Parenthood and Leaving Home in Young Adulthood

Abstract

With increases in nonmarital fertility, the sequencing of transitions in early adulthood has become even more complex. Once the primary transition out of the parental home, marriage was first replaced by nonfamily living and cohabitation; more recently, many young adults have become parents before entering a coresidential union. Studies of leaving home, however, have not examined the role of early parenthood. Using the Young Adult Study of the 1979 National Longitudinal Survey of Youth (n = 4,674), we use logistic regression to analyze parenthood both as a correlate of leaving home and as a route from the home. We find that even in mid-adolescence, becoming a parent is linked with leaving home. Coming from a more affluent family is linked with leaving home via routes that do not involve children rather than those that do, and having a warm relationship with either a mother or a father retards leaving home, particularly to nonfamily living, but is not related to parental routes out of the home.

With the rise in nonmarital fertility late in the twentieth century, the sequencing of transitions in early adulthood has become increasingly complex (Liefbroer and Toulemon 2010; Fussell and Furstenberg 2005), as have the number of choices at each transition point. The traditional transition from parental home to marriage has been replaced by sequences that insert nonfamily living (Goldscheider and DaVanzo 1989), nonmarital cohabitation (Billari and Liefbroer 2010) and increasingly, parenthood (Wu and Wolfe 2001) between living with parents and living with a spouse. As once needy young couples lived with the parents of one or the other, do needy young single parents remain home, possibly longer than if they had not become parents?

Despite the increases in age at leaving home of the 1980s and 1990s (Goldscheider 1997) and the high levels of nonmarital parenthood in the United States, studies of leaving home in young adulthood have not yet examined the role of early nonmarital parenthood in the nestleaving process. For young people making the transition to adulthood early in the twenty-first century, does parenthood retard leaving home or hasten it? Or does it perhaps retard leaving home for very young, unmarried parents but not for those becoming parents at older ages? Do patterns differ for women and men, with their often different responsibilities?

This paper focuses on the relationship between parenthood and leaving home in early adulthood. The data come from the linked Child–Mother and Young Adult samples of the NLSY79. Together, these data provide information from birth through young adulthood about the children born to the women of the NLSY79. Following the tenets of life course and rational choice theories, we first consider the links between parenthood and leaving home, and then consider the determinants of leaving home following a recent birth relative to leaving childless.

Background and Hypotheses

The transition to adulthood includes completing schooling, beginning full-time employment, and entering long-term family roles. Leaving home is an important transition as part of this process. These transitions define the beginning of the adult life course. During the post-World War II baby boom years, these transitions occurred within a relatively short span (Modell et al. 1976); the process has become both more drawn out and more complexly sequenced in recent decades (Fussell and Furstenberg 2005).

The analysis of leaving home began when marriage ceased to be the dominant route out of the home. As recently as the late 1970s, it was reasonable to equate the timing of leaving home with that of marriage (Glick 1977). Analyses that included alternative routes out of the home solved the puzzle that arose during the 1970s: the age at leaving home was decreasing even though the age at marriage was increasing because more left home to nonfamily living. Later research distinguished various nonmarital routes, including semiautonomous living in dormitories or military barracks (Goldscheider and DaVanzo 1989) and living alone or with nonrelatives in independent dwellings (Stone et al. 2011).

Living arrangements during early adulthood, however, continued to diversify. Cohabitation displaced marriage as a first coresidential union among young adults (Raley 2000), and parenthood has increasingly preceded marriage, as well (Wu and Wolfe 2001). Central to the life course perspective is the idea that the timing and sequencing of events in the life course are critical to later outcomes (Elder 1998). Given that two of the central life course events in early adulthood are the splitting off of youth from their parental household and the formation of new families via parenthood, this paper analyzes the links between parenthood and leaving home.

In addition to taking the life course approach, we consider the decision to change living arrangements as a rational choice by young adults.¹ Those making a residential decision have to consider the resources they will have in each living arrangement (Duncan and Hoffman 1990). Parents are more likely to expect to support coresident children than those who have left (except students), even if they are in financial difficulty (Goldscheider et al. 2001). Those who leave the parental home to live with a partner can expect to share expenses and have some access to their partner's resources, whereas those who leave without a partner expect to assume total responsibility. Own and partner resources are likely to be particularly important for young parents' living arrangement decisions, given that children are expensive and child care responsibilities compete with work effort.

These resources, of course, are used to realize preferences. The U.S. nuclear family system, the legacy of the early European-origin settlers to the U.S. (Hajnal 1965; Reher 2005), has incorporated the influences of other cultures and circumstances (such as slavery and its consequences), but, fundamentally, the family formation system values family independence and privacy, which were traditionally achieved via marriage. Hence, those forming families expect to establish a new household. Today, cohabitation rather than marriage is considered the first step, and is nearly as strongly linked with residential independence as is marriage (Goldscheider et al. 1993). It seems likely that unmarried parenthood will also be seen as the basis for a new household.

In the following sections, we address the likely effects of becoming a parent on leaving home and on leaving via different routes, controlling for other sociodemographic measures. We also consider gender differences, given that differences in child custody may mean that the relationship between parenthood and leaving home differs between young men and young women; as well as age differences. Although most new parents would want to leave home to attain an independent residence for their new family, very young parents might need support from their own parents more than they need privacy. Finally, we consider what factors might be linked with leaving home as a parent or not.

Parenthood and Leaving Home

Given the important link between family formation and residential independence, if a birth occurs prior to forming a union, moving out allows young adults to capture the “ideal” family type by establishing an independent family household in which to rear the child. Hence, we propose:

Hypothesis 1

We expect that becoming a parent prior to leaving home will increase the likelihood of leaving the parental home, with or without a romantic partner.

Among the unpartnered (neither married nor cohabiting), however, we expect the association to be stronger for women than for men. Now that single and absent parenthood have become common for females and males, respectively, it is likely that the privacy needs linked with parenthood will increase the likelihood of leaving home for women, but either have no effect, or a negative effect, for young men. Although studies are consistent in showing that young women leave home earlier than young men (for a review, see Chiuri and Del Boca 2010), this is primarily linked with women's younger age at entering a romantic coresidential partnership, and hence with establishing a separate residence; it may not be the same for other young parents. Further, the association between gender, parenthood, and leaving home appears to have changed over time. In a study using 1970s data, males who were already parents were more likely to leave home than those who were not, with no effect for females (Goldscheider and DaVanzo 1989), whereas one using 1980s data found the gender difference reversed, with females more likely to leave (Avery et al. 1992). It will be important to learn if this reversal has continued, or if gender differences have attenuated.

Hypothesis 2

We expect that becoming a parent before leaving home will increase the likelihood of leaving more for women than for men.

Finally, it seems likely that the effects of parenthood on leaving home might be greater for those who have reached adulthood than for younger adolescents, who are likely to need parental support both in terms of reduced housing costs and in terms of child care, which can be more easily attained through coresidence. Further, although adolescents were able to live separately prior to the late 1990s, a result of the uncapped support for unmarried teenage mothers from the welfare programs of that time (Duncan and Hoffman 1990), such support ended with the passage of Temporary Assistance for Needy Families (TANF) in 1996, requiring adolescents receiving support to live at home (Hofferth et al. 2005). Hence, we may expect increased residence at home for younger mothers.

Hypothesis 3

We expect that becoming a parent prior to leaving home will increase the likelihood of leaving less for younger parents than for older ones.

Parenthood as a Route Out of the Home

Above we have considered the arrival of a new baby as a trigger for leaving home. But in a nuclear family system, a baby is also the basis for a new living arrangement and hence should be considered a separate route out of the home. When marriage was the primary route out of the home, analyses that included marriage as a determinant of leaving home obtained coefficients that were too powerful to interpret; only by redefining marriage as a separate route out of the home could the factors associated with getting married be distinguished from other factors linked with the likelihood of leaving home (Goldscheider and DaVanzo 1989). Thus, in addition to considering parenthood as a factor linked with the likelihood of leaving home, we also consider the creation of a single-parent family as a separate route. No research, as far as we can determine, has taken this alternative approach.

This approach also allows us to overcome a limitation of the data, which is that surveys occurred only every 2 years during the key family forming period, so that the sequences between the birth and the observed change in living arrangements may be incomplete. Importantly, this approach also allows us to build on theory and research on early parenthood to help us predict how factors that contribute to becoming a parent at a young age will influence the likelihood of leaving home to an independent residence with children. We will examine the factors predicting leaving home by various routes, including that of becoming an unmarried parent, whether coresident with the child (for women) or not (for men).

The Role of Parental Resources and Relationships

Although own and parental resources influence the decision to leave home, modeling their effects is problematic, as parents can transfer resources to their children who leave home, increasing privacy for both generations (Angelini and Laferrere 2013; Cobb-Clark et al. 2011). Furthermore, the resources needed in the new living arrangement depend on how much is provided by potential coresidential partners, whose characteristics are rarely known in advance (at least to researchers). Women leaving home to marry often wait until a suitably providing spouse has been found (Duncan and Hoffman 1990). Further, in the case of those leaving home without a partner, parental resources might have different effects for those leaving home to live without a child compared with those doing so with a new infant, as many parents may prefer to share a residence rather than provide support for establishing another household with a tenuous chance of success. All of these decisions, of course, might also be subject to the quality of parent–child relationships.

Measures of socioeconomic resources have been regularly included in prior studies of this aspect of the transition to adulthood. The results, however, have been contradictory. This has occurred in part because the effects of parental income on their children's residential independence appear to vary sharply over the transition to adulthood. Both Waite and Spitze (1981) and Iacovou (2010) found that greater parental income reduced the likelihood of leaving home for marriage among very young adults, but increased the likelihood of unmarried residential independence at older ages. The present study avoids this life course problem because it focuses only on leaving home in very early adulthood.

Parental education may also increase the likelihood of leaving home. Although not as directly transferable as financial capital, education as human capital has been linked with attitudes giving priority to privacy over family relationships (Goldscheider et al. 1993), with effects similar in magnitude to that of family income. Mothers' work hours might also have a separate effect, given that greater hours provide a role model of female independence and increase family income; hence, they should increase young adults' leaving home, likely via nonfamily routes. In contrast, having a mother who began childbearing at a young age may increase leaving home via parental routes. Such early childbearing has often been found to be strongly correlated with early parenthood of one's own children, possibly because it is correlated with otherwise unmeasured dimensions encouraging early parenthood (Barber 2001). Consistent with this argument, sons and daughters of mothers who bore a child at an older age are less likely to experience an early first birth themselves (Hofferth and Goldscheider 2010).

When young people had little choice about leaving home until a “suitable” marriage partner was found, relationships with parents likely had to be quite difficult to prompt leaving home early. With the increasing range of choices, however, parental relationships might matter far more. There is no clear consensus about the effects of young adults' relationships with their parents on leaving home. Some studies have found that more negative relationships increase young adults' likelihood of leaving home (Turley et al. 2010; Seiffge-Krenke 2006). One study, however, found the opposite, that poor relationships (“failure to establish… relatedness in adolescence”) predict “difficulty separating from parents in young adulthood” (O'Conner et al. 1996).

It is also not clear how much leaving might depend on relationships with mothers versus fathers. Previous research has indicated that the maternal relationship is more important for the transition to parenthood than the paternal relationship (Hofferth and Goldscheider 2010); however, fathers may be relatively more important in the decision to leave home than they are for the transition to parenthood. We will examine the effects of relationships with each parent during adolescence on leaving home, as well as whether the effects of these relationships differ by route out of home and between young men and women.

Other Family Background and Control Variables

In addition to the measures of parenthood, parental resources and relationship quality, we control for a rich set of sociodemographic indicators measured from early childhood through adolescence prior to the transition to adulthood, reducing the potential for unobserved differences and reciprocal child influence to bias our results. A disrupted childhood family structure has been found to speed young adults' leaving home (Goldscheider and Goldscheider 1998). This may occur because (1) it reduces how much youth benefit from remaining close to their family of origin, (2) the lower resources of most single-parent families provide parents less leverage over their children's behavior, or (3) it increases childbearing (Hofferth and Goldscheider 2010).

Growing up in an urban area delays the transition out of the home (Goldscheider and DaVanzo 1989), likely because of greater access to other opportunities. Racial/ethnic differences in values and expectations may also contribute. Both Black and Hispanic youth have been shown to leave home later than nonHispanic Whites (e.g., Aquilino 1991; Buck and Scott 1993). We also control for indicators of other activities in adolescence and young adulthood, including early sexual experience and participation in delinquent activities. Early sexual experience increases the chance of a birth, as might participation in delinquent activities, which, for girls, often implies early independence from parental control. Educational enrollment has been shown to delay and full-time employment to hasten moving out (Berrington et al. 2013).

Method

Data

The data source is the 1979 National Longitudinal Survey of Youth (NLSY79), which collected data annually from 1979 through 1994 and biannually thereafter; this study used data through 2006. Beginning in 1986, the NLSY79 obtained detailed information every 2 years from the mothers on their children, tested the children using standardized assessments, and obtained information directly from children ages 10–14 (the Child–Mother study). Starting in 1994, children 15 and older were interviewed using questionnaires similar to those given their mothers, and beginning in 2000, these young adults were interviewed as a separate research project, the Young Adult Study (Mott 2002). These data provide information on sons and daughters whom we observe in their late teens and early/mid-twenties, and on their mothers, the women who were 14–21 in 1979 and the subjects of the original study. Specifically, we drew a sample of 5,802 youth (2,949 young adult men and 2,853 young adult women) born in 1978 or later who were ever 14–28 in one of the years from 1988 to 2006, representing 36,598 person years. We did not include young adults born prior to 1978, the earliest year for which complete data on their mothers were collected. The oldest young adults are necessarily the children of very early childbearers and none can be the children of very late childbearers. These constraints mean that no mothers were older than 34 at birth; most were much younger. Therefore, until most of these women's children have become young adults, the cohort of young adults is disproportionately disadvantaged.

Our final sample included 4,674 youth, 2,308 males and 2,366 females, representing 27,519 person years; 8,079 person years (22 % of 36,598) were missing. The main cause of this loss (more than half) was that a significant number of children (4,827 person years, 13 %), were not interviewed between ages 10 and 14, when questions about the children's relationship with mothers and fathers were asked. Another 1,689 person years (4.6 %) were missing on residence of the young adult after the birth (the dependent variable) and had to be dropped. Finally, 1,563 person years (4.2 %) were missing other variables: 221 were missing the number of parental transitions in childhood, 560 were missing age of mother at first birth, and 782 were missing the number of delinquent acts. Because a transition was only counted using stated data, we created family structure indicators only if we had a minimum of 3 years of stated data. An additional twenty cases that did not meet this criterion were dropped. Based upon the literature that has examined the implications of different methods of treating missing data, we first assessed whether data appeared to be missing at random and, based upon this assessment (described below) primarily used list-wise deletion for cases with missing data. In this case, our sample represents a random subsample of the original sample and the estimates will be unbiased with appropriate standard errors (Allison 2002).

There were, however, several variables that we addressed through either mean substitution or substitution with dummy variable adjustment. Three of our variables were unlikely to be missing at random: income of spouse/partner, education of spouse/partner, and relationship of the child to his/her father. We substituted the mother's education for that of the father/partner based upon the high correlation between mothers' and fathers' educational levels. We could not use the same strategy for father's income because we wanted to include an indicator of the mother's contribution to family economic well-being (her annual work hours) separate from that of a spouse or partner. Consequently, in the occasional year missing father/partner income, we substituted the average income of the mother's spouse/partner over the years the child was 0–14. This assumes that the incomes of fathers are available to children even in years in which men are not residing with their children and that the incomes of all potential father-figures, biological, or non-biological, matter; given the importance of father's income for child support and the contributions stepfathers make to stepchildren, this seems reasonable. However, five percent of children never resided with a father. Such children are not randomly selected. For them, an average of the incomes of all fathers in the study across all years was substituted and a dummy variable indicator was included that indicates that this family never reported that a male father-figure resided in the household (1 = never reported a father at any wave; 0 = reported a father at least once).

A similar strategy was taken for another variable, the relationship of the child to his/her father. This variable was taken from the age 10–14 child data. The child could report on the relationship with a residential stepfather, as well as with a biological father (regardless of residence). If the child did not report on the quality of the relationship with either a biological father or a stepfather, we assumed that the quality was unlikely to be good and assigned the poorest relationship code to this variable. Because a dummy variable indicator controls for never residing with a father-figure and this situation was infrequent, the specific value assigned to data missing these three variables should not be overly influential. Simulations of imputation strategies for data missing at random have demonstrated that dummy variable substitution is less desirable than multiple imputation methods (Allison 2002). However, in a recent applied analysis using a family and household data set, the coefficients obtained using mean substitution with or without dummy variable adjustment did not differ from those obtained using list-wise deletion or multiple imputation (Johnson and Young 2011). As expected, simple imputation resulted in slightly smaller standard errors than either of those two alternatives.

Our strategy appears to have been successful. Comparing the original sample with our final sample of 28,519 person years using a t test for mean/proportion differences, we found no differences on most key variables: there were no significant differences in the proportion that left home (6.4 %) and no differences in the proportion that had a child (8.4 %). There were statistically significant differences on a few of the SES indicators, suggesting that the final sample is slightly more advantaged than the original sample. The final sample has slightly higher spousal education and income, older age of mother at first birth, more maternal work hours, and is more likely to have Caucasian ancestry. In all cases, the differences were very small, only in the first decimal place. All data were weighted in our analyses to represent the national population of the appropriate ages, and converted to person years for the multivariate analysis of leaving home.

Measures

Routes Out of the Home

Our outcome measures focus on the type of residence when first observed to be living away from the parental home. Files were checked at each interview to establish whether young adults resided with parents or parental figures (e.g., aunts, uncles, and grandparents) or in some other arrangement. If they were not living with parental figures, they were coded as having left home between surveys. Their living arrangements at the following survey were divided into four categories: (1) living with a partner and first child; (2) living with a partner but no children; (3) living autonomously, having become a parent, either with the child (if female) or separate from the child (if male), which we call “unmarried parental independence”; and (4) living autonomously, in a separate dwelling or group quarters without having become a parent.

Few in this last group had left to live in a college dormitory (one of a long list of living arrangements, including a vast array of relatives and nonrelatives). This is not so unusual, given the negatively selected nature of this sample. Further, using data from the Panel Study of Income Dynamics, Mulder and Clark (2002) showed that fully two-thirds of those who attended college remained home throughout, and many of the rest who did leave did so later in their college experience.

Finally, we note that almost no single coresidential fathers or absent mothers were observed in this file immediately after leaving home, hence they were omitted from the analyses. (Such families apparently evolve later.) Given the strong normative pressure on young adults to leave home early in adulthood (Billari and Liefbroer 2007), we assume that young adults are leaving voluntarily. However, it is always possible that parents put pressure on them to leave before they were ready.

Given that the surveys were spaced 2 years apart, we undoubtedly are missing short-term departures and returns. Therefore, we do not refer to these departures as “first leaving,” although, because of their age, it is highly likely that this is the first time that the young adults have left home. Studies have varied greatly in the length of interval employed to examine leaving home as a result of different survey reinterview schedules, including 4 months, 1 year, and 2–3 years. When the data points are closely spaced, studies of changing living arrangements in young adulthood find much churning (Goldscheider et al. 1993). The probability of returning home, of course, varies greatly by these durations. This 2-year interval also obscures short-term changes after leaving home, as when they separate from the partner they moved out to join, or gain a partner after having left by themselves. Unfortunately, we are unable to observe these changes.

Key Independent Variable

The key independent variable for this analysis is whether the young adult had recently become a parent for the first time. The 2006 Young Adult fertility and relationship data were used to obtain young adults' ages at the birth of their first child. Each wave within 2 years of that age was checked to establish whether young adults resided with their first child at the time of birth and whether they then were married or cohabiting. Because births occurred between survey waves, parental status was determined at the first wave following childbirth, beginning when they were age 14. An indicator for having had a first birth was coded for each wave.

The NLSY79 was unusually successful in obtaining reliable information on young men's fertility (Joyner et al. 2012), and it is likely that the Young Adult Survey, which followed the same protocols, was similarly successful. Although the high mobility of young people in their early twenties makes them more difficult to find and contributes to underreporting of births, this was not a problem for those who have not yet left home who are the major focus of our analysis. Response rates in the Young Adult Study were about 80 % in 2010 (U.S. Bureau of Labor Statistics 2013). Hence, it seems likely that reporting should be particularly reliable. We assume that if no birth were reported, no birth occurred.

Control Variables

In order to determine whether the effect of becoming a parent on leaving home was independent of other background factors, we controlled for socioeconomic conditions when growing up, parent–child closeness, family structure, race/ethnicity, demographic characteristics (urban exposure, age), as well as behaviors in adolescence and early adulthood. For socioeconomic conditions when growing up, we computed average values of the income and education of the residential father, and the mother's work hours over all child observations between birth and age 14. This provides a long term and stable summary of early socioeconomic resources.

The education of the spouse/partner of the mother was averaged over the years the young adult was 0–14. Similarly, the average income of the spouse or partner was obtained across all years the child was 0–14. Mothers' work hours indicate her contribution to the family's economic well-being across all ages 0–14. Average annual maternal work hours for all years in which the young adult was 0–14 years of age were divided by 50 in order to estimate average weekly hours. We included the age of the young adult's mother when she had her first child as an indicator of other unmeasured dimensions of parental socioeconomic status; it was reported by the mother and recorded in the NLSY data.

Information on closeness to parents was collected from children aged 10–14 in a self-administered questionnaire. If information was available for more than 1 year from these biannual surveys, we took the more recent. Scales of parental closeness were created using the sum of three items: How close do you feel to your parent (separately for mother/biological father/stepfather)? (1 = not very to 4 = extremely); how well do you share ideas and talk about important things with that parent? (1 = not very well to 4 = extremely well); and how often does the parent miss important events and activities? (1 = a lot to 3 = almost never). The residential father was used if the respondent reported on multiple fathers. The score for father closeness was calculated as the difference between the mother's and the father's score to reduce collinearity between the measures of parental relationships. Because children were closer to mothers than to fathers, the larger the value the less close the child was to the father.

Family structure was measured by the number of parental transitions (measured continuously). To calculate it, we examined living arrangements across all years from birth to age 14. Comparing household records across subsequent waves, we counted instances in which the biological father left the household and when a new father-figure entered the household. These transitions were summed. For those with no transitions, we determined whether the biological father was in the household for all years when the young adult was 0–14, referred to as father always there, and those in which the biological father was never in the household over the 0–14 period and no stepfather was ever in the household, referred to as father never there. For race/ethnicity, we distinguished Hispanics and nonHispanic Blacks, compared with nonHispanic Whites and others, and for residence, we used the proportion of years lived in urban areas between ages 0 and 14.

Our measures of adolescent behaviors include sexual experience and delinquent acts. Data come both from the self-administered questionnaires completed when they were 10–14 and from their interviews as young adults. Young adults were asked the age when they first had sex. If the response indicated that they first had sex before age 15, a dummy variable (1, 0) was included. The delinquent acts scale is based on the sum of nine items asked of 10–14 year olds in the self-administered questionnaire, including such items as: “stayed out later than parents said,” “hurt someone bad enough to need doctor,” “lied to parents about something important,” and “took something without paying.” Item responses ranged from 0 = never to 3 = more than twice. We think it likely that delinquency is underreported, but we have no way to assess bias in underreporting. To make the data more representative of adolescence, we selected the latest year in which the youth participated. For young adult behaviors, we also created a set of variables for whether they were enrolled in school and whether they were employed each year.

Methods

This analysis of leaving home uses descriptive tabulations, life tables, and multivariate discrete-time event history analysis. The life table analysis is based upon the leaving home experience of individual young men and women and used the actuarial approach. This life table was calculated for each gender and then stratified by parental status within gender.

Our event history analysis file of leaving home consisted of a separate observation for each year a young person was present in the NLSY young adult study and had not yet left home, beginning at age 14. Once measured as having left home, or 2 years after the birth of the first child if still at home, young adults are no longer considered eligible for a transition out of the home linked to a recent birth. At that point they are no longer observed and their information no longer contributes to the analysis. Thus, young men and women have as many observations as the number of years living at home either with no child or a very recent birth at the beginning of the year, 14,345 person years for men and 14,174 for women. Robust standard errors were obtained in the software package, STATA, to adjust for clustering within families and across years.

Analysis Plan

After examining the descriptive tabulations and the life table pattern of transition to separate residence by parental status, we move to multivariate analyses of the determinants of the transition out of the home. Using multinomial logistic regression, we first analyze the effect of becoming a parent on exit from the parental home for men and for women, distinguishing whether their new arrangement included a partner. Second, using multinomial logistic regression, we examine whether the young adult was living with a partner and/or a child when first observed living independently. In both cases the comparison group consists of those who did not leave home. We also analyze differences by gender and by age at transition.

Results

Descriptive Analyses

Variable and Subpopulation Descriptions

The weighted means and standard deviations of the independent variables based on the person year file are presented for the total, and separately by whether or not the young adult had had a recent birth (Table 1). Considering the total, first, we see that only 6 % of the person years involved a recent birth; whereas 4 % involved leaving home. Their background characteristics are consistent with what would be expected among the families of mothers who averaged age 21 at first birth, about 2 years younger than other women during that period (Mathews and Hamilton 2009). Respondents are from generally less affluent families (fathers' educations averaged 12.6 years and their incomes averaged $34,500). The young adults had averaged more than one transition in childhood family structure (1.26) and relationships were warmer with mothers than with fathers (a difference of 1.98), indicating that the average value for fathers was 7.04 compared with 9.02 for mothers. Although the young adult file averaged only 17 years of age, less than half of the person years living with their parents were spent enrolled in school (44 %); 24 % of their person years were spent in employment. The average level on the delinquency scale was 3.6 out of a possible 27.

Table 1. Means and standard deviations of the independent variables (N = 28,519 person years, total; 26,161, no recent birth; 2,358, recent birth).

Measures	Total		No recent birth		Recent birth
	Mean	Std	Mean	Std	Mean	Std
Leave home between surveys	0.04
Recent birth (within the past two years)	0.06	0.24	0.00	na	1.00	na
Family resources
Mother's spouse education (years)	12.56	2.21	12.64	2.21	11.35	1.81
Mother's spouse income (10,000 s)	3.45	2.52	3.52	2.57	2.42	1.45
Mother's weekly work hours, birth to age 14	21.12	13.60	21.33	13.59	17.96	13.28
Mother's age at first birth	21.11	3.50	21.27	3.50	18.64	2.48
Parental relationships
Maternal closeness scale (range 3–11)	9.02	1.73	9.05	1.72	8.65	1.93
Difference between parents in closeness scales	1.98	2.80	1.93	2.77	2.68	3.16
Controls
Number of parental transitions, birth to age 14	1.26	1.82	1.22	1.80	1.83	1.96
Father never there, birth to age 14	0.05	0.21	0.05	0.21	0.10	0.29
Percent of years in urban areas, birth to age 14	0.74	0.37	0.74	0.37	0.75	0.37
Hispanic	0.08	0.27	0.08	0.27	0.12	0.32
Black	0.17	0.37	0.16	0.36	0.30	0.46
Age of the young adult	17.19	2.64	17.05	2.59	19.37	2.36
First sex < age 15	0.16	0.37	0.14	0.35	0.41	0.49
Employed at Time 1	0.24	0.43	0.22	0.42	0.53	0.50
Enrolled in school or college at time 1	0.44	0.49	0.46	0.50	0.26	0.44
Male	0.52	0.50	0.54	0.50	0.36	0.48

Men and women with no recent birth closely resemble the total, given that only 6 % of the total person years were of those who had experienced a recent birth. Those who had a recent birth, however, were distinctive on a number of dimensions. They were from even more disadvantaged families (their fathers had considerably less education and income) than the sample as a whole. They had experienced 50 % more transitions in family structure (1.8 vs. 1.2), and twice as many had grown up in a family in which a father was never present (10 vs. 5 %). Their mothers had been much younger at their first birth (18.6 vs. 21.2) and they were much more likely to be members of minority groups (30 vs. 16 % were Black; 12 vs. 8 % were Hispanic). As would be expected, they were considerably less likely to be male.

Those individual person years with a recent birth were of young adults who were much more likely to have had sex before age 15 than others (41 vs. 14 %). They had somewhat less warm relationships with their mothers (8.6 vs. 9.0 on the closeness scale), and a larger gap in warmth between their relationships with their fathers and mothers (2.7 vs. 1.9). Partly reflecting the timing of first births in this person year file, youth with a recent first birth were more than 2 years older than those who had not had a first birth. They were much more likely to be employed (53 vs. 22 %), and less likely to be enrolled in school (26 vs. 46 %).

Life Table Analysis

A substantial proportion of young adults made the transition out of the parental home over the ages we observed, more among women and more among those who had had a baby than among men or the childless (Fig. 1). Few in any category had left home by age 18, about a quarter had left home among those observed at age 21, and about 60 % had left by age 24. About half the childless men had left home by age 24 compared with 68 % of young mothers. By age 26, whereas 80 % of those who had become parents (both males and females), and 80 % of childless women, were living independently of their parents, only 68 % of childless men were doing so.

Fig. 1. Proportion of young adults leaving home by gender and whether had a recent birth.

Across these ages, those with a recent birth left home more rapidly, although this difference is greater for females than for males, at least through age 21. At older ages, the effects of parenthood increase for males and attenuate for females. We will test the significance of this non-proportionality in the multivariate analyses.

Describing Routes

Those who left home took a wide variety of routes away from their parents. In Table 2, we distinguish the four pathways taken (with and without a partner; with and without a child), for the total and separately for males and females. The dominant route out of the home for these young adults is to autonomous (nonfamily) living, with about half overall (51.5 %) leaving without having become a parent or taken a partner. This is more common for males than for females, however (59 vs. 45 %).

Table 2. Percent leaving the parental home by route and gender, young adult offspring of the NLSY79 (N = 1125, total; 528 males; 597 females).

Route out	Total (%)	Males (%)	Females (%)
To live with a partner and child(ren)	19.6	20.5	18.9
To live with a partner but no child(ren)	18.0	17.4	18.6
Forming a female-headed family (living with child if female, separately if male)	10.8	3.2	17.6
To autonomous living (neither partner nor child)	51.5	58.9	44.9
Total	100.0	100.0	100.0

Most of the rest left home to form partnerships, which were predominantly cohabiting (results not shown), with few gender differences. About a fifth left home with a partner and children (20 % of males and 19 % of females) and slightly fewer left with a partner but without children (17 % of males and 19 % of females). The largest gender difference is in leaving in conjunction with the formation of a female-headed family. More than one in six female nest-leavers left to form a female-headed family, whereas only 3 % of males left home in conjunction with becoming fathers of children they were not living with (absent fathers). This difference likely reflects women's younger age at parenthood and perhaps a weaker connection between becoming a parent and leaving home among young unpartnered men than otherwise comparable women. It may also result from men's greater likelihood of underreporting births of own children, particularly if they do not live with them (Joyner et al. 2012). Using the NLSY79 data, the study by Joyner and colleagues, however, found evidence of only a small amount of unreported paternity, which was concentrated in the early 20 s rather than in the teenage years. Gender is clearly a factor in an analysis of the relationship between parenthood and leaving home, and in the multivariate analyses, we will distinguish factors that affect women from those that affect men.

Multivariate Analyses of Leaving Home and Routes Taken

We begin our multivariate analysis of leaving home with models that include having a recent birth as an independent variable (Table 3), testing Hypothesis 1. We examine the determinants of leaving home and, consistent with earlier research, the determinants of leaving home with a partner and without, compared to those who do not leave home (the omitted category). We provide separate results for males and females, and test for gender differences. Then we examine the factors that are linked with whether young adults leave home in conjunction with forming a family that includes their first child.

Table 3. Factors predicting leaving home, by gender (odds ratios) (N = 28,519 person years total; 14,345 males; 14,174 females).

Predictors	Model I			Model II
	Leave vs. remain home			Leave to live with a partner vs. remain home			Leave to live without a partner vs. remain home
	Total	Males	Females	Total	Males	Females	Total	Males	Females
Recent birth	3.71**	4.09***	3.47***	8.93**	14.15**	6.30**	1.56**	0.65	2.10**
Controls
Family resources
Mother's spouse education	1.07**	1.07*	1.08*	1.03	1.03	1.03	1.09**	1.08	1.10*
Mother's spouse income	1.00	1.03	0.96	0.91*	0.98	0.85**	1.03	1.04	1.01
Mother's work hours	1.00	1.00	1.00	1.00	1.00	1.00	0.99	1.00	1.00
Mother's age at first birth	1.02	1.06**	0.99	0.97	1.00	0.95	1.05*	1.09**	1.01
Parental relationships
Maternal closeness	0.94*	0.96	0.92*	0.97	0.95	0.97	0.92*	0.96	0.89**
Parental closeness difference	1.06**	1.06	1.05	1.01	1.03	0.99	1.09**	1.07	1.10**
Other controls
Number of transitions	1.09**	1.09	1.11*	1.09	1.11	1.07	1.05**	1.07*	1.05**
Father never there	1.12	1.24	1.07	1.12	1.12	1.12	1.12	1.12	1.1
Urban	0.93	0.92	0.94	0.83	1.02	0.74	1.02	0.88	1.12
Hispanic	0.64**	0.68*	0.59**	0.56**	0.54*	0.54*	0.68**	0.78	0.60*
Black	0.45**	0.51***	0.39***	0.25**	0.34**	0.17**	0.63**	0.68	0.58**
Age of the young adult	1.45*	1.44***	1.48***	1.43**	1.40**	1.48**	1.47**	1.46**	1.50**
First sex < age 15	1.13	0.91	1.36	1.27	0.99	1.48	1.03	0.82	1.30
Delinquent activities	1.01	1.01	1.01	1.02	1.02	1.01	1.01	1.01	1.02
Employed	3.62**	4.01***	3.16***	4.69**	6.33**	3.64**	3.18**	3.25**	2.92**
Enrolled in school or college	0.67*	0.56*	0.72	0.51	0.36	1.00	0.71	0.58	0.83
Male	0.75**			0.71*	1.00	1.00	0.76*	1.00	1.00

Bold: Significant difference between male and female coefficients (p < .05). Underline: Significant difference between leaving with and without a partner (p < .05). Italics: Significant difference between males and females re leaving with and without a partner (p < .05)

^**p < .01;

^*p < .05, 2-tailed test

The Effects of Parenthood

In this first multivariate analysis (Table 3, Model I, columns 1–3), we see that parents left home much more rapidly than nonparents. Consistent with the basic expectation of Hypothesis 1, having a baby increased the odds of leaving home by three to four times. Contrary to our expectation, however, the difference between males and females was not significant; gender differences only appeared when we distinguished partner status. The largest effects of becoming a recent parent on leaving home appear when a partner is involved; effects on leaving to live with a partner were significantly greater than for leaving to live without a partner. The large positive effects of parenthood on leaving to live with a partner were large for both young men and women, but the effects for young men were significantly larger than for young women. The significant effect on leaving home without a partner, however, only characterized young women; parenthood had no impact on young men's leaving home unless he entered a union as well.

We also tested the age dimension in Hypothesis 3, that parenthood would have a stronger effect on leaving home among those who became parents at the older end of the age range (20–26) than among those at the younger end (15–19), a pattern that appeared in the life table analysis. Once all the control variables were included, however, the interaction between age and parenthood was not significant, and none of the other coefficients changed in any substantial way (results not presented). This may have resulted from the small number of births among those in their mid- to late-teens, and the small number of person years at the older ages in this sample. Hence, although there are clear differences in the effects of parenthood on leaving home by gender, there is much less evidence of differences by age.

The effects of parental resources are complex. Paternal education was associated with a greater chance of leaving home for both males and females, with nearly all of its effect on leaving to live without a partner; each additional year of the father-figure's education increased those odds by 8–10 %. Greater paternal income had no effect on the group as a whole, but (nonsignificantly) retarded young women's departure and (nonsignificantly) and accelerated that of young men, leading to a significant gender difference. The negative effect of income for women was due to its effect on delaying leaving home with a partner, which was statistically significant. A higher maternal age at first birth was associated with an increased chance of leaving home for men, but had no effect for women, with a significant gender difference. Mothers' work hours had no impact on overall leaving home for either sex or by any route.

In sum, all of the effect of family resources for men worked to accelerate leaving home without a partner. For women, parental resources increased leaving without a partner and retarded living with a partner compared with remaining in the parental home. Having a warmer relationship with one's mother retarded leaving home, although the effect was only significant for women and, among women, only for leaving home without a partner. The difference in warmth between the parents, indicating lower paternal warmth, accelerated leaving home for both sexes; as with maternal warmth, all of the effect of paternal warmth were on leaving home without a partner. In results not presented, we found that the quality of relationships with parents had a much more powerful effect on leaving home during the years prior to age 20 than it did thereafter. This was the only substantial difference by age that emerged when we considered whether the correlates of leaving at very young ages differed from those at later ages.

Consistent with previous studies, a greater number of transitions when growing up sped up leaving home. Growing up in an urban area has no effect. Members of minority groups, particularly Blacks, were much less likely than the White majority to leave home, whether with or without a partner. Not surprisingly, older respondents were more likely to leave home than younger respondents.

Early sexual experience had no effect on leaving home, by either route, for either men or women (controlling for whether it leads to parenthood or not). Own personal resources accelerated leaving. Being employed increased nestleaving for both sexes, making nestleaving three to four times more likely for both males and females, with strong and significant effects for both routes. Staying in school delayed leaving home, although this was only significant for men.

Parental Routes Out of the Home

Now that it is clear that having a recent birth increases leaving home overall, we can address the predictors of leaving via routes that differ by whether the young adult had a recent birth. For this analysis, we subdivided the two major routes out of the home (to live with a partner and to live with no partner) by whether a child is involved, giving us four routes to residential independence from parents: to live with a partner and a new child; to live with a partner without a child; to live with no partner, having had a new child (coresident for women but not men); and to live without a partner and without a child. The omitted category is comprised of young adults who continue to live at home with parents. Hence, the recent birth becomes part of the structure of routes out of the home, rather than a correlate of leaving home. These results are presented in Table 4. We show only the effects of family resources and parental relationships, as there were differences in these results between Tables 3 and 4; the other controls were also included but are not shown as there were no substantive differences between the two tables (results available on request). We present results for men and women separately (testing for gender differences). We also include tests for factors that link with the likelihood of leaving with a partner differentially depending on whether a child is involved or not, and test as well for factors that are differentially linked with the likelihood of leaving with no partner, depending on whether a recent child is involved or not. These tests are our major focus. We also include indicators for tests of leaving home by a given route relative to remaining in the parental home; we will note the interesting findings.

Table 4. Factors predicting first leaving home, by gender and first post-parental living arrangements (odds ratios) (N = 28,519 person years total; 14,345 males; 14,174 females).

Predictorsa	Males				Females
	Partner + kids	Partner, no kids	1-parent independence (absent fatherhood)	Nonfamily independence	Partner + kids	Partner, no kids	1-parent independence	Nonfamily independence
	vs. remain home				vs. remain home
Family resources
Mother's spouse education	0.98	1.01	1.44	1.08	0.95	1.01	0.90	1.15**
Mother's spouse income	0.88	0.95	0.89	1.04	0.70**	0.92	0.84	1.02
Mother's work hours	1.00	1.00	0.96	1.00	0.99	1.00	0.96**	1.01
Mother's age at first birth	0.85**	0.94	0.72*	1.11**	0.91	0.94	0.86**	1.04#
Parental relationships
Maternal closeness	0.87	1.02	0.72	0.97	0.97	0.97	0.92	0.87**##
Parental closeness difference	1.06	1.00	1.14	1.07	0.98	1.01	1.10	1.10**

Bold means that the two coefficients within a given partner category are significantly different from each other (p < .05)

^aModels also include the same control variables as in Table 3

^**p <.01;

^*p < .05, 2-tailed test (against remaining home)

^##p < .01;

^#p < .05, 2-tailed test (against effect for males)

The omission of recent parenthood from the factors predicting overall leaving home did not alter in a major way the results for covariates compared with those in Table 3 (results not presented). The effects of parental resources were smaller when parenthood was no longer in the model, indicating that part of the effects of parental resources on leaving home operated through their impact on reducing early parenthood, and, hence, that impetus to leave. There were also not a large number of differences between routes involving children and those that did not; for many purposes it would be appropriate to combine them. Nevertheless, it is clear that in order to understand the role of recent parenthood on routes taken out of the home, it is important to examine the parental route directly.

Parental socioeconomic background clearly shaped those routes out of the home that involved parenthood, with more effects for young women than for young men. Overall, greater resources led to greater nonfamily independence (all odds ratios are above 1.0) and to less leaving in conjunction with becoming a parent (odds ratios either very close to 1.0 or well below). Higher parental education significantly increased the likelihood young women left home via nonfamily independence relative to remaining home (as indicated by the conventional asterisks). Young women with more educated parents who left home without a partner were significantly more likely to do so without children than with children, indicated by the two bold odds ratios (1.15 vs. .90).

Parental earnings, in contrast, have their major effect on distinguishing between young women who left with a partner and children and those who established a partnership without children. An additional ten thousand dollars a year of the resident father's earnings were associated with a reduction in both likelihoods, but the effect was significantly stronger for having a child early in (or even before) the partnership's establishment, reducing the odds of leaving with a partner and child by fully 30 %, far more than the 8 % reduced odds for a childless partnership. (See the bold odds ratios of .70 and .92.) And each additional hour mothers worked per week during the childhoods of these young adults was associated with a reduction in the likelihood young adults left in conjunction with unpartnered parenthood (OR = .96), although, despite identical odds ratios, only the result for young women was significant.

The age of the mother at first birth shaped the pathways out of the home for both males and females. The older his or her mother was at first birth, the more likely the youth was to leave to nonfamily independence, relative to remaining home (only significant for males), and the less likely to leave via any family-related route (significant for both males and females). Most of the effects were in the same direction for men and women.

Hence, parental resources contributed to young adults' leaving home, but much less so to leaving by routes involving children. The resources of both fathers (education and earnings) and mothers (work hours and the age they started childbearing) made a contribution. The associations were stronger for females than for males.

The effects of closeness to parents are less dramatic in this table than in Table 3, and the results were only significant for females. Having a close relationship with the mother reduced the transition to nonfamily independence, an effect significantly greater for young women than for young men. Similarly, having a less close relationship with the father accelerated leaving home to nonfamily independence for both men and women.

Discussion

This analysis has identified factors linked to departure from the parental home for young men and young women, focusing in particular on the role of parenthood in the process. We model the transition to parenthood both as a factor linked with leaving home and as one of the routes out of the parental home. In our analyses treating parenthood as a determinant of leaving home, we distinguished leaving with and without a partner. In our analyses treating parenthood as a route out of the home, we distinguished four route combinations, varying by whether a recent birth or a partner was involved.

Support for our hypotheses was quite strong. The results were consistent with our theoretical approach, based both on the life course perspective, which focuses on sequences of transitions early in the life course, and on rational choice theories, which focus on the effects of resources in the context of strong preferences for family privacy. As we expected in Hypothesis 1, becoming a parent increased the likelihood both young men and young women left home; contrary to our expectations in Hypotheses 2 and 3, there was little difference in its effects on leaving home between men and women or between younger and older young adults. The change that occurred between the pattern evident in the 1970s, favoring men (Goldscheider and DaVanzo 1989), and that in the 1980s favoring women (Avery et al. 1992) not only did not intensify, but actually reversed, so that gender made no difference. Having recently become a parent increased leaving home for both young men and young women.

There were gender differences, however, when we distinguished whether the young adult left home to form a partnership. Men who reported a recent birth were more likely to leave home with a partner than were otherwise comparable young women, and young women with a recent birth were more likely to leave home without a partner than young men. Perhaps young men who become a parent while living with their parents have to balance their need for independence from the parental home linked to having made the transition to parenthood with the increased financial burden many of them have taken on. For young women, in contrast, an important route out of the home was to live separately from parents, but with a child. This may reflect a continued difference in expectations for males and females, with having a baby being a true transition to independence for a young woman but not for a young man. Previous research has suggested that living with one's own mother is not a stable situation for a young mother because of conflicts over childrearing (Furstenberg and Crawford 1978).

As with our expectation of gender differences, our expectation from life course theory that the effects of becoming a parent might differ depending on age at the event was not supported. There were almost no significant differences by age; those who became parents in their middle/later teens were as likely to leave home as those who waited until their early twenties. The only age difference that appeared was in the effect of parental relationship quality, suggesting that parents have more impact on the decisions of their teenage children than on their older children. Finding no difference in the effect of recent parenthood by age is particularly surprising, as one of the aims of the passage of TANF was to reduce adolescent parents' ability to maintain a separate residence. The risk of becoming a parent as a teenager is considerably less than among older young adults, but among those who do, parental support is associated with their living arrangement decisions and, thus, of policy relevance.

We examined a wide range of parental resources, including education and income, maternal work hours, and mothers' age at first birth. Parental resources had a strong impact on leaving home and on whether the route taken involved parenthood. However, the results depended on the type of resources, gender, and whether a partner was involved. The effect of parental earnings was primarily to reduce leaving home at these relatively young ages. Consistent with the findings of Iacovou (2010), they particularly reduced leaving home among those who had a recent birth without establishing a partnership, and hence in conjunction with the formation of single-parent families. Perhaps these additional resources not only made young people less likely to become unpartnered parents at these young ages, but made them rationally choose to remain close to parental resources by living at home. Maternal work hours operated somewhat like parental income, reducing the likelihood of young adults' forming a new residence as a single-parent family.

Consistent with the findings of Barber (2001), mothers' age at first birth had very powerful associations, with the children of women who began childbearing early more likely to leave home with children, either with a partner or in conjunction with the formation of a single-parent family. In contrast, greater parental education was associated with increased leaving home for both young men and young women, but only via nonfamily routes. Greater parental education transmits more information and access to opportunities other than family formation, such as furthering one's education and starting a career, with which childbearing would interfere.

These findings underline the ways parental resources or lack thereof lead to the reproduction of privilege and poverty. They thus reinforce the findings of studies looking at growing inequality in other dimensions of the transition to adulthood, such as teenage parenthood (Furstenberg 2007; McLanahan 2004) and unmarried parenthood (Edin and Kefalas 2011; McLanahan 2004), as well as socioeconomic well-being (Danziger and Rouse 2007).

Economic resources are not the only important resources for children. The warmth of parental relationships is also important, and implicated in the growing inequality in access to fathers, as a positive relationship reduced the likelihood that young adults left home without a partner. This was the case both for leaving after a recent birth or transitioning to nonfamily living, consistent with the findings of several studies (e.g., Turley et al. 2010; Seiffge-Krenke 2006). Unlike the transition to early parenthood (Hofferth and Goldscheider 2010), both parents' relationships had quite similar effects. The results suggest that parental relationships serve as important resources for young parents, so that the growth in father absence contributes to inequality. These findings also reinforce the parent–child axis of the family, relative to the weakening couple axis (Swartz 2009), and perhaps the growing importance of parental resources in young adults' outcomes (Guldi et al. 2007).

This study has several weaknesses. First, these young adults are still relatively young, and the relationships between parenthood and leaving home might change with age beyond the window we could observe. Their youth may be the reason we failed to find major differences in the factors influencing leaving home by age. It will be important to see whether these results hold up as more enter parenthood and/or leave home. Second, although we have used weights to correct this to the extent possible, these young adults, most particularly the oldest, are selectively the children of women who became mothers at an early age. Third, although the NLSY79 remains one of the best data sources for our research, the 2-year period of time between survey waves means that some transitions may have gone unreported.

Our analyses did not focus in any detail on race/ethnic differences. American families are increasingly diverse, as the Black–White divide has been altered by new arrivals from Asia and Latin America. Although we find predicted differences across these groups in the routes out of the parental home, a full examination of the ways family resources, parental relationships and recent parenthood interact to influence leaving home is needed.

There are also other economic factors likely to influence the decision to leave home. Recent research on Great Britain has shown that student debt reduces residential independence (Andrew 2010) and that changes in welfare policies have increased the likelihood of coresidence between parents and their adult children (Berrington et al. 2013). In the United States, the “great recession” appears to have greatly increased economic and personal instability for young adults (Qian 2012).

Nevertheless, this study has many strengths. These include the use of data on young adults' parental resources collected prospectively from multiple informants over their entire lifetimes; measurement of relationships with both mothers and fathers; the focus on a recent and policy-relevant group of young adults; and the ability to qualify the living arrangement of the young adult by the presence of partner and child. In the future, examining changes in these transitions in the context of economic conditions would be an important addition to the literature.

Based on these results, we draw the following conclusions. First, although delays in marriage and parenthood are undoubtedly delaying leaving home among many young adults, the growth in unmarried parenthood among some has spurred leaving home, at least for disadvantaged young women, even in the twenty-first century. Further, the growth in socioeconomic inequality is not only contributing to inequalities in the timing of parenthood (Martin 2000), it is likely contributing to inequalities in the timing and sequencing of leaving home, particularly to fragile family formation. Finally, in this era of multiple routes out of the parental home, the better the quality of relationships between parents (both mothers and fathers) and their children emerging into adulthood, the lower the likelihood that young adults leave home early in adulthood.

This paper is the first to elaborate how young parenthood relates to the home leaving process, and brings an examination of the increasing role of parental resources and relationships with both mothers and fathers on the nestleaving process to a new cohort. It documents the role that parental resources play in the lives of young men and women in moving to independent or to partnered living, and points to the continued influence of socioeconomic disadvantage on the transition to adulthood.