). Individuals with poor health, drug and alcohol problems, or involvement in the criminal justice system may have difficulty handling the adult responsibilities that accompany independent living (i.e., paying bills on time, following rental agreements, maintaining relationships with roommates and partners), increasing their risk of residential instability while limiting their living arrangement options. Overall, poorer health may increase mothers’ need to move back home and may be deemed particularly worthy of parental (G1) aid, since health problems are not generally due to individual behavior or fault. Risk-taking (binge-drinking, drug use, and criminal activity) may increase fathers’ need to move back home, but substance issues may be less obvious, and both substance use and criminal activity may not be considered as worthy of aid. Thus, considering gender differences in risk-taking and health leads to the following hypothesis:

Hypothesis 3: Accounting for health and risk-taking factors attenuates or reduces any baseline gender differences in returns to the parental home.

The parenting context of young mothers and fathers

Hypotheses 2 and 3 largely address how disadvantage, in one form or another, may be the primary mechanism driving both coresidence overall and gender differences in coresidence more specifically. Another angle, though, would be that differences in returns to the parental (G1) home are driven primarily by issues regarding the context of parenting: child-related factors and union stability. Considering child-related factors first, there is a large literature stressing the role of grandparents in providing direct and indirect assistance to grandchildren, positioned in a framework of functional solidarity (e.g., Cherlin & Furstenberg, 1992; Fuller-Thompson, Minkler, & Driver, 1997; Luo, LaPierre, Hughes, & Waite, 2012). By default, young parents have young children, and the depth and intensity of the needs of young children are well-known. As such, young parents may need extensive social, economic, and instrumental support, and the aid provided by their own family members is an important source of relief and support (Pebley & Rudkin, 1999). These needs, however, might be gendered. On the one hand, mothers more often live with their children than fathers, which might elevate young mothers’ odds of returning to their own parent’s home to the extent that mothers’ needs stem from instrumental and logistical support. Additionally, grandparents might feel more obligated to take in their adult child and grandchild(ren) than their adult child alone (Eggebeen & Hogan, 1990). On the other hand, mothers’ tendency to be the primary parent might increase their sense of competency and thus reduce the need to move back home relative to fathers, who as secondary parents may feel more comfortable interacting with their children in the presence of other adults. Other relevant factors include whether the young parent (G2) lived with their own parent (G1) at their child’s (G3) birth, which is likely more common among young mothers than fathers given age differences at first birth even among teen parents (Scott et al., 2012); living at home at birth would increase the odds of moving back home to the extent that multigenerational living is an established experience. Finally, grandchildren’s ages are inversely associated with intergenerational assistance (Hogan, Eggebeen, & Clogg, 1993), and although having more grandchildren increases the odds of assistance, this assistance is largely restricted to nonresidential care (Luo, LaPierre, Hughes, & Waite, 2012). Thus, having additional children as well as having older children (both of which are more common among young mothers than fathers given age differences in birth and subsequent exposure to additional fertility) likely reduces the odds of multigenerational living. Together, consideration of child-related factors leads to this hypothesis:

Hypothesis 4: Accounting for child-related factors attenuates or reduces any baseline gender differences in returns to the parental home.

Finally, the instability of young parents’ (G2) unions (Edin & Tach, 2012) may be a major factor leading to returning to the parental (G1) home. Goldscheider and Goldscheider (1999) identify marriage as a primary reason for moving out of the parental home, and dissolution is likely a primary reason to return. In general, married and cohabiting young adults are less likely to be living with their parents than those who are unpartnered (Fry, 2013) because couples tend to need and receive less financial support than single individuals (Hogan et al. 1993). Young parents are rarely married - the majority of births to individuals under 30 are nonmarital, and while more than half of such births are to cohabiting couples, these relationships are often unstable (Child Trends, 2015). Union dissolution, in turn, is often accompanied by residential instability, and moving back into the parental home is a common behavior following a break-up (Cherlin & Furstenberg, 1992). Although women are more likely than men to move out of their parent’s (G1) home because they are forming a relationship (Goldscheider & Goldscheider, 1999), women also tend to initiate divorce more often than men (Sayer et al., 2011). If mothers are initiating dissolutions, they may be less likely to move elsewhere, as the custodial (and primary) parent is less likely to leave the family’s home after a break-up (Mulder & Wagner, 2010). Thus, for parents, dissolution may be a risk factor for coresidence with one’s own parent (G1) for young fathers (G2) more strongly than for mothers. Differences in child coresidence may also be captured by union factors (Cherlin & Furstenberg, 1992), as both stably single mothers and those experiencing dissolution may retain physical custody whereas fathers in either situation would be less likely to live with their children. This body of work leads to the following hypothesis:

Hypothesis 5: Accounting for union status and stability attenuates or reduces any baseline gender differences in returns to the parental home.

In sum, I hypothesize that young mothers will be more likely to move back in with their own parents than young fathers but that accounting for socioeconomic characteristics, health and risk-taking, child-related factors, and union stability will mediate gender differences. These factors are will be examined separately to determined which are more relevant for both returns to parental home overall and for explaining gender differences. If returns to the parental home mirror reasons for leaving (Goldscheider & Goldscheider, 1999), union factors will be more strongly related to multigenerational living, but if grandchildren are the primary reasons for intergenerational assistance among young parents (Cherlin & Furstenberg, 1992), then including child-related factors will explain returns to parental home better. If coresidence is largely driven by aspects of disadvantage, then including either socioeconomic factors or health and risk-taking will be most strongly predictive of coresidence. Thus, indicators of model fit will help parse out the relative importance of each set of factors. Finally, it should be acknowledged that Hypotheses 2–5 are contingent upon baseline difference between young mothers and fathers in their return to their own parents’ home. Should there be no initial differences when controlling for only basic demographic factors such as age, race, and family background, I have no a priori expectations about whether including these factors will actually introduce gender differences.

Data and methods

I use the National Longitudinal Survey of Youth 1997 (NLSY97), a representative cohort study of individuals born 1980–84 who were first interviewed in 1997, with yearly interviews thereafter. The original sample of 8,984 respondents included oversamples of black and Hispanic youth. Parents are identified as those who had a child by the round 15 interview in 2011, when respondents were aged 26–32, with an average age of 28.8 years. Of those with children (N = 4,089), the average age at first birth was 22.7 years, so this is indeed a sample of young parents. The analytical sample is further restricted to those for whom their first child was alive and not adopted/in foster care by the last interview in which the respondents participated (excluding 124 cases) and whose first birth appeared prior to their last interview round (excluding 257 cases).

The dependent variable is whether the respondent (G2) lives with his or her biological or adoptive parent(s) (G1). Individuals who were not living with parents at age 12 (those in foster care, group care, living with an unidentified adult, or in some other situation) are excluded (265 cases) since it is unclear whether they have parents to which they could return. Parental coresidence is taken from the household roster created at the annual survey and is defined as living with one or more biological parent(s), including single parent and stepparent families. Because the dependent variable is measured only at the time of survey, the risk of missing parental coresidence increases when the respondent did not participate in multiple survey years, so analyses are further restricted to those with no more than a one-year gap between any surveys (excluding 510 cases). All independent variables are also taken at the time of survey, though many incorporate information from retrospective histories and/or index the full year preceding the survey, as discussed in more detail below. A small number of young parents lived with their parents at birth and continuously through their last survey (N = 212); these are also excluded from the multivariate analysis to produce a final sample size of 2,721 mothers and fathers.

The analyses include a five sets of variables, corresponding to the five hypotheses. The first set includes time-invariant measures of the respondent’s demographic and background characteristics (Hypothesis 1). In addition to gender, these include the respondent’s (G2) age at first birth (categorized as ≤17, 18–22. 23–27, or ≥28 years based on preliminary analyses), race-ethnicity (white, black, foreign-born Hispanic, native-born Hispanic, and other), family structure at age 12 (both biological parents, stepfamily, and single parent), a dichotomous indicator of whether there were grandparents or other relatives in the household at the 1997 interview (to proxy experience living in multigenerational household), mother’s (G1) education (less than high school, high school/GED, Associate’s degree/some college, or Bachelor’s degree or higher), and whether the mother (G1) had a first birth prior to age 18. The second set of measures are time-varying indicators of the respondent’s own socioeconomic status (Hypothesis 2). Employment is taken from the detailed employment histories collected across waves and includes information for the full year rather than only at the time of the survey. For respondents who worked more than one job during the year, the average of total hours across all jobs is used to determine full-time vs. part-time status, with part-time defined as less than 35 hours a week. There are five categories: not employed, employed part-time part of the year, employed part-time year-round, employed full-time part of the year, or employed full-time year-round. Additional indicators include educational attainment (with the same categories as maternal education), enrolled in school at the time of the survey, and number of months of receiving any government assistance in the last year. The third set of variables are time-varying indicators of the respondent’s health and behaviors (Hypothesis 3): self-reported health (with higher scores reflecting poorer health), any binge-drinking in the 30 days preceding the interview, any drug use in the 30 days preceding the interview, and any arrests since the last interview. Binge-drinking, drug use, and health are lagged one year to assess a causal association between health, risk-taking and multigenerational living since these measures are more strictly indexed to the survey date than other variables; using same-year indicators yielded substantively similar conclusions. If a respondent did not participate in the prior interview, they are indexed to the closest prior interview.

The last two sets of variables relate to the respondent’s children and unions. For Hypothesis 4, in addition to a time-invariant indicator of whether the respondent (G2) lived with their parents (G1) at first birth, time-varying child-related measures include the age of the oldest child, the number of children (categorized as only one child, two children, or three or more), and whether the respondent lives with all of his/her children. Finally, union status (Hypothesis 5) is measured by a time-varying indicator of stability since last interview: stably single; single to cohabiting/married; cohabiting/married to single (i.e., dissolved union); and stably cohabiting/married. Alternative specifications yielded no significant differences between cohabitations and marriages in the odds of returning home. As with the lagged measures above, if a respondent did not participate a particular round, union status is indexed to round in which they were last interviewed. Note that these indicators do not explicitly measure whether unions are with the same partner from year to year or with their first child’s other biological parent.

Missing data were present for some family background and socioeconomic measures (nativity, mother’s (G1) education and age at first birth, respondent’s household structure at age 12, months of welfare receipt, and school enrollment status). However, missingness was fairly low (less than 8% for any variable) and data were imputed using the multiple imputation commands in Stata 13.1. All descriptive and multivariate analyses use this imputed data.

Approach

The analysis begins by briefly discussing the living arrangements of young parents before describing the differences in the covariates across gender. The weighted descriptive statistics include the five sets of variables described above; for the time-varying measures, these are measured at the year the respondent entered the analysis. The next stage of the analysis involves event history models predicting the odds that young parents (G2) will live with their own parents (G1) again. I present 6 models, beginning with a baseline model (Model 1), which has gender and the basic demographic/family background covariates, to test Hypothesis 1. Models 2–5 each add a set of factors to Model 1 separately. Model 2 (testing Hypothesis 2) includes the respondent’s own socioeconomic variables. Model 3 (Hypothesis 3) adds health and risk-taking to the baseline model. Models 4 and 5 add child-related factors and union characteristics, respectively, to Model 1 to test Hypotheses 4 and 5. Model 6 is the full model.

The data files were converted to person-years. Individuals enter the analysis the survey year following the survey in which they reported their first birth (if they were not living with their parents) or the year after the first survey in which they reported they did not live with their parents after they became parents themselves (if they were living with their parents at the survey year in which the first birth appeared). Respondents exit the analysis when they first report living with their parents again (the dependent variable) or at their last survey if they remain living independently. Years of observation (ranging from 1–15 with an average of 4.3 years), is included as a categorical variable (1 year, 2 years, 3 years, and 4 or more years).

Results

Figure 1 displays five different living arrangements: 1) those who were not living with their own parents at first birth and did not return home by the last survey, 2) those who were not living with their own parents at first birth but returned home, 3) those who were living with own parents at first birth, subsequently moved out, and had not returned by the last survey, 4) those who were living with own parents at first birth, subsequently moved out, but who returned by the last survey, and 5) those who lived with their own parents continuously from first birth to the last survey. This latter group is included for comparison but are not in the multivariate analyses.

Figure 1.

Multigenerational Living among Young Parents by Gender

*not included in multivariate analysis

A sizeable minority of young parents were not residentially independent between the birth of their first child and their last round of interviewing. Only 58% of fathers and 52% of mothers were residentially independent when they had their first child and had not returned home by the last survey, with the overall proportion (not shown) at 54%. For young parents, then, multigenerational living was a fairly common experience, slightly more so among young mothers. Sixteen percent of young mothers and 15% of young fathers reported living outside of their parents’ home at birth but returning home. Although a greater percentage of young mothers (17%) reported living at home at first birth but subsequently moving – and staying – out than young fathers (12%), a greater percentage of young mothers living at home but who moved out also returned back to live with their parents again. Finally, though not included analytically, about 6% of young parents lived continuously at home from birth until their last interview.

Descriptive Statistics

Table 2 shows the weighted descriptive statistics for each set of explanatory variables for those mothers and fathers who lived independently at birth or thereafter in the first two columns, with the last column displaying the proportion who returned home by those same characteristics. Significant differences in characteristics between men and women are indicated by the superscript “A,” with the superscript “B” indicating significant bivariate differences in returning home within categories of each variable or across values (in the case of linear variables). Of these young parents, the sample is skewed female (58%), and at the bivariate level, there is no significant gender difference in returning home, contrary to Hypothesis 1.

Table 2.

Odds Ratios of Returning Home among Young Parents Living Independently in the NLSY97 (N=2,721)

		Model 1		Model 2		Model 3		Model 4		Model 5		Model 6
Female		0.86	*	0.87	*	0.88		1.24	**	0.84	**	1.03
Background and demographics
Respondent’s (G2) age at 1st birth
≤17 years		1.77	***	1.23	***	1.72	***	1.51	***	1.54	***	1.33	*
18–22 years		1.62	***	1.47	***	1.59	***	1.46	***	1.43	***	1.31	**
23–27 years		--		--		--		--		--		--
≥28 years		0.44	*	0.55		0.45	*	0.47	*	0.50	*	0.62
Race-ethnicity-nativity
Non-Hispanic white		--		--		--		--		--		--
Non-Hispanic black		1.29	***	1.30	***	1.29	***	1.16		1.00		0.98
Native-born Hispanic		1.40	***	1.41	***	1.41	***	1.40	***	1.47	***	1.48	***
Foreign-born Hispanic		1.34	*	1.37	*	1.38	*	1.37	*	1.43	**	1.46	**
Other		0.93		0.94		0.93		0.94		0.93		0.98
Family structure at age 12
Two biological parents		--		--		--		--		--		--
Stepfamily		1.02		0.95		1.00		0.96		0.94		0.88
Single parent family		1.01		0.96		0.99		0.98		0.96		0.91
Lived w/ other relatives at 1st interview		0.91		1.91		0.90		0.93		0.96		0.98
Maternal (G1) education
< High school		1.05		1.01		1.04		1.05		1.07		1.05
High school/GED		--		--		--		--		--		--
AA degree/some college		1.07		1.11		1.06		1.11		1.11		1.16
College or higher		0.82		0.91		0.82		0.81		0.83		0.76
Mother (G1) had a birth <18		0.93		0.91		0.93		0.93		0.88		0.87
Respondent (G2) socioeconomic characteristics (time-varying)
Education
< High school				1.13								1.07
High school/GED				--								--
AA degree/some college				0.59	**							0.61	*
College or higher				0.64	*							0.76
Currently enrolled in school				0.82								0.75	*
Months of gov't assistance in past year				0.99								0.99
Employment status
Not employed during year				1.16								1.13
Employed part-time, part of the year				1.87	***							1.72	***
Employed part-time, year-round				1.35	*							1.28
Employed full-time, part of year				1.80	***							1.63	***
Employed full-time, year-round				--								--
Respondent (G2) health and behaviors (time-varying)
Self-reported health at last interview						1.05						1.01
Binge drinking at last interview						0.96						0.92
Drug use at last interview						1.26	*					1.06
Arrested in the past year						1.29						0.84
Child-related characteristics
Respondent (G2) lived with parents (G1) at birth								1.58	***			1.50	***
Age of oldest child (G3, time-varying)								0.90	***			0.91	***
Number of children (G3, time-varying)
One child								--				--
Two children								0.85	*			0.91
Three or more children								0.68	***			0.78	*
Respondent (G2) lives with all children (G3, time-varying)								0.43	***			0.67	***
Respondent (G2) union stability (time-varying)
Stably single										2.42	***	2.09	***
Single to cohabiting/married										0.65		0.57	*
Cohabiting/married to single										9.17	***	8.02	***
Stably cohabiting/married										--		--
Duration
1 year		--		--		--		--		--		--
2 years		0.71	***	0.71	**	0.71	***	0.80	***	0.73	**	0.83
3 years		0.42	***	0.44	***	0.42	***	0.53	***	0.41	***	0.52	***
≥ 4 years		0.23	***	0.25	***	0.23	***	0.40	***	0.23	***	0.39	***
Constant		0.08	***	0.06	***	0.06	***	0.15	***	0.05	***	0.06	***
Person-years		16,176		16,176		16,176		16,176		16,176		16,176
Pseudo R2		0.0472		0.0578		0.0489		0.0630		0.1181		0.1320

^*p≤.05

^**p≤.01

^***p≤.001

Although mothers were younger at first birth and less likely to live with both biological parents at when first interviewed than young fathers, most of the significant gender differences lie not in background demographics but in the young parent’s own characteristics. Most of the characteristics in this table were measured when the respondent first entered the analysis, either at first birth if not living at home (75% of the analytical sample, not shown) or the first year they reported not living with their parents after having their first birth; the patterns were virtually identical when using characteristics at birth (not shown). Young mothers were advantaged in some ways but disadvantaged in others relative to fathers. On the one hand, young mothers were more educated than their fathers (23% of mothers had some college or more, compared to only 17% of fathers), and more mothers than fathers were enrolled in school (12% and 9%, respectively). On the other hand, more fathers (35%) were stably employed full-time year-round than mothers (11%), and mothers received more months of government assistance than fathers. For health and risk-taking, we again saw contrasting gender patterns. Mothers reported significantly poorer health on average than fathers (higher scores mean poorer health), but a lower proportion reported binge-drinking (15% vs. 36%) and drug use (12% vs. 17%). Significantly more fathers (6%) also reported a history of arrests than mothers (2%). Turning to child-related and union characteristics, more mothers reported living with their own parents (G1) at birth than fathers (29% and 23%, respectively), and mothers’ oldest child was slightly but significantly older on average than fathers. More mothers also reported having additional births. Consistent with other work, significantly fewer fathers (80%) lived with their oldest child than mothers (98%) when they entered the analysis. Although the proportion married was essentially identical across gender, slightly more young fathers (31%) were cohabiting than young mothers (27%), so slightly more young mothers were neither cohabiting nor married than fathers.

As shown in the last column of Table 1, at the bivariate level, many of these characteristics were associated with returning home: younger parents, race-ethnic minority parents, those from non-intact families, and those whose own mother had low levels of education displayed higher proportions returning home. Over 40% percent of young parents without a high school degree returned home compared to only 8% of those with a college degree. Stable full-time year-round employment seemed to be protective against returning home, whereas receiving government assistance was positively associated with returning home. Poor health, drug use, and arrests were also linked to multigenerational living; for instance, 45% of young parents who had been arrested returned home compared to only 26% of those with no arrests.

Table 1.

Weighted Descriptive Statistics of Young Parents Living Independently by Gender and Proportion Returning Home by Key Characteristics

		Men	Women	Returned Home
Overall distribution		42.0%	58.0%
Returned to the parental (G1) home		24.9%	27.4%
Background and demographics
Respondent’s (G2) age at 1st birthA B
≤17 years		2.2%	9.8%	49.1%
18–22 years		38.0%	47.7%	38.1%
23–27 years		50.1%	34.8%	15.0%
≥28 years		9.7%	7.7%	3.7%
Race-ethnicity-nativityB
Non-Hispanic white		62.4%	62.7%	22.9%
Non-Hispanic black		19.9%	19.0%	32.1%
Native-born Hispanic		13.0%	12.0%	33.9%
Foreign-born Hispanic		2.4%	2.8%	36.5%
Other		2.3%	3.5%	22.1%
Family structure at age 12A B
Two biological parents		49.2%	42.0%	22.9%
Stepfamily		6.8%	8.0%	28.9%
Single parent family		44.0%	50.0%	29.4%
Multigenerational household at 1st interview
No		93.0%	94.0%	26.2%
Yes		7.0%	6.0%	28.7%
Maternal (G1) educationB
< High school		22.6%	22.3%	35.3%
High school/GED		42.5%	40.8%	25.0%
AA degree/some college		23.2%	24.6%	25.7%
College or higher		11.8%	12.2%	16.1%
Maternal (G1) age at first birth
< 18 years old		11.2%	11.9%	25.9%
≥ 18 years old		88.8%	88.1%	30.0%
Respondent (G2) socioeconomic characteristics
Education when enter analysisA B
< High school		17.1%	16.6%	41.9%
High school/GED		65.5%	60.3%	27.7%
AA degree/some college		4.7%	7.4%	13.6%
College or higher		12.7%	15.7%	8.3%
Enrolled in school when enter analysisA
No		90.7%	87.8%	26.1%
Yes		9.3%	12.2%	25.2%
Months of gov't assistance when enter analysisA B		.53 mos	2.26 mos
Employment status when enter analysisA B
Not employed during year		27.0%	33.4%	33.1%
Employed part-time, part of the year		7.8%	24.5%	30.9%
Employed part-time, year-round		10.3%	7.2%	19.4%
Employed full-time, part of year		20.3%	23.9%	28.2%
Employed full-time, year-round		34.7%	11.1%	13.7%
Respondent (G2) health and behaviors when enter analysis
Self-reported healthA B		2.16	2.31
Binge drinkingA
No		63.7%	85.0%	25.6%
Yes		36.2%	15.0%	28.9%
Drug useA B
No		82.9%	87.6%	25.0%
Yes		17.1%	12.4%	35.9%
Arrested in past yearA B
No		94.1%	98.3%	25.7%
Yes		5.9%	1.8%	45.4%
Child-related characteristics
Multigenerational household (G1, G2) at 1st birthA B
No		77.5%	71.1%	21.8%
Yes		22.5%	28.9%	39.4%
Age of oldest child (G3) when enter analysisA B		.58 years	.74 years
Additional childrenA B
No		45.1%	34.0%	21.7%
Yes		54.9%	66.0%	29.3%
Oldest child (G3) lives with respondentA B
No		20.3%	1.7%	46.5%
Yes		79.7%	98.3%	24.3%
Respondent (G2) union stability when enter analysisAB
Not cohabiting or married		17.9%	22.9%	41.3%
Cohabiting		31.4%	27.3%	33.5%
Married		50.8%	49.8%	16.1%
N		1,131	1,590	2,721

^ASignificant differences between mothers and fathers at p≤.05

^BSignificant differences in proportion returning home within categories at p≤.05

Child and union characteristics were also strongly associated at the bivariate level with returning home. Nearly 40% of those living with their own parents (G1) when their first child (G3) was born returned home compared to only 22% of those living independently at birth. Although the age of the oldest child was negatively associated with returning home, having additional children was positively associated, with 29% of those with more children returning home but only 22% of those without additional children. Child coresidence was important – more of those living apart from their oldest child (47%) returned home relative to those who lived with their oldest child (24%). Finally, union status was associated with returning home – only 16% of married parents returned home compared to 34% of those cohabiting and 41% of those not in a coresidential union.

The descriptive statistics provided little evidence of gender differences in returning home among young parents, but there were significant differences in the characteristics of young mothers and fathers. These characteristics seemed to be differentially associated with the odds of returning home. To parse out whether mothers and fathers differ in the odds of multigenerational living once accounting for variation in key characteristics, I turn to multivariate models.

Multivariate Results

Table 2 displays a series of nested event history models predicting the odds of returning to the parental home. Model 1 is a baseline model, controlling for gender and basic background factors. Model 2 adds the respondent’s own time-varying socioeconomic characteristics to Model 1, and Model 3 adds the respondent’s time-varying health and risk-taking to Model 1. Models 4 and 5, respectively, add time-varying child-related characteristics and union stability. Model 6 is the full model.

In Model 1, we see that in a multivariate setting, gender was indeed significantly related to the odds of returning home, but the direction of the association contradicts the expectation laid out in Hypothesis 1. Young mothers were less likely to return home than young fathers, by 14%. Background characteristics were associated with returning home, in the expected directions. Relative to those whose first birth occurred between ages 23–27, those who were younger at first birth were more likely to move home whereas those who were 28 or older were less likely to move home; the differences between young teens and those 18–22 were not statistically significant (not shown). Black and Hispanic parents were about 30–40% more likely to move back in with their parents than their white counterparts. Duration was also important – compared to the first year of being a residentially independent parent, the odds of returning home declined over time. The differences between further years of observation were also significant (not shown).

Model 2 added measures of the respondent’s own socioeconomic characteristics to Model 1. Given that the gender differences in the baseline model were in the opposite direction of what was predicted, it was unclear whether adding such characteristics would actually magnify gender differences rather than eliminate them. This did not appear to be the case: Despite differences in these measures, gender functions in the same direction – young mothers remained less likely to return home than fathers. Socioeconomic factors were important in their own right, with better socioeconomic status associated with a lower likelihood of moving in with one’s parents. Those with at least some post-secondary education were about 35–40% less likely to move back home relative to those with only a high school degree. Compared to stably employed adults, those who were employed only part of the year (part- or full-time) were nearly twice as likely to move home, and those who worked part-time throughout the year were about 1.4 times as likely to move home.

Model 3 adds health and related behaviors to Model 1, and gender became insignificant, though it just missed statistical significance at p=0.58. Though this suggests accounting for health and risk-taking somewhat mediates gender differences in returns to the parental home, as laid out in Hypothesis 3, the association was not as expected. The lack of differences between young mothers and fathers in returning home in this model was not because young mothers’ poorer health led to coresidence among women, as I hypothesized; women were less likely to return home in the earlier models, and health was not significantly associated with returning home. Rather, those who reported using drugs at their last interview were more likely to return home in the following year (OR = 1.26) compared to non-users rather than remain living independently. For both Models 2 and 3, the pseudo-R² suggested that these two sets of factors provided little predictive power in terms of explaining why young parents return to the parental home relative to the pseudo-R² in the baseline model.

Model 4 includes child-related characteristics in addition to the baseline covariates, and here an interesting finding emerges: Mothers were then 24% more likely to move back home than fathers. In models not shown, in which each indicator was entered separately, this was entirely due to indicator of whether the respondent lived with all of his/her children. Those who lived with all their children were substantially and significantly less likely to experience multigenerational living (OR = 0.43) than those with one or more nonresident children. Put differently, having any nonresident children increased the odds of moving home. Because men were considerably less likely to be living with all their children than women, accounting for nonresident children reversed the gender differences in the odds of returning home, even when controlling for the number of children. The other child-related indicators were also linked to returning home. Young parents who were living at home when their first child was born were nearly 60% more likely to move back home than those living independently, and having more than one child also reduced the odds of moving back home. The older the age of the first child, the less likely the respondent was to move home (OR = 0.90). Background factors changed little, but the introduction of child-related factors eliminated the black-white difference in the odds of returning home relative to remaining residentially independent. Despite the fact that child-related factors reversed the gender difference seen in earlier models and were highly significant, the pseudo-R² remained quite low (0.0630), only slightly more than in Model 2 (0.0578).

Model 5, which added union instability to Model 1, provided another interesting wrinkle: Mothers were again less likely to return home than fathers when controlling for union stability, by 16%. Compared to those who were cohabiting or married at both the last interview and the current interview, young parents who were stably single were 2.4 times as likely to move back home, and those who experienced the dissolution of a coresidential union were 9.2 times as likely to return home. Conversely, those who formed a new coresidential partnership were no different than those who were stably partnered in the odds of returning home. As with Model 4, black-white differences were no longer significant. Finally, compared to Models 1–4, a larger proportion of the variance was explained by the inclusion of union factors in Model 5 – the pseudo R² is 0.1181.

The full model (Model 6) is presented in the last column, and gender again became insignificant. In models not shown, gender remained significant if all covariates and either union status or living with all children was included but not both. In Model 6, we also saw some changes from earlier models – there were no longer differences between those who had their first child in their mid-twenties and those whose birth occurred later, nor were there black-white differences. Only respondents with some college differed from those with a high school degree in the odds of returning home (OR = 0.61), with being currently enrolled at school emerging as significant (OR = 0.75). Employment continues to influence the odds of returning home but only for those who worked part of the year. Other than parity (having two children was no longer significantly different than having only one child), child-related characteristics remained significant. In particular, living with all of one’s children remained negatively associated with moving home but is reduced in magnitude (OR = 0.67), and this was true even in the presence of controls for union status/stability. The odds for union stability changed fairly little from Model 5 to Model 6 for the stably single and those dissolve a union. Those who moved from single to cohabiting or married, though, were significantly less likely to move in with their parents than remain living independently compared to their stably cohabiting or married peers. Finally, the pseudo-R² changed only modestly from Model 5 (0.1181) to Model 6 (0.1320), suggesting that adding the indicators of economic disadvantage, health and risk-taking, and child-related factors have done little to explain gender differences beyond union factors.

I also disaggregated the full model to examine whether the covariates were differentially associated with returning home by gender (Table 3); significant differences (from Wald chi-square tests comparing coefficients) between mothers and fathers were indicated with the superscript ‘A’. Although some covariates only reached statistical significance as a predictor of returning home for one gender, one characteristic emerged as sharply different between mother and fathers: union status/stability. For all young parents, being stably single or breaking up increased the odds of returning home relative to those who were stably partnered, but the odds were substantially higher for fathers in both these situations. For dissolution, in particular, the odds ratio was about twice as high – mothers were about 6 times as likely to move back home after a dissolution whereas fathers were 12 times as likely. Additionally, for mothers, forming a union significantly reduced the odds of moving back into the parental home compared to those who were stably partnered (OR = 0.38). Supplementary exploration of those who moved back home (not shown) revealed that 45% of these young mothers but only 33% of fathers were moving back home with a partner. Conversely, a greater proportion of young fathers were moving back home after a dissolution (36%) than young mothers (25%) even though more mothers (31%) than fathers (27%) experienced dissolution during the analytical period (not shown). Notably, there was no evidence that child coresidence or any other child-related characteristics work differently for mothers than fathers.

Table 3.

Odds Ratios of Returning Home among Young Mothers (N=1,590) and Fathers (N=1,131) Living Independently in the NLSY97

		Fathers		Mothers
Background and demographics
Respondent’s (G2) age at 1st birth
≤17 years		1.16		1.45	*
18–22 years		1.19		1.34	*
23–27 years		--		--
≥28 years		0.40		0.77
Race-ethnicity-nativity
Non-Hispanic white		--		--
Non-Hispanic black		0.92		1.02
Native-born Hispanic		1.43	*	1.53	***
Foreign-born Hispanic		1.11		1.58	**
OtherA		2.12	*	0.71
Family structure at age 12
Two biological parents		--		--
Stepfamily		0.83		0.95
Single parent family		0.92		0.95
Lived w/ other relatives at 1st interview		0.71		1.17
Maternal (G1) education
< High school		1.11		1.01
High school/GED		--		--
AA degree/some college		1.40	*	1.04
College or higher		1.04		0.77
Mother (G1) had a birth <18		0.77		0.93
Respondent (G2) socioeconomic characteristics (time-varying)
Education
< High school		0.90		1.22
High school/GED		--		--
AA degree/some collegeA		0.19	**	0.82
College or higher		0.69		0.80
Currently enrolled in school		0.83		0.75
Months of gov't assistance in past year		0.99		1.00
Employment status
Not employed during year		1.21		1.02
Employed part-time, part of the year		1.85	**	1.52	**
Employed part-time, year-round		1.24		1.21
Employed full-time, part of year		1.84	***	1.39	*
Employed full-time, year-round		--		--
Respondent (G2) health and behaviors (time-varying)
Self-reported health at last interview		1.09		0.97
Binge drinking at last interview		0.84		1.03
Drug use at last interview		1.04		1.06
Arrested in the past year		0.76		0.98
Child-related characteristics
Respondent (G2) lived with parents (G1) at birth		1.47	**	1.50	***
Age of oldest child (G3, time-varying)		0.95		0.90	**
Number of children (G3, time-varying)
One child		--		--
Two children		0.95		0.95
Three or more children		0.85		0.86
Respondent (G2) lives with all children (G3, time-varying)		0.64	**	0.87
Respondent (G2) union stability (time-varying)
Stably singleA		2.78	***	1.78	***
Single to cohabiting/marriedA		0.99		0.38	**
Cohabiting/married to singleA		12.06	***	6.17	***
Stably cohabiting/married		--		--
Duration
1 year		--		--
2 years		0.71		0.87
3 years		0.46	***	0.53	***
≥ 4 years		0.29	***	0.45	***
Constant		0.05	***	0.05	***
Person-years		5,880		10,296
Pseudo R2		0.1896		0.1067

^*p≤.05

^**p≤.01

^***p≤.001

^ASignificant differences between mothers and fathers at p≤.05

Discussion

This research examined an often-overlooked component of multigenerational households – the middle generation (G2), those living with their own parents but who are parents themselves, with a focus on young parents. Young adults in general often live with their parents as they make the transition to adulthood, and this seems to be particularly the case in more recent cohorts, with “boomeranging” back home a fairly common behavior. However, most research on boomerang kids has focused on individual economic or aggregate structural issues (Newman, 2013), yet about half of young adults have children. The possibility that young adults’ returns home are related to their own family formation issues has been overlooked in recent research despite earlier work identifying family transitions as key to home-leaving and returning (Goldscheider & Goldscheider, 1999). Young parents are particularly likely to need the instrumental, social, and financial support of their own parents.

The descriptive results have indeed shown that young parents utilized their parents as a safety net, with over 40% living with their parents at birth or at some point thereafter. Further, there were important theoretical and substantive reasons to think multigenerational living among young parents would be gendered. Young mothers may face more challenges supporting themselves given the negative impact of parenthood on women’s economic prospects. Young mothers may also benefit from social support and childcare services provided by their own parents (G1) since mothers (G2) overwhelmingly live with their children (G3). Together, this suggested that young mothers were more likely to return to the parental (G3) home than young fathers. This also corresponds with the broader literature on intergenerational assistance, which has shown that daughters are more likely to receive more support than sons (Davey, Janke & Savla, 2004; Suitor, Pillemer, & Sechrist, 2006). The overarching hypothesis, then, was that mothers would be more likely to move back home than fathers, though differences in key indicators of disadvantage and the context of parenting were hypothesized to mediate the gender difference.

The results initially showed significant gender differences but contrary to Hypothesis 1 – young mothers were significantly less likely to move back home than young fathers, even when accounting for socioeconomic factors (Hypothesis 2). Adding health and risk-taking (Hypothesis 3) mediated the gender difference, largely because young adults who used drugs at last interview (disproportionately male) were more likely to move back home. However, when looking only at child-related factors (Hypothesis 4) or union status and stability (Hypothesis 5), gender again became a significant predictor. In particular, it seems multigenerational living works across gender in countervailing directions. Union status and child coresidence tap into similar issues that link gender to returning home in opposite ways. Mothers were more likely to move home when accounting for children’s coresidence (which itself is negatively associated with moving home), but since mothers more often live with their children when single or after a dissolution, union changes alone essentially serve to make gender a proxy for coresidence (but measured in the opposite direction of the actual child coresidence indicator). Including both child coresidence and union stability, as in the last model, again reduced gender to nonsignificance.

The greater improvement to model fit when including union factors was consistent with Goldscheider & Goldscheider’s (1999) work stressing the salience of relationship factors in young adults’ residential independence – just as young adults moved out to form unions, they moved back home when those unions end. The complication seemed to be when those unions involved children – young parents were far more likely to move back home if they did not live with their children, which was more often the case for fathers than mothers. Mothers who were stably single or who experienced a breakup tended to live with their children, which seemed to inhibit returning home despite the potential need for assistance in caring for children, consistent with other work on residential changes after a dissolution (Mulder & Wagner, 2010). It may be easier for the secondary parent to move out than uproot multiple people, or it may be simpler for the G1 generation to accommodate a single person rather than multiple people.

Still, union dissolution seemed to result in returns to the parental home more strongly for fathers than mothers, even when controlling for child coresidence, and when mothers returned to the parental home, they were more often bringing their partners with them than were fathers. If young fathers were moving in with either their own parents or their partner’s parents, then this might be part of a larger story about young men overall – what challenges are they are facing in transitioning to stable adulthood in recent years, and is the transition to adulthood becoming increasingly gendered? Kimmel (2008) argued that young men today are often stuck in a limbo between childhood and adulthood, pointing to cultural changes making adulthood increasingly undesirable for middle-class young men. Economic changes and the restructuring of the labor market have also limited less advantaged young adults’ – and particularly young men’s – ability to achieve some of the markers of adulthood, such as a stable, well-paying job and residential independence (Silva, 2013). Women have also been affected by these changes, but different occupational profiles across gender and women’s higher rates of post-secondary education seem to provide women more avenues to transition to adulthood (Oesterle et al., 2010). The variation in the transition to adulthood by gender has resulted in gendered patterns of multigenerational support. Young men seem to rely on their parents for ongoing support in the form of coresidence (Fry, 2013), reflecting an inability or an unwillingness to manage the responsibilities of residential independence (such as housework, meal preparation, and grocery shopping) and the financial obligations. This is not to say that young women, especially young mothers, are not receiving support from their parents or their partner’s parents; rather, their support may be take the form of services (such as babysitting) or irregular, as-needed financial support.

Though not the primary focus of this paper, there were few differences by race-ethnicity and indicators of disadvantage in the likelihood of returning the parental home; this is interesting given the larger narrative focusing on multigenerational households among poor and minority communities (e.g., Burton, 1990; Chase-Landale, Brooks-Gunn, & Zamsky, 1994). In this analysis, only young Hispanics parents were more likely to return home than whites; there were no black-white differences, and education (either the G1 or G2 generation) is largely unrelated to returning home. The only evidence that disadvantage is linked to returning home is the finding that young parents who only work part of the year (likely experiencing a job loss) were more likely to move back home. Part of the explanation for the lack of differences lies in selection processes – who has an early birth? That is, there are socioeconomic and race-ethnic differences in early fertility, but among young parents, socioeconomic differences may be minimal. The other potential explanation is that the most disadvantaged young parents – those most likely to need to live with their parents – are those not analyzed here because they lived continuously with their own parents from birth, never establishing residential independence.

Limitations

There are several limitations to the analysis. Foremost is the use of yearly residence information. Residence at the survey following the first birth is not necessarily the residence status at first birth if respondents moved between birth and survey. The use of yearly residence information also almost certainly underestimates the prevalence of returning back home as short-term spells between surveys are missed. Gaps in survey participation can also lead to missed spells of coresidence. To limit missed spells, I included those with no more than a one-year gap between any surveys; alternative specifications with only perfect responders (N = 2,170) yielded slight variations in the earlier models but virtually identical results for the full model, and alternate specifications including all individuals regardless of gaps (N = 3,231) yielded virtually identical results across all models. Multigenerational living is also underestimated because the dependent variable is limited to the respondent’s parents and does not include moving in with a partner’s parents. Further, the direction of coresidence changes is unclear; it is possible, though unlikely in such a young sample, that parents are moving in with their adult children rather than adult children moving back home; this may be more likely the case for partnered young parents (Dunifon, Ziol-Guest, & Kopko, 2014). It was also not possible to test whether certain configurations of grandparental (G1) households are associated with moving back home, as parental family/household structure is only known in the years in which respondents live at home. Finally, because the analyses used yearly measures of union status, I did not include measures of whether the respondent’s cohabitations or marriages were with the same partner as the prior surveys, nor did the analyses identify whether newly formed coresidential unions were with the child’s other biological parent.

Conclusion

This research studied a particular group of boomerang kids – young parents – and moved beyond a focus on economic and structural factors to consider other risk factors for moving back home. Young adults in general rely upon their parents extensively for social, instrumental, and economic support, and young parents are no exception. Although the primary contention that moving back home would be more common among women than men was not supported, I found evidence that returns to the parental home are indeed gendered to some extent, with the context in which young parents are raising their children the primary driver for returning home.

For young parents living with their own parents, gender may influence the household and interpersonal dynamics. Among young mothers, most of whom will be bringing their children (and perhaps their partners) with them, performing the parental role while still functioning in the child role themselves may strain G1 and G2 relationship ties and inhibit development of the G2 and G3 relationship. For fathers, most of whom will not have their children all of the time, they may function primarily in the child role while only periodically experiencing the parental role when they have visitation with their children; moreover, since they have fewer opportunities to develop confidence in their own parenting skills, they may rely upon their own parents (G1), further inhibiting the development of parental competency and a strong father identity.

In any case, these multigenerational households warrant more attention to their duration, composition, and function. Returning home may be a temporary phenomenon or more a long-term living arrangement. The relationships within these households may be quite complex, especially when young parents bring their partners into their parents’ home. Another layer of complexity occurs in the households they are moving into – many in the grandparent generation (G1) are single, potentially dating, or are repartnered, and stepfamily relationships can be complicated (Sweeney, 2010). Yet another wrinkle emerges if the G1 generation has other children (minor or adult) living at home; these would be the G2 generation’s siblings and could represent a draw on household resources or a potential source of support. The emerging body of research on family complexity has largely looked at “horizontal” complexity (sibling ties) or two-generation complexity (stepparents and stepchildren) (Manning, Brown, & Stykes, 2014), but family complexity can also emerge – and be quite fluid – based on multigenerational ties and coresidence. Further, family structure instability has been associated with poorer outcomes for children (Cavanagh & Huston, 2008), and there is evidence that these multigenerational families were similarly unstable (Dunifon, Ziol-Guest, & Kopka, 2014). Additional research is needed testing how changes in living arrangements and multigenerational contacts affect child well-being.