Men’s and Women’s Pathways to Adulthood and Their Adolescent Precursors

Abstract

This study compared men’s and women’s pathways to adulthood by examining how role transitions in education, work, marriage, and parenthood intersect and form developmental pathways from ages 18–30. The study investigated how sociodemographic factors and adolescent experiences were associated with these pathways. We used latent class analysis to analyze longitudinal data from a gender-balanced panel of 808 contemporary young adults. We found three similar latent pathways for both genders, but men and women differed in the timing of marriage and when they began to live with children and the likelihood of combining both roles. The present study points to the continued, though differential, relevance of marriage and family in the transition to adulthood for men and women.

Consideration of the diverse ways young people move from adolescence to adulthood is important because different pathways have potentially important implications for functioning and quality of life later in adulthood (Macmillan & Eliason, 2003; Shanahan, 2000). Studies of the transition to adulthood have described the changing character of this stage in the life course, finding that it has become more diverse, individualized, and destandardized in Western societies over the past decades (Buchmann, 1989; Elzinga & Liefbroer, 2007; Shanahan, 2000). Despite evidence for greater variability in the timing and sequencing of transitions into adult roles and responsibilities, recent empirical investigations have identified a limited number of distinct transition pathways differentiated primarily by the timing of family formation and participation in postsecondary education (Macmillan & Copher, 2005; Osgood, Ruth, Eccles, Jacobs, & Barber, 2005; Sandefur, Eggerling-Boeck, & Park, 2005). This line of research has suggested that, despite greater variety in transition sequences and weaker age norms that no longer clearly prescribe the timing of transitions into adult roles, there are only a few typical pathways that describe the transition to adulthood for the majority of young adults.

Despite generally greater gender equality in U.S. society, these pathways continue to differ by gender. Research on the life course indicates that men and women differ in the timing of transitions, particularly to family roles such as marriage and raising children, and in how they sequence and combine such roles (Elder, 1998; Moen, 2001). Today’s young women still tend to have children earlier and marry earlier than men. There are also growing concerns about young men’s ability to successfully move into adulthood (Kimmel, 2008). Yet knowledge about how pathways to adulthood vary by gender is limited. Furthermore, many studies of the transition to adulthood are dated, focusing on cohorts that transitioned into young adulthood during the 1980s or earlier (e.g., Macmillan & Eliason, 2003; Marini, 1984; Mouw, 2005; Rindfuss, Swicegood, & Rosenfeld, 1987). Comparatively little is known about the transition to adulthood of contemporary young adults.

To examine the pathways to adulthood of contemporary men and women, the present study analyzed data from a gender-balanced longitudinal panel of 808 young adults followed from age 10 in 1985 to age 30 in 2005. The study used a multidimensional approach by examining how transitions in education, work, and the family intersected and formed pathways to adulthood from age 18 to age 30, and how such pathways differed by gender.

Pathways to Adulthood

Completing school, moving into full-time employment, getting married, and becoming a parent are key transitions in young adulthood in Western cultures (Booth, Crouter, & Shanahan, 1999; Cohen, Kasen, Chen, Hartmark, & Gordon, 2003; Macmillan & Copher, 2005). The acquisition of new roles and statuses during the transition to adulthood represents a salient and normative developmental task expected to be completed during this life period (Neugarten, Moore, & Lowe, 1965; Roisman, Masten, Coatsworth, & Tellegen, 2004). Arnett (1998) has argued that these role transitions have lost their relevance in marking the entry into adulthood and that more individualistic and personal factors, such as a sense of independence, autonomy, and responsibility, define a period of emerging adulthood. Although this may be true for contemporary young adults who postpone family formation for a college education or career advancement, many young adults marry and raise children, and it is likely that moving into family roles indicates adult status to the self as well as others. In contrast to the domains of school and work, which are salient already in adolescence, most young people take on marriage and raising children for the first time during young adulthood, which makes such role transitions distinct indicators of the new adult status (Roisman et al., 2004). Recent studies have found that role transitions, particularly into a parenting role, continue to be important markers of young adulthood for the young adults themselves but that associated individualistic experiences are relevant as well (Johnson, Berg, & Sirotzki, 2007; Shanahan, Porfeli, & Mortimer, 2005).

Elder (1998) has conceptualized the life course as “a sequence of socially defined, age-graded events and roles that the individual enacts over time” (p. 941). This framework recognizes that transitions in different domains, such as education and family, are interdependent within and across time and form social pathways of linked developmental trajectories. The timing of one transition often has cascading consequences for other transitions (Masten et al., 2005). For example, postsecondary education often results in postponement of family formation (Mortimer, Oesterle, & Krüger, 2004; Rindfuss et al., 1987); very early parenting hinders not only completion of high school but also continuation in postsecondary education (Haggstrom, Kanouse, & Morrison, 1986; Upchurch, 1993); and teenage parenting often occurs outside of marriage, yet when marriage happens very early, it increases the probability of divorce and separation during the transition to adulthood (Macmillan & Eliason, 2003). Few studies have operationalized the complex developmental nature of the transition to adulthood consistent with this life course conceptualization. Many studies have focused on one transition at a time (e.g., the transition to parenthood; Cohen et al., 2003; Woodward, Fergusson, & Horwood, 2006) or individual transition sequences such as school to work (Cooksey & Rindfuss, 2001; Mortimer et al., 2004). By taking a multidimensional and person-centered perspective, the present study sought to examine the transition to adulthood more comprehensively and draw a more complex and accurate empirical picture of pathways to adulthood consistent with life course theory.

Studies that have used person-centered typological approaches, such as latent class analysis, cluster analysis, sequence analysis, and trajectory analyses, have suggested that participation in postsecondary education is a major dividing factor that distinguishes those who move quickly into forming their own families from those who postpone family formation, especially parenthood (Sandefur et al., 2005). Among those with limited postsecondary education, a second division exists between those who have children early, as teenagers and often outside of marriage, and those who have children later, beginning in their mid-20s. Recent studies based on national and regional samples found that one third to one half of young adults followed a pathway characterized by investment in postsecondary education and postponed parenthood, though some were cohabiting or married, and another one half to two thirds had limited postsecondary education with earlier family formation (Amato et al., 2008; Macmillan & Eliason, 2003; Osgood et al., 2005).

Gender Differences in Pathways to Adulthood

Because of the demographic changes in women’s lives over the past several decades (Fussell & Furstenberg, 2005), men’s and women’s work and educational pathways in the transition to adulthood have become more similar (Fussell & Furstenberg, 2005; Johnson, Oesterle, & Mortimer, 2001). Men’s and women’s life courses still differ, however, with respect to family roles such as marriage and parenthood (Moen, 2001; Williams & Umberson, 2004). It is surprising that few studies of multidimensional pathways to adulthood have examined gender differences. Some studies have restricted analyses to women only (Amato et al., 2008; Fussell & Gauthier, 2005; Macmillan & Copher, 2005). Others combined men and women in the analysis (Osgood et al., 2005; Ross, Schoon, Martin, & Sacker, 2009) or analyzed them separately but concluded that the findings were similar for men and women (Mouw, 2005; Sandefur et al., 2005). These studies have left an incomplete understanding of how contemporary young men’s and women’s transitions to adulthood differ. Examination of the results from Sandefur et al.’s (2005) study, for example, suggests an important gender difference in the link between marriage and parenthood during young adulthood. Among those with limited postsecondary education, unmarried young adult men were much less likely than unmarried young adult women to have children, which suggests that fertility is less tied to marriage for women than for men. Women not only give birth to but also increasingly raise children outside the context of marriage, much more so than men do (Seltzer, 2000). Overall, men marry and have children later than women (Cohen et al., 2003; Woodward et al., 2006), but are also much less likely than women to live with their children and to have the primary responsibility for raising their own or their partner’s children (Coltrane, 2000; Hochschild & Machung, 1989).

These gender differences in the timing and combining of marriage and parenting responsibilities are likely to have important implications for men’s and women’s pathways to adulthood. Specifically, we would expect that women marry and take on parenting responsibilities (as indicated by living with children) earlier than men and, consequently, are more likely to have moved into these adult family roles by age 30. We would also expect that more women than men live with children without being married at any point during young adulthood. At the same time, we would expect that women differ little from men in their participation in paid employment outside the home and involvement in postsecondary education. This implies that women are more likely than men to transition into multiple adult roles simultaneously, for example, combining school, work, and parenting, and that men are more likely to sequence transitions in different domains, for example, completing postsecondary education before getting married and living with children only after marriage. We examined these hypotheses by using a multidimensional analysis approach that takes into account that transitions in education, work, marriage, and the family intersect within and across time.

Sociodemographic and Adolescent Precursors

A person’s social location, experiences of early adversity, and adolescent experiences and behaviors greatly influences the timing of transitions into adult roles. Transitions to adult roles and statuses have been found to differ markedly by sociodemographic characteristics, including race and ethnicity, socioeconomic status and family structure in childhood, and adolescent experiences such as school performance and involvement in substance use and crime. African Americans tend to be on a track of early parenting more so than Whites and Asian Americans, but they tend to marry later (Macmillan & Copher, 2005; Schoen, Landale, Daniels, & Cheng, 2009). Young adults growing up in families of lower socioeconomic status, measured by parental education and family income, move more quickly toward financial independence; into committed romantic relationships, including marriage and parenthood; and are less likely to participate in postsecondary education than young adults from higher socioeconomic backgrounds (Guldi, Page, & Stevens, 2007; Osgood et al., 2005; Sandefur et al., 2005; Schoen et al., 2009). Students from more highly educated families are more likely to invest in postsecondary education in part because they do better in high school, but students’ own academic aspirations, expectations, and performance in high school are also independently associated with pathways characterized by higher educational achievement during young adulthood (Amato et al., 2008; Osgood et al., 2005; Ross et al., 2009). Other characteristics of the family of origin are related to how young people transition to adulthood, including experiencing early adversities and family disruptions during childhood and adolescence, such as parental divorce or death, and having been born to a teenage mother. Young adults who grew up in divorced families or stepfamilies are less likely to be involved in postsecondary education and more likely to start their own families at an earlier age (Ross et al., 2009; Wolfinger, 2003). Both daughters and sons of very young mothers are also likely to have their children early and often outside of marriage (Barber, 2001).

Drug use and delinquency in adolescence have been linked with precocious transitions to adult roles, including early and risky sexual behavior, teenage parenthood, and leaving high school early, as well as problems with the assumption of adult roles and less socioeconomic success, such as unemployment, single parenthood, lower educational attainment, and welfare receipt (Brook, Richter, Whiteman, & Cohen, 1999; Krohn, Lizotte, & Perez, 1997; Newcomb & Bentler, 1988). Osgood et al. (2005) found that illegal behavior during the young adult years, including illicit drug use and crime, was most prevalent among single young adults and “slow starters,” a group of young adults who were least likely to have transitioned into adult roles by age 24. Illegal behavior was least prevalent in groups that had already taken on family roles such as marriage and parenting. To account for selection into pathways to adulthood, the present study examined sociodemographic indicators, experiences of early adversity in the family, adolescent academic experiences, and illicit behaviors in adolescence as predictors of multidimensional pathways to adulthood.

Method

Data

The present study used prospective data from the Seattle Social Development Project, a longitudinal panel study of the development of prosocial and antisocial behaviors. In 1985, 18 Seattle elementary schools that served students from high-crime neighborhoods were identified. During this study, the Seattle School District used mandatory busing to achieve racial balance in schools. Thus, all schools in the study served a heterogeneous population of students drawn from at least two different neighborhoods. The study population included all fifth graders in these schools (N = 1,053). A total of 808 students (77% of the identified population) and their families agreed to participate in the study.

The panel was interviewed in 12 waves (annually during school Grades 5–10, in Grade 12, and every 3 years thereafter) from 1985 through 2005, when most subjects were 30 years old (standard deviation [SD] = .52). One parent of the respondents was also interviewed in six annual waves when panel members were in Grades 5–10. Student questionnaires were group administered in school in Grades 5 and 6. Parents and, in later years, panel members were interviewed individually and in person. Respondents who moved out of state (about 25% by 2005) were tracked and interviewed. Retention rates for the sample have remained greater than 91% since 1989, when panel members were 14 years old. The sample included about equal numbers of men (n = 412) and women (n = 396) and was ethnically diverse: 47% White, 26% African American, 22% Asian American, and 5% Native American. Of those groups, 5% were Hispanic. A substantial proportion of the participants grew up in low-income families. Forty-six percent of participants’ parents reported a maximum family income of less than $20,000 per year in 1986. About 52% of panel members participated in the National School Lunch/School Breakfast Program between the ages of 10 and 12.

The study included a universal preventive intervention aimed to reduce youths’ health risk behaviors. The multicomponent intervention consisted of teacher training, parent education, and social competence training for children (Hawkins, Catalano, Kosterman, Abbott, & Hill, 1999). In 1981, students and teachers were randomly assigned to Grade 1 classrooms, and Grade 1 classrooms were randomly assigned to receive the intervention or serve as controls in eight Seattle public schools. Students in intervention and those in control classrooms were followed up prospectively until 1985. When these students entered Grade 5 in the fall of 1985, the study was expanded to include Grade 5 students in 10 additional schools. New schools added for the panel study were matched to the original eight schools with respect to grades served and inclusion of students drawn from high-crime neighborhoods. Schools added for the panel study were nonrandomly assigned to conditions to achieve balanced numbers across conditions. From that point on, all Grade 5 students in each of the 18 schools received the intervention according to their school’s intervention assignment. On the basis of this design, a nonrandomized controlled trial with four conditions was created. The full intervention group (n = 156) received the intervention from Grade 1 through Grade 6. The late intervention group (n = 267) received the intervention only in Grades 5 and 6, and the control group (n = 220) received no special intervention. A parent-training-only group (n = 141) was offered only the “Preparing for the Drug Free Years” curriculum during Grades 5 and 6. Twenty-four participants in the longitudinal study could not be classified into any of these groups because they left participating schools in Grade 5 before spending at least one semester there. Prior examinations of the data have shown that the intervention had significant long-term effects on some of the markers of the transition to adulthood, including a lower rate of early parenting, greater engagement in work and school, and decreased crime and drug use by age 21, as well as higher educational and economic attainment by age 27 (Hawkins, Kosterman, Catalano, Hill, & Abbott, 2005, 2008; Lonczak, Abbott, Hawkins, Kosterman, & Catalano, 2002). Although differences in prevalences and means have been observed between the intervention and control groups, prior analyses have shown few differences in the covariances among variables between the groups (Hill, Hawkins, Catalano, Abbott, & Guo, 2005; McCarty et al., 2009).

Measures

Pathways to Adulthood

To examine pathways to adulthood, the analysis used data collected at five time points when most panel members were 18, 21, 24, 27, and 30 years old (1993–2005). At each time point, participants’ school attendance, employment status, marital status, and whether they were living with children were assessed. Information from closed-ended interview questions about current status (e.g., “What is your current marital status?”) and information reported on a life-history calendar (LHC) (Axinn, Pearce, & Ghimire, 1999; Freedman, Thornton, Camburn, Alwin, & Young DeMarco, 1988) were combined (e.g., month and year of marriage). The LHC was administered at each interview and covered the 3 years since the prior data collection wave, reaching retrospectively from age 18 to age 30.

School attendance

At age 18, respondents were grouped into three mutually exclusive categories: (a) attended high school, (b) participated in a General Education Development (GED) program or attended community college, or (c) did not attend school. At ages 21–30, we assessed whether participants were attending any kind of school for 4 or more months during the previous 12 months (coded 1) or for fewer than 4 months (coded 0) using month-by-month data from the LHC. We used this cutoff to capture involvement in an educational institution that was more than occasional or a onetime enrollment in classes.

Employment status

Respondents indicated on the LHC for each month of the 12 months prior to each interview whether they were employed full-time (35 or more hours per week), employed part-time (fewer than 35 hours per week), unemployed, or a full-time homemaker (no paid employment outside the home). For each of the five time points, respondents were grouped into three mutually exclusive categories: (1) no paid employment for the entire year (including homemaking and unemployment), (2) employed full-time for 6 or more months (two or more part-time jobs in the same month were counted as working full-time for that month), or (3) employed part-time (i.e., held only part-time jobs or were employed full-time for fewer than 6 months). Calendar data on employment status were not available for age 18, yet we were able to create a comparable measure using closed-ended interview questions referring to the “regular job” over the previous year.

Marital status

On the basis of the year of first marriage as indicated on the LHC and in combination with responses to the question “What is your current marital status?” at each interview wave, we determined cumulatively for each of the five time points whether respondents were (a) never married, (b) married (including remarried and separated but still married), or (c) divorced.

Living with children

Because not all young adults who give birth to or father a child are regularly and actively involved in the child’s life, we measured the adoption of parenting responsibilities indicated by living with children rather than fertility as indicated by the birth of a biological child. We identified for each time point whether respondents lived with children (coded 1) or not (coded 0) at any point during the 12 months prior to the interview as indicated on the LHC, including biological, step, adopted, and foster children. Table 1 shows the categories of all variables and the distribution of men and women on each measure for each time point.

Table 1.

Percentage of Women (N = 387) and Men (N = 395) in Role Statuses, Ages 18–30

		Age 18		Age 21		Age 24		Age 27		Age 30
		Women	Men	Women	Men	Women	Men	Women	Men	Women	Men
School attendance	Not in school	8.0	4.2	—	—	—	—	—	—	—	—
(Age 18)	In high school	84.8	86.9	—	—	—	—	—	—	—	—
	GED/Com. collegea	7.2	8.9	—	—	—	—	—	—	—	—
School attendance	Not in school	—	—	53.1	63.2	74.3	76.0	80.0	80.1	86.5	84.8
(Age 21–30)	In school	—	—	46.9	36.8	25.7	24.0	20.0	19.9	13.5	15.2
Employment status	Not employed	33.7	35.7	13.6	9.4	15.7	10.2	21.0	11.3	20.4	13.5
	Employed part-time	59.2	54.3	36.6	29.2	13.6	9.5	11.7	12.4	9.2	7.9
	Employed full-time	7.1	10.0	49.7	61.4	70.8	80.3	67.4	76.3	70.4	78.5
Marital status	Never married	99.7	99.7	94.7	98.1	71.0	82.5	54.3	70.3	38.6	50.0
	Married	0.3	0.3	4.8	1.6	24.4	15.4	38.6	25.7	51.9	42.0
	Divorced	0.0	0.0	0.5	0.3	4.5	2.1	7.2	4.0	9.5	8.0
Living with children	No	80.9	94.5	57.0	82.1	48.3	76.0	41.6	68.3	35.2	62.8
	Yes	19.1	5.5	43.0	17.9	51.7	24.0	58.4	31.7	64.8	37.2

^aGED/Com. college = GED program/community college.

Sociodemographic Factors and Adolescent Experiences

We included several sociodemographic factors and adolescent experiences that have been found to be associated with the transition to adulthood. Self-reported racial-ethnic group was represented by mutually exclusive dummy variables for African American, Asian American, and Native American, with White being the reference category. Measures of economic status in childhood included participation in the National School Lunch and School Breakfast program at some point in the fifth, sixth, or seventh grade (1 = participated, 0 = did not participate) and parents’ average years of schooling (1–18). We measured academic performance in high school by taking the average of the self- and school-reported grade point averages in Grades 9–12 (0 = F to 4 = A). Parental divorce, separation, or widowhood (1 = family disruption, 0 = no family disruption), as reported by the parent, indicated family disruption during adolescence (Grades 5–10). Whether respondents were born to a teenage mother was coded as 1 = mother’s age at birth was 19 years or younger or 0 = mother’s age at birth was older than 19 years (the panel member reported age of mother). Measures of adolescent risk behavior included self-reported arrest by Grade 12 (1 = arrested, 0 = never arrested) and the average frequency of monthly substance use between Grades 7 and 12. The summary measure of adolescent substance use was created by averaging the self-reported frequency of use in the past 30 days across five drugs (alcohol, binge drinking, cigarette smoking, marijuana, and hard drugs—each with a possible range from 0 to 30) and five grades from Grade 7 to 12 (no data were collected in 11th grade). To examine the effect of the intervention, we included three dummy variables in the analysis for (a) the full intervention group, (b) the late intervention group, and (c) other conditions (parent-only intervention and nonclassifiable students). The control group was the reference category. Table 2 shows descriptive statistics for the predictor variables by gender.

Table 2.

Descriptive Statistics for Predictors of Latent Pathways

	Women			Men
	(N = 387)			(N = 395)
Predictor Variables	%	M	SD	%	M	SD
White (0/1)	47.5			46.8
African American (0/1)	25.1			24.8
Asian American (0/1)	20.4			24.6
Native American (0/1)	6.7			3.8
Free/reduced-price school lunch (0/1)	54.0			48.9
Parental education (1–18)		12.7	2.8		12.8	2.8
Family disruption (0/1)	47.3			44.3
HSa academic performance (0–4)		2.6	0.8		2.4	0.8
Born to teen mother (0/1)	13.7			14.7
Missing mother’s age (0/1)	10.9			14.4
Arrested in adolescence (0/1)	11.9			30.9
Missing arrest data (0/1)	6.2			10.1
Adolescent substance use (0–12)		0.6	1.1		0.8	1.4
Control condition (0/1)	25.8			28.6
Full intervention (0/1)	19.4			19.2
Late intervention (0/1)	34.6			31.6
Other conditions (0/1)	20.2			20.6

^aHS = High school.

Analysis

We used latent class analysis (Clogg, 1995) to examine how the four different role statuses interrelated across time and formed distinct life course trajectories. We entered all 20 variables (i.e., four role statuses measured at five time points) as indicators of a single latent variable. We conducted analyses separately for men and women, using Mplus 5.2 (L. K. Muthén & Muthén, 1998–2007). We also performed multiple-group analyses to test whether model parameters could be constrained to be equal for men and women. We compared alternative models using a chi-square difference test based on log likelihood values and scaling correction factors obtained with the MLR estimator in Mplus (L. K. Muthén & Muthén, n.d.).

We assessed the number of latent classes and relative model fit using the log likelihood, the Bayesian information criterion (BIC), and the Vuong-Lo-Mendell-Rubin likelihood ratio test (VLMR-LRT) (Nylund, Asparouhov, & Muthén, 2007). The log likelihood increases with increasing numbers of latent classes, but so does the complexity of the model. The BIC statistic takes into account both the log likelihood and the complexity of the model relative to the sample size. A lower BIC value indicates a better fitting model. The VLMR-LRT compares the fit of a model (e.g., a three-class model) with the fit of a model with one less latent class (e.g., a two-class model). A statistically significant p-value (≤ .05) indicates that a model with one more class fits better than a model with one fewer classes. We assessed quality of classification using the entropy statistic, which is comparable to a measure of internal reliability (B. O. Muthén, 2004; Petras & Masyn, 2010). We used multiple random starting values to avoid local solutions (Hipp & Bauer, 2006): 20 initial-stage iterations, 5,000 initial-stage random values, and 50 final-stage optimizations. First, we estimated different models with an increasing number of latent classes. Once we found the best fitting latent pathways model, we reestimated the model including sociodemographic factors and adolescent experiences as predictors of the latent pathways. Including predictors did not change the interpretation of the latent classes, which suggests that the model was appropriately specified (Nylund & Masyn, 2008; Petras & Masyn, 2010).

Missing Data

The proportion of missing data was low in the present study because of the high retention rate in the study and generally low nonresponse. Of the 412 men in the sample, 395 had available data for the analysis, 268 (68%) of which had complete data on the pathway indicators across all five time points; of the 396 women in the sample, 387 had available data, 288 (74%) of which had complete data. Of the total number of data points (20 variables × sample size), 5% were missing for women and 8% for men. Adjusted for missing data (N′ = N × [1 – proportion missing]; Graham & Hofer, 2000), the sample sizes were 362 for men and 366 for women. We used full information maximum likelihood estimation to include all available data and obtain unbiased estimates of model parameters and their standard errors (Schafer & Graham, 2002). Two exogenous predictor variables had somewhat higher rates of missing data and were recoded to increase the sample size available for analysis (110 respondents had missing data for their mother’s age at their birth and 65 respondents had missing arrest data). We incorporated missing cases on these variables by creating three mutually exclusive dummy variables for (a) “born to teen mother” (1 = yes, 0 = no) and “mother’s age missing” (1 = yes, 0 = no), with “not born to teen mother” (1 = yes, 0 = no) being the reference category; and (b) “arrested” (1 = yes, 0 = no) and “arrest data missing” (1 = yes, 0 = no), with “never arrested” (1 = yes, 0 = no) being the reference category.

RESULTS

Role Statuses by Gender

Figure 1 shows the percentage of men and women in the sample attending school, being employed full-time, being married, and living with children at each time point. School attendance dropped rapidly after age 18 for both men and women and was followed by a proportional increase in the percentage being employed full-time. By age 30, 15% of men and 14% of women were still attending school, and 79% of men and 70% of women were employed full-time. Men and women in the sample had similar rates of part-time employment (8% and 9%, respectively, at age 30), but more women (20%) than men (14%) were not employed year-round at age 30. The greatest difference between men and women was the timing of living with children. At age 18, 20% of women in the sample were already living with children, compared to 6% of men. Women also moved more rapidly than men into a parenting role during that period. By age 30, 65% of women but only 37% of men in the study were living with children. Women also became married somewhat earlier than men, although by age 30, 52% of women and 42% of men were married. The most striking difference between men’s and women’s roles during the transition to adulthood was the discrepancy between the percentage who were married and the percentage who lived with children at a given time. By age 30, 42% of men in the study were married and 37% were living with children, whereas 52% of women were married but 65% were living with children. Figure 1 considers each role status separately at each time point, however. To examine to what extent these roles intersected across time and how much variability in longitudinal pathways existed between genders, we turned to the multivariate latent class analysis.

Figure 1.

Role Statuses by Gender, Age 18–30.

Latent Pathways by Gender

The latent class analysis examined how school attendance, employment status, marital status, and living with children intersected across time and formed latent pathways to adulthood. Table 3 shows the fit statistics for models estimated separately by gender, including the simultaneous effect of sociodemographic and adolescent predictors on the latent pathways. The three-class models provided the best substantive and statistical fit for both genders. Subsequent multiple-group analyses indicated that models that constrained some or all of the conditional probabilities for the latent pathways to be equal for men and women fit significantly worse than the fully unconstrained model. Therefore, we present the results for men and women separately and describe similarities and differences below.

Table 3.

Fit Statistics for Latent Pathway Models

Number of Classes	Log Likelihood	Number of Free Parameters	BICa	VLMR- LRTb	Entropy
Women (N = 366)
1	−4,804.74	30	9,788.69	n/ac	n/a
2	−4,226.83	76	8,906.49	p < .001	.891
3d	−4,031.76	122	8,790.45	p = .004	.921
4	−3,905.43	168	8,811.87	p = .752	.941
5	−3,829.12	214	8,933.35	p = .763	.936
Men (N = 362)
1	−4,149.29	30	8,478.84	n/a	n/a
2	−3,695.28	76	7,844.96	p < .001	.889
3*	−3,512.60	122	7,754.62	p = .198	.849
4	−3,420.89	168	7,846.24	p = .761	.889
5	−3,338.63	214	7,956.75	p = .797	.916

^aBIC = Bayesian information criterion.

^bVLMR-LRT = Vuong-Lo-Mendell-Rubin likelihood ratio test.

^cn/a = not applicable.

^dSelected model.

Figures 2 and 3 plot each pathway’s probability profile for school attendance, full-time employment, being married, and living with children. For both genders, movement into marriage and living with children most clearly distinguished the three pathways. Involvement in postsecondary education was also a differentiating factor but to a lesser degree. The three pathways differed the least with respect to employment status. For both genders, one pathway was defined by a high probability of remaining unmarried through age 30 and having limited involvement in postsecondary education after high school. Twenty-seven percent of women and 26% of men were estimated to follow this pathway. A second pathway was defined by a high likelihood of marriage and living with children by the mid-20s, and by limited involvement in postsecondary education. Twenty-nine percent of women and 32% of men were estimated to be on this pathway. Investment in postsecondary education and postponement of family formation characterized a third pathway. This pathway was the most common transition pattern for both women (43%) and men (42%) in the sample. Tables 4 and 5 show the conditional probabilities for each of the latent pathways.

Figure 2.

Latent Pathways for Women (N= 366).

Figure 3.

Latent Pathways for Men (N= 362).

Table 4.

Latent Pathways and Conditional Probabilities, Women (N = 366)

	Unmarried Early Mothers (27.4%)					Married Mothers (29.3%)					PSa-Educated Women Without children (43.4%)
Age	18	21	24	27	30	18	21	24	27	30	18	21	24	27	30
School attendance
Not in school	0.14	0.73	0.84	0.85	0.91	0.11	0.63	0.78	0.83	0.87	0.01	0.35	0.66	0.75	0.83
In schoolb	0.70	0.28	0.16	0.15	0.09	0.81	0.37	0.22	0.17	0.13	0.97	0.66	0.34	0.25	0.17
GED/Com. collegec	0.15	—	—	—	—	0.08	—	—	—	—	0.02	—	—	—	—
Employment status
Not employed	0.44	0.29	0.17	0.32	0.27	0.32	0.12	0.26	0.31	0.32	0.29	0.04	0.07	0.07	0.07
Employed part time	0.45	0.31	0.11	0.07	0.05	0.58	0.32	0.18	0.14	0.10	0.68	0.43	0.12	0.13	0.11
Employed full time	0.11	0.40	0.72	0.61	0.68	0.11	0.56	0.56	0.54	0.58	0.03	0.53	0.81	0.80	0.82
Marital status
Never married	1.00	1.00	1.00	0.95	0.69	0.99	0.84	0.24	0.00	0.00	1.00	0.98	0.84	0.65	0.46
Married	0.00	0.00	0.00	0.05	0.31	0.01	0.14	0.61	0.82	0.77	0.00	0.02	0.15	0.32	0.49
Divorced	0.00	0.00	0.00	0.00	0.00	0.00	0.02	0.15	0.19	0.23	0.00	0.00	0.01	0.04	0.05
Living with children
No	0.57	0.19	0.09	0.02	0.05	0.77	0.45	0.19	0.06	0.04	0.99	0.91	0.94	0.92	0.77
Yes	0.43	0.81	0.91	0.98	0.95	0.23	0.55	0.81	0.94	0.96	0.01	0.09	0.07	0.08	0.23

^aPS = Postsecondary.

^bAt age 18, “In school” was “In high school.”

^cGED/Com. college = GED program/community college; only measured at age 18 (see Table 1).

Table 5.

Latent Pathways and Conditional Probabilities, Men (N = 362)

	Unmarried Men With Limited PSa Education (26.2%)					Married Fathers (31.6%)					PSa-Educated Men Without Children (42.2%)
Age	18	21	24	27	30	18	21	24	27	30	18	21	24	27	30
School attendance
Not in school	0.10	0.91	0.92	0.88	0.87	0.04	0.76	0.81	0.82	0.91	0.00	0.34	0.61	0.74	0.81
In schoolb	0.73	0.09	0.08	0.12	0.13	0.87	0.24	0.20	0.18	0.10	0.98	0.66	0.39	0.26	0.19
GED/Com. collegec	0.18	—	—	—	—	0.10	—	—	—	—	0.02	—	—	—	—
Employment status
Not employed	0.34	0.10	0.19	0.22	0.32	0.38	0.11	0.09	0.09	0.08	0.36	0.08	0.06	0.07	0.06
Employed part-time	0.50	0.26	0.06	0.23	0.11	0.46	0.19	0.03	0.07	0.05	0.61	0.38	0.15	0.10	0.09
Employed full-time	0.16	0.65	0.75	0.56	0.57	0.16	0.71	0.89	0.84	0.87	0.03	0.54	0.79	0.83	0.85
Marital status
Never married	1.00	1.00	1.00	1.00	0.87	0.99	0.94	0.45	0.14	0.00	1.00	1.00	1.00	0.94	0.67
Married	0.00	0.00	0.00	0.00	0.13	0.01	0.05	0.48	0.74	0.76	0.00	0.00	0.00	0.06	0.33
Divorced	0.00	0.00	0.00	0.00	0.00	0.00	0.01	0.07	0.12	0.24	0.00	0.00	0.00	0.00	0.00
Living with children
No	0.87	0.71	0.68	0.60	0.66	0.94	0.69	0.58	0.33	0.29	1.00	1.00	0.98	1.00	0.87
Yes	0.13	0.29	0.33	0.40	0.34	0.06	0.31	0.42	0.67	0.71	0.00	0.00	0.02	0.00	0.14

^aPS = Postsecondary.

^bAt age 18, “In school” was “In high school.”

^cGED/Com. college = GED program/community college; only measured at age 18 (see Table 1).

Despite the similarities in the definition and prevalence of each pathway, men’s and women’s pathways also differed in important ways. Men and women on the first pathway were similar with respect to having limited involvement in postsecondary education, remaining unmarried, and having varied employment statuses; yet men and women on this pathway differed dramatically in the probability of living with children (see Figures 2 and 3). Many of the women on this pathway took on a mother role early, already living with children at age 18 (probability π = .43). By age 21, the majority of the women on this pathway were parenting (π = .81). In contrast, few of the men on this pathway were likely to live with children at age 18 (π = .13) or later on during their 20s (probabilities ranged from π = .29 to π = .40). To reflect these differences and contrast them with the other pathways for each gender, we labeled this pathway “Unmarried Early Mothers” for women and “Unmarried With Limited Postsecondary Education” for men. Although both men and women on this pathway remained predominantly unmarried until age 27, more women than men had moved into marriage by age 30 (π = .31 versus π = .13, respectively). Finally, women in this group also were more likely than men to attend postsecondary schooling at age 21 (π = .28 versus π = .09, respectively) and consequently were less likely to work full-time at that age. After age 21, however, men’s and women’s school attendance and employment status were very similar on this pathway.

Moving into both marriage and living with children by the mid-20s characterized the second latent pathway, labeled “Married Mothers” for women and “Married Fathers” for men. This pathway differed, however, for men and women in the timing of when these family transitions occurred, particularly with respect to taking on parenting responsibilities, as indicated by living with children. Women in this group began living with children earlier than men: by age 21, more than half of the women in the “Married Mothers” category were likely to live with children (π = .55), compared to about one third of the “Married Fathers” (π = .31). By age 27, almost all of the women in this group were likely to live with children (π = .98), compared to about two thirds of the men (π = .67). It is important to note that, even on this pathway, where the majority of men and women were married by age 30, women were more likely to live with children than to get married, whereas men were more likely to combine the two family roles of marriage and living with children (see Figures 2 and 3). Last, married mothers and married fathers had almost identical educational trajectories, but men were more likely to be employed full time, particularly after age 21, when almost half of the women were likely to work part-time or not work outside the home. It is also interesting to note that the employment pattern of married mothers was different from that of the unmarried early mothers, who had similarly high probabilities of living with children (π ≤ .98 and π ≤ .96, respectively) but were somewhat more likely to work full-time (π ≤ .72 and π ≤ .58, respectively) and less likely to work part-time (π ≤ .45 and π ≤ .58, respectively); however, the two groups had similar patterns of not being employed.

Men and women were most similar with respect to the third pathway to adulthood, characterized by involvement in postsecondary education and postponement of family formation, particularly living with children. This pathway was labeled “Postsecondary-Educated without Children” for both genders. Men and women on this pathway differed, however, in the timing of marriage. Women on this pathway tended to marry earlier than men: about one third of the women were likely to be married by age 27 (π = .32), compared to just 6% of the men (π = .06). Women were also somewhat more likely than men to have children by age 30 (π = .23 versus π = .14, respectively), but for both genders, the probability of living with children was low. It is important to note that those men and women on the “Postsecondary-Educated Without Children” pathway who did move into family roles displayed ways of combining family roles that were similar to those found for the other two latent pathways. Women were more likely to live with children than to get married, but marriage clearly preceded taking on the role of living with children for men on this pathway.

Sociodemographic and Adolescent Predictors

Sociodemographic factors and experiences during adolescence predicted membership on the three latent pathways. Results in Tables 6 and 7 present the logit coefficient (B), its standard error (SE B), and the odds ratio (OR), which is the exponentiated logit coefficient (e^B), from the multivariate multinomial logistic regression of latent pathways on predictor variables.

Table 6.

Multivariate Multinomial Logistic Regression of Latent Pathways on Predictor Variables, Women (N = 366)

	Unmarried Early Mothers vs. Married Mothers			Unmarried Early Mothers vs. Women Without Children			Married Mothers vs. Women Without Children
Predictor variables	B	SE B	ORa	B	SE B	ORa	B	SE B	ORa
African Am. (1) vs. White (0)	.42	.41	1.52	.92*	.44	2.51	.51	.42	1.67
Asian Am. (1) vs. White (0)	.05	.73	1.05	−.50	.71	.61	−.54	.57	.58
Native Am. (1) vs. White (0)	.41	.82	1.51	.92	.81	2.51	.51	.91	1.67
Free/red. school lunch (1/0)	.15	.38	1.16	.42	.42	1.52	.27	.41	1.31
Parental education (1–18)	−.00	.07	1.00	−.22**	.08	.80	−.22**	.08	.80
Family disruption (1/0)	.62	.36	1.86	1.12**	.36	3.06	.50	.33	1.65
HSb academic performance	−.28	.28	.76	−1.14***	.31	.32	−.86***	.25	.42
Born to teen mother (1/0)	.06	.47	1.06	.57	.51	1.77	.52	.49	1.68
Missing mother’s age (1/0)	−.37	.63	.69	−.69	.63	.50	−.32	.57	.73
Arrested in adolescence (1/0)	.91	.70	2.48	.50	.71	1.65	−.42	.87	.66
Missing arrest data (1/0)	−.03	.63	.97	.18	.63	1.20	.21	.54	1.23
Adolescent substance use	−.02	.14	.98	.47	.32	1.60	.50	.33	1.65
Full intervention (1) vs. control (0)	−.08	.61	.92	−.54	.66	.58	−.46	.52	.63
Late intervention (1) vs. control (0)	−.68	.52	.51	−.44	.59	.64	.24	.43	1.27
Other (1) vs. control (0)	−.84	.48	.43	−.18	.58	.84	.65	.52	1.92

^aOR = Odds ratio (= e^B).

^bHS = High school.

^**p < .01.

^***p < .001.

Table 7.

Multivariate Multinomial Logistic Regression of Latent Pathways on Predictor Variables, Men (N = 362)

	Unmarried Men vs. Married Fathers			Unmarried Men vs. PSEa Men Without Children			Married Fathers vs. PSEa men without children
Predictor variables	B	SE B	ORb	B	SE B	ORb	B	SE B	ORb
African Am. (1) vs. White (0)	1.34	1.11	3.82	.73	1.54	2.08	−.61	.63	.54
Asian Am. (1) vs. White (0)	.79	1.21	2.20	−.21	1.41	.81	−1.01	.64	.36
Native Am. (1) vs. White (0)	1.18	1.36	3.25	2.79	1.46	16.28	1.61	.93	5.00
Free/red. school lunch (1/0)	.29	.44	1.34	.52	.51	1.68	.23	.39	1.26
Parental education (1–18)	.08	.18	1.08	−.11	.17	.90	−.19**	.07	.83
Family disruption (1/0)	.66	.59	1.93	.65	.56	1.92	−.01	.35	.99
HSc academic performance	−.96**	.35	.38	−1.55**	.48	.21	−.59	.51	.55
Born to teen mother (1/0)	1.03	.67	2.80	1.81*	.80	6.11	.77	.89	2.16
Missing mother’s age (1/0)	−1.18	.72	.31	−1.20	.90	.30	−.01	.46	.99
Arrested in adolescence (1/0)	.65	.65	1.92	1.39	1.03	4.01	.74	.76	2.10
Missing arrest data (1/0)	−.25	.77	.78	.62	.76	1.86	.88	.52	2.41
Adolescent substance use	.10	.18	1.11	.33	.31	1.39	.23	.22	1.26
Full intervention (1) vs. Control (0)	.48	.78	1.62	.62	1.01	1.86	.14	.51	1.15
Late intervention (1) vs. Control (0)	.06	.59	1.06	−.27	.87	.76	−.34	.52	.71
Other (1) vs. Control (0)	.87	.64	2.39	.38	.74	1.46	−.49	.55	.61

^aPSE = Postsecondary educated.

^bOR = Odds ratio (= e^B).

^cHS = High school.

^*p < .05.

^**p < .01.

Parental education and high school academic performance were common predictors of both men’s and women’s pathways to adulthood. Men and women whose parents had a higher level of education were more likely to invest in postsecondary education and postpone family formation during the young adult years. Specifically, women from more highly educated families were less likely to be among on the pathways of “Unmarried Early Mothers” and the “Married Mothers” than the “Postsecondary-Educated Women Without Children.” Men with more highly educated parents were less likely to be on the “Married Fathers” than on the “Postsecondary-Educated Men Without Children” pathway. Men’s and women’s own academic performance in high school was independently and positively associated with being on a pathway characterized by investment in postsecondary education and postponement of family formation. Women with higher grade point averages in high school were more likely to be on the “Postsecondary-Educated Without Children” pathway than either the “Married Mothers” or “Unmarried Early Mothers” pathway. Similarly, men with higher high school grade point averages were less likely to be among the “Unmarried Men With Limited Postsecondary Education” than either of the other two pathways, both involving greater investment in postsecondary education.

Other factors predicted pathways for only women or men. African American women were 2.5 times more likely than White women to be on the “Unmarried Early Mothers” pathway than on the “Postsecondary-Educated Without Children” pathway, yet they were equally likely to be among the “Married Mothers” (see Table 6). Race and ethnicity were not associated with men’s pathways to adulthood after controlling for all of the other covariates, yet in bivariate analyses (not shown), African American men were significantly more likely than White men to be among the “Unmarried Men With Limited Postsecondary Education” than among the other two pathways involving either family formation or postsecondary education.

For women, having experienced a family disruption during adolescence increased the likelihood of being on the “Unmarried Early Mother” trajectory more than threefold as compared to being on the “Postsecondary Education Without Children” pathway. For men, experiencing a family disruption as an adolescent was not significantly associated with any of the pathways, with or without controlling for other covariates. Having been born to a teenage mother, however, greatly increased men’s likelihood of being on the “Unmarried” pathway compared to being among men with postsecondary education and postponed family formation. Having been born to a teenage mother was not associated with women’s pathways to adulthood once we included the other predictors in the analysis. However, in bivariate analyses (not shown), having been born to a teenage mother was significantly and negatively associated with the “Postsecondary-Educated Without Children” pathway compared to both of the pathways involving the parenting role, as indicated by living with children. These findings suggest that having been born to a teenage mother was associated with earlier family formation and less investment in postsecondary education for men and women. Respondents with missing “born to teen mother” and missing arrest data did not differ significantly from the reference groups in their membership on latent pathways (see Tables 6 and 7).

Because of the covariation among the predictor variables, coming from a low-income family, as measured by participation in the free or reduced-price school lunch program, having been arrested in adolescence, adolescent substance use, and intervention status did not have significant additive effects on the likelihood of membership on the latent pathways for men or women in this study once we took into account all other factors. Bivariate analyses (not shown) indicated, however, that most of the predictor variables were significantly associated with pathways to adulthood in expected directions. For both men and women, more frequent substance use in adolescence, having been arrested during adolescence, and family poverty were significantly associated with a lower probability of being on the third pathway, characterized by investment in postsecondary education and postponement of marriage and living with children, than on one of the other two pathways. Despite findings that intervention status was associated with some of the indicators of the pathways (Hawkins, Kosterman et al., 2005; Lonczak et al., 2002), intervention status was not associated with the multidimensional latent pathways in this study.

DISCUSSION

This study sought to identify and compare contemporary young men’s and women’s pathways to adulthood and to examine their sociodemographic and adolescent precursors. Investigating how role transitions in education, work, marriage, and parenthood intersected and were linked across time to form pathways from age 18 to age 30 led to the identification of three latent pathways for both men and women. For both genders, participation in postsecondary education and timing of family formation distinguished the three pathways. A pathway characterized by involvement in postsecondary education and postponed family formation, as indicated by marriage and living with children, existed for both men and women. It was the most common pathway in this sample, characterizing 43% of women and 42% of men. The second most common pathway (for 29% of women and 32% of men) entailed getting married and living with children by the mid-20s, with much less involvement in postsecondary education. Slightly more than a quarter of women (26%) and men (27%) fit a third pathway, also characterized by limited participation in postsecondary education but without getting married.

Overall, the three pathways had similar characteristics for men and women and were about equally prevalent between both genders; however, men and women differed in the timing of marriage and in when they began to live with children and the likelihood of combining both roles. As expected, women moved into both marriage and living with children earlier than men and were more likely to be married and to live with children by age 30 on all pathways. The latent class analysis also showed that women were more likely to live with children without being married, whereas marriage often preceded or was more closely timed to living with children for men. Only a small proportion of men on the “Unmarried With Limited Postsecondary Education” pathway lived with children throughout their 20s without being married. It is striking to see the relatively large proportion of young adult women raising children outside of marriage, whereas most young adult men who were living with children were married.

This study also found that socioeconomic family background, early adversity, and antisocial behavior in adolescence predicted membership in the latent pathways. Although these factors were significantly associated with latent pathways in bivariate analyses, once we controlled all factors simultaneously, only educational factors, race-ethnicity, and early adversity (including family disruption and having been born to a teen mother) uniquely distinguished pathways to adulthood. The set of predictors included in the analysis best differentiated pathways with and without postsecondary education but had less predictive power for distinguishing between the two pathways with limited postsecondary education and earlier family formation. Future research may help identify childhood and adolescent factors that predict who is more likely to follow a pathway of young adult family formation than postponed marriage and early parenthood with limited postsecondary education.

The present study had several important strengths. The multidimensional, person-centered approach to studying the intersection of multiple role transitions simultaneously across time allowed for a more complex and accurate empirical picture of pathways to adulthood consistent with life course theory. Furthermore, the longitudinal design of the study, the contemporary panel of young adults, and the inclusion of nearly equal numbers of men and women allowed the comparison of contemporary men’s and women’s pathways to adulthood over the entire young adult period from age 18 to age 30. The sample also included a substantial representation of different ethnic groups, a large proportion of participants from poverty and from single-parent households, and had excellent retention.

The study also had limitations. In particular, it was based on a community sample of urban children. Generalizations of findings from this study must be made with caution. Prevalences of pathways and other characteristics apply to the panel studied here, though they are comparable to national rates (Rumbaut & Komaie, 2007). Many of the results from this analysis are comparable to those reported in other studies of pathways to adulthood using national and regional samples and similar person-centered methods (Amato et al., 2008; Macmillan & Eliason, 2003; Sandefur et al., 2005). Although generalizations from a single study must be made with caution, we have no reason to expect that the observed relationships among constructs should be found only in this sample.

The findings from this study suggest that, despite many social changes over the past decades, including a more diverse and individualized life course, three distinct pathways can describe the transition to adulthood for most contemporary young men and women. Despite discussions about the decline of the family and a retreat from marriage in U.S. society (Schoen & Cheng, 2006; Seltzer, 2000), the present study has indicated the continued relevance of marriage and raising children in the transition to adulthood, though perhaps differentially for men and women. The prevalences and timing of marriage and of living with children differed markedly with gender in this sample. Living with children outside of marriage was much more prevalent during young adulthood among women than among men. Marriage and living with children were far more likely to occur together among men than among women. To fully judge the degree to which young people today are moving away from marriage and family, it will be important to follow this sample into midlife to understand the degree to which those young men and women who have not gotten married or taken on the parenting role of living with children during their 20s are postponing or forgoing family formation entirely.