Pathways of Early Fatherhood, Marriage, and Employment: A Latent Class Growth Analysis

Abstract

In the National Longitudinal Survey of Youth 1979 (NLSY79), young fathers include heterogeneous subgroups with varying early life pathways in terms of fatherhood timing, the timing of first marriage, and holding full-time employment. Using latent class growth analysis with 10 observations between ages 18 and 37, we derived five latent classes with median ages of first fatherhood below the cohort median (26.4), constituting distinct early fatherhood pathways representing 32.4% of NLSY men: (A) Young Married Fathers, (B) Teen Married Fathers, (C) Young Underemployed Married Fathers, (D) Young Underemployed Single Fathers, and (E) Young Later-Marrying Fathers. A sixth latent class of men who become fathers around the cohort median, following full-time employment and marriage (On-Time On-Sequence Fathers), is the comparison group. With sociodemographic background controlled, all early fatherhood pathways show disadvantage in at least some later-life circumstances (earnings, educational attainment, marital status, and incarceration). The extent of disadvantage is greater when early fatherhood occurs at relatively younger ages (before age 20), occurs outside marriage, or occurs outside full-time employment. The relative disadvantage associated with early fatherhood, unlike early motherhood, increases over the life course.

Background

Using methods to uncover latent life course pathways, this article demonstrates that men who become fathers at an early age are distinct from other men and comprise several groups that are distinct from one another. Using a life course conceptual model for linkages among social origins, pathways to adulthood, and subsequent well-being, we test hypotheses about the concomitants of varying pathways to adulthood among men that include early fatherhood. Our results illustrate the importance of understanding how the timing and sequencing of life course transitions unfold. Our approach constitutes a model for how researchers grappling with the determinants and consequences of decisions taken at key developmental junctures can navigate the analytic difficulties posed by joint decision making.

Changes in the Marital Context of Births

A recent cross-sectional survey found that 25% of children were living with a single parent in 2004, up from less than 10% in 1950 (Kreider 2007). Longitudinal analysis suggests that more than one-half of American children spend some time in a single-parent family, a figure that is higher than in any other industrialized country (Heuveline et al. 2003). Most children in single-parent families are living with their mothers. This is cause for concern given the accumulating evidence that children living in single-parent households, especially those headed by a never-married mother, experience negative consequences (Cherlin and Furstenberg 1994; Fomby and Cherlin 2007; McLanahan and Sandefur 1994).

The changes in marriage and fertility behavior that result in this trend in children’s living arrangements are well-understood, but why these changes have taken place is debated. Labor market shifts, feminism, and state policy have all been implicated (Mason and Jensen 1995). Women’s expanding labor market opportunities raise the opportunity costs of marriage and childbearing (Ruggles 1997; Schoen et al. 2002). The decline in wages and opportunities for men with low levels of education make such men less attractive potential partners (Oppenheimer 2003). Some aspects of welfare policy are incentives to the formation of single-mother families (Hao et al. 2007; Moffitt 1998; Moffitt et al. 1998).

To these explanations for the changes in marriage and fertility behavior, McLanahan (2004) added two key insights. First, the societal changes that have been linked to changes in marriage and fertility behavior in the latter half of the twentieth century had different impacts on young people from the top of the socioeconomic ladder than on those at the bottom. Second, one crucial element of this differential impact was how these changes affected the incentives to delay or not delay family formation—particularly parenthood. Those who are well-off delayed family formation to invest in human capital, while the less-affluent had fewer incentives to delay parenthood but also reduced incentive to marry.

Emerging Adulthood

McLanahan’s (2004) contention that structural factors led young adults from different socioeconomic origins to take distinct pathways to adulthood bears on an issue in recent debates about the utility of the concept of “emerging adulthood.” Arnett (2000) proposed that modern industrial societies exhibit a new developmental stage, occurring after adolescence, that is characterized by instability, identity formation, self-focus, exploration, and “feeling in-between.” As with other life stages, the age boundaries of emerging adulthood cannot be specified precisely, but it is generally regarded as beginning in the late teens and extending through the mid- and even late 20s. Bynner (2005), by contrast, maintained that many disadvantaged young people are systematically excluded from the opportunity for exploration and identity formation occurring during emerging adulthood. Therefore, for structural reasons, less-affluent youth begin the transition to full adulthood early.

Especially relevant to this study, Arnett interpreted early adoption of adult roles—particularly teen parenthood—as an action by individuals that interrupts a developmental process that has become essential for socioeconomic success in modern societies. Arnett wrote:

Early Parenthood, Marriage, and Work

How does early parenthood affect later-life chances, and what is the role of marriage and employment in these effects? The current research consensus about women is that although early motherhood and subsequent economic disadvantage are correlated, the causal role of early childbearing in bringing this disadvantage about is small, if it exists at all (Astone and Upchurch 1994; Furstenberg 1991; Geronimus 1994; Jaffee 2002). Also, it appears that in the years immediately following a birth, young mothers are disadvantaged compared with their peers who delay childbearing; still, many resilient young mothers recoup, and the differences between young and older mothers are not so profound in mid-adulthood (Furstenberg et al. 1987). Because so many young mothers are also unmarried (Astone 1993; Barber and Emens 2006), some research focuses on the consequences of young, unmarried motherhood specifically (Beutel 2000; Moore et al. 1998). Women who become unmarried mothers are disadvantaged relative to married mothers regardless of their age at first motherhood (Foster et al. 1998). These latter studies, however, have not explored the extent to which this disadvantage persists over the life course.

Much of the literature on early fatherhood concerns its determinants (Lamb and Elster 1986; Lerman and Ooms 1993; Marsiglio and Cohan 1997; McLanahan and Carlson 2004). Research on the later lives of early fathers is limited to three principal studies. This research has been attentive to how the later-life experience of young fathers varies by the marital context of the early birth. In a British 1970 birth cohort followed to age 30, when social background was controlled, young fathers (who became fathers before age 22) who were married or cohabiting at the time of the birth did not fare significantly better on four employment-related outcomes than fathers who were not in unions. Sigle-Rushton argued, nonetheless, that the pattern of nonsignificant better employment outcomes for fathers in unions, especially marriage, “does lend some credence to the hypothesis that strong social bonds might be protective” (Sigle-Rushton 2003:747).

In a study of educational attainment among teen fathers in the National Longitudinal Survey of Youth 1979 (NLSY79), in simple bivariate association, 39% of teen fathers who were married at the time of the birth had completed high school at ages 20–24 (i.e., as of the 1984 wave), while 62% of the unmarried had (Marsiglio 1987). Almost three-quarters of teen fathers, however, had completed high school or dropped out before the birth. In a further analysis in the subsample in which the birth preceded high school completion or dropping out, and controlling for demographics, living with the child after the birth (about two-thirds of whom were married) was not associated with high school completion. A related analysis likewise found no association with highest grade completed at ages 20–24 (Marsiglio 1986).

Finally, Nock (1998) found unmarried fatherhood through age 25 was associated with decreased odds of subsequent marriage and increased odds of cohabitation. Compared with men without unmarried births, those having unmarried births at ages 12–19 completed less education and had lower odds of year-round employment at ages 29–36. Those having unmarried births at ages 20–25 additionally showed lower earnings and were more likely to be poor. Other analyses suggested that a principal mechanism by which nonmarital fatherhood has disadvantaged later sequelae is via decreasing the likelihood of subsequent marriage.

Although the marital context of early parenthood has received some attention, little research exists on how the correlates of early parenthood vary by employment status at the time of birth. For women, this gap probably occurs because mothers have lower rates of labor force attachment than nonmothers at any age, and theory does not lead to any obvious a priori hypothesis about how this association differs by the age at motherhood (Astone 1993). In the case of fathers, the paucity of research on how the correlates of young fatherhood vary by the work context is more surprising given men’s traditional role as breadwinner with resident children as well as the current concern about men financially supporting children with whom they do not live.

Some recent work, however, does examine how marital context relates to employment among fathers. In the year before becoming a father, married men work more weeks per year and many more hours per week than either cohabiting or nonresidential men who are soon to be fathers (Percheski and Wildeman 2008). Five years later, however, these differences no longer hold: unmarried men increase their work, while married fathers maintain their work levels. When background variables are controlled, differences at baseline and five years later no longer exist for the number of weeks worked per year; yet, married men still maintain a significant lead in hours worked over unmarried men even though the values in both groups diminished. Other research supports the finding that increased work effort (in hours worked) is associated with becoming a first-time father for unmarried men but not for married men (Astone et al. 2010).

Selection

Whether early parenthood—or other aspects of the transition to adulthood—is independently associated with subsequent well-being is a key question for public policy. If it is, and the association can be demonstrated as causal, then interventions to delay parenthood (e.g., family planning) or change other aspects of the transition to adulthood (e.g., job training programs) might be able to enhance social mobility.

A good deal of evidence exists, however, to support the proposition that the nature of the transition to adulthood is a consequence, rather than a cause, of social inequality (Lawlor and Shaw 2002). In the studies reviewed earlier, selection factors account for much of the poorer later-life outcomes experienced by men who become fathers when young, unmarried or both compared with those who do not, although some differences remain (Marsiglio 1986, 1987; Nock 1998; Sigle-Rushton 2003). A conceptual model within which the consequences of transitions to adulthood varying in the timing of parenthood can be studied, taking selection into account, is needed.

Conceptual Model

Our conceptual model is illustrated in Fig. 1. The transitions to fatherhood as part of the transition to adulthood more broadly, as life course transitions, are interdependent (Hagestad and Call 2007), particularly across work and family domains (Moen et al. 2008). Men take varying pathways to adulthood that are characterized by the timing and sequencing of these multiple transitions. Varying patterns of transition to adulthood are an outgrowth of a young man’s social origins, and simultaneously influence later-life circumstances net of social origins.

The theoretical basis for our expectation that early fatherhood pathways are associated with more disadvantaged subsequent lives is provided by Arnett’s notion that teen parenthood prevents individuals from carrying out the developmental tasks of emerging adulthood. By extension, although parenthood in the early 20s does not completely exclude some period of emerging adulthood, it foreshortens it. Thus, we expect all early fatherhood pathways to be linked to subsequent life disadvantage. At the same time, disadvantage is likely to be greater when birth occurs in the teens than when birth occurs in the early 20s.

Opposing hypotheses are possible about whether early marriage heightens or mitigates the foreshortening of emerging adulthood associated with early fatherhood. On the one hand, the addition of early marriage to early parenthood may represent more of a “disrupted pathway” (Sigle-Rushton 2003), in part because early marriage is associated with higher levels of divorce and lower levels of socioeconomic attainment for both men and women (Alexander and Reilly 1981; Bartz and Nye 1970; Call and Otto 1977; Kerckhoff and Parrow 1979; Marini 1978). The alternative view is that marriage provides social support from the partner, kin, and others that fosters later positive outcomes for the individual. For early fathers, remaining unmarried may represent a “retreat from adult responsibilities” associated with diminished subsequent well-being (Sigle-Rushton 2003). Because there is more evidence for the latter interpretation (Sigle-Rushton 2003), we hypothesize that early fatherhood pathways including early marriage will be associated with less disadvantaged later-life circumstances than pathways without it.

Early employment may provide benefits promoting later-life advantages for young fathers, similar to those potentially provided by early marriage. In addition, early labor force attachment promotes later labor force attachment as associated positive correlates. We therefore hypothesize that being employed around the time of the birth is protective for early fathers.

Research Questions and Hypotheses

Two research queries guide this study. First, do all young fathers have similar early life pathways? If not, how do their pathways vary in terms of the timing of first fatherhood (earlier vs. later), the timing of first marriage, and their full-time employment status across time, and how are these varying pathways related to social origins?

Second, how are early fatherhood pathways associated with subsequent well-being, with social origins controlled? We hypothesize that men who take an early fatherhood pathway have more disadvantaged subsequent lives than those who take a pathway to adulthood with normative timing and sequencing of fatherhood, marriage, and full-time employment (Hypothesis 1).

Among those whose pathway to adulthood includes an early transition to fatherhood, more disadvantaged subsequent lives are associated with pathways characterized by first fatherhood as a teen versus in the early 20s (Hypothesis 2); pathways in which the first birth occurs outside marriage (Hypothesis 3); and pathways in which the first birth occurs when the father does not have full-time employment (Hypothesis 4).

Finally, we also hypothesize that the later-life disadvantages associated with early fatherhood decrease over the life course as they do for women (Hypothesis 5).

Data and Methods

Sample

The 1979 cohort of the National Longitudinal Survey of Youth (NLSY79), a nationally representative sample of youth aged 14 to 21 in 1979, is the data source for this study. These youth were interviewed annually until 1992 and biennially since. These analyses are limited to the “cross-sectional”¹ sample representative of the noninstitutionalized civilian population of young people born from 1957 to 1964. We excluded female respondents, resulting in a sample size of 2,800 men who were black, white, or Latino.

Variables for Latent Class Analysis

We created three binary variables for each of 10 approximately evenly spaced ages (18, 20, 22, 24, 26, 28, 30, 32, 35, and 37): ever-fatherhood, ever-married, and full-time employment status.

Ever-Fatherhood Status

We used the birth dates of the respondent and his oldest child to calculate the respondent’s age at first biological fatherhood. Respondents who never transitioned into fatherhood by age 37 were coded 0 on this variable for all observed ages. Men who became fathers were coded 1 for the age at first fatherhood and all subsequent ages. For example, men who became fathers at 19 were coded 1 for ages 19 through 37. There were 1,873 (66.9%) fathers.²

Ever-Married Status

We use a similar strategy for marriage. More than 2,000 (2,110; 75.4%) were ever-married. Given that the construct was ever-married, men who separated or divorced were still coded as 1.

Full-Time Employment Status

We aggregated the weekly labor force activity data to calculate each man’s median yearly work hours. At each observation, men who worked 1,440 hours or more a year (consistent with working 30 hours per week for 52 weeks) were coded as 1, and men who worked fewer hours were coded as 0. Unlike the marital and fatherhood status variables, full-time work status is allowed to vary over time from age 18 to 37.

The analytic technique to be used in the analyses, latent class growth analysis, requires that respondents have complete data on all role status variables in the model. This would result in listwise deletion of one-third of the sample; variables missing values tend to be employment variables at earlier ages. We therefore used missing data imputation with the EM algorithm (SPSS 17.0, Missing Data Analysis module) for variables on which respondents were missing values. Allison (2000) suggested that this algorithm is appropriate when less than 5% of data are missing; for the role status variables, 1.2% of the data were missing (980 observations of 84,000).

Demographic Background Characteristics and Covariates

Race/Ethnicity

Our analyses include men of three racial/ethnic backgrounds: white/other (non-Hispanic whites, Asian Americans, missing ethnicity), black (non-Hispanic blacks), and Latino. The sample consisted of 79.9% white/other, 12.4% black, and 7.8% Latino male youth.

Youth Poverty

We used youth poverty status variables from 1978/1979³ (1 = in poverty; 0 = not in poverty). These variables were created based on measures of family income at the time each youth entered the study (from ages 14 to 21). Approximately 11.5% of the sample experienced youth poverty.

Family Structure at Age 14

We recoded living arrangements at age 14 into four categories: living with both biological parents (76.1%), one biological parent only (13.4%), one biological parent and a stepparent (8.2%), and neither biological parent (2.3%).

Mother’s Characteristics

Mother’s highest level of educational attainment, a continuous variable, was collected in 1979. On average, mothers completed less than a high school education (11.7 years, SD = 2.7). We also use mother’s age at the time of the respondent’s birth. On average, respondents’ mothers were 26.3 years old (SD = 6.5) when they gave birth to respondents.

There were some missing data on demographic background characteristics. We again used missing data imputation with the EM algorithm (SPSS 17.0, Missing Data Analysis module) for variables on which respondents were missing values. For this set of background variables, 3.4% of the data were missing (860 observations of 25,200).

Region

For use as a covariate, we used region (South, Northeast, North Central, and West) at the age of the later-life circumstances examined in this study (see next section)—for example, region at age 26 for earnings at age 26, and region at age 37 for earnings at age 37. The exception is for highest educational attainment, for which we used region at the start of the survey (1979) because educational attainment is influenced more by early environment than by region of current residence.

The demographic background characteristics of the study sample are reported in Table 1. Distributions are reported including imputed values, but Ns of respondents completing each item are also provided.

Later-Life Circumstances, Respondent’s Earnings, Earnings per Child, and Current Marital Status: Ages 26 and 37

In addition to later-life circumstances, three other variables were assessed at ages 26 and 37. First, annual earnings for each survey year were calculated based on wage data for each respondent. The sample average earnings were $16,882 (SD = $12,434) at age 26, and $39,336 (SD = $36,565) at age 37. Second, as mentioned previously, men reported their earnings at ages 26 and 37. For men who were fathers at ages 26 and 37, earnings per child were computed: $12,302 (SD = $10,058) at age 26, and $23,076 (SD = $23,225) at age 37. Third, men reported their current marital status at ages 26 and 37. At age 26, 41.6% were married. Nearly three-fifths reported being married at age 37 (58.9%).

Respondent’s Highest Educational Attainment at Age 37

Respondents reported their highest grade of educational attainment at each observation. At age 37, the sample average was 13.5 years of education (1.5 years beyond high school; SD = 2.5).

Incarceration by Age 26

Given that the NLSY79 does not contain an item that directly asks men whether they ever spent time in prison or jail, incarceration by age 26 was created by using residence items from age 18 to 26.⁴ If a man ever reported that he was currently residing in jail or prison, he was coded as having a history of incarceration (N = 179; 6.4%). This is an underestimate given that respondents could have been incarcerated between survey years but not at survey years.

Limited Work

If a man reported that by age 26 and by age 37, his work was limited by his health or that he was disabled or unable to work, he was coded as being limited in work at that age (by age 26, N = 436, 15.6%; by age 37, N = 655, 23.4%).

Analytic Strategy: Latent Class Growth Analysis

This method empirically identifies varying life course patterns: in this case, patterns of the acquisition of roles over the life course, marriage, and full-time employment. In LCGA, each identified class represents a pattern of behaviors across the ages examined. We derive distinct classes (or subgroups) of men who are homogeneous with respect to patterns of these indicators over time. Each LCGA class represents a configuration of both the ordering of the three transitions and the age at which each transition occurs.

At each time point, with three dichotomous variables, there are eight possible combinations of statuses. The total number of possible combinations across all 10 time points is 8¹⁰. Although backward transitions for fatherhood and marriage (e.g., being scored 1 for transition to fatherhood at age 20 but scored 0 at later ages) are excluded, the number of theoretically possible combinations is nonetheless unmanageably large. LCGA reduces these combinations into a smaller number of latent classes representing common patterns of the ordering, timing, and spacing of first fatherhood, first marriage, and full-time employment over the period from age 18 to age 37.

The current analyses were conducted by using Proc LCA for SAS 9.1.

Analytic Strategy: Linear and Logistic Regression

For continuous later-life correlates, we use linear regression to determine whether latent classes (interpreted here as “pathways”) are associated with later-life correlates after controlling demographic background characteristics and other covariates. For the dichotomous measures (marital status at age 26 and at age 37, and incarceration by age 26), we use logistic regression.

Results

Are There Distinct Pathways of Early Fatherhood?

The Alaike information criterion (AIC) and Bayesian information criterion (BIC) for LCGA models with different numbers of classes were used to determine the best fitting model. The model with 15 classes had the best fit (AIC = 13,461, BIC = 16,216, compared with 13,995 and 16,566 for the 14-class model, and 13,501 and 16,440 for the 16-class model) and was also the most interpretable.

The median age at first fatherhood for five of the 15 classes was lower than the median age of first fatherhood (26.4) in the National Survey of Family Growth (NSFG) 2002 for men of the NLSY79 cohort (i.e., those born between 1958 and 1965). These five classes, constituting 32.4% of the sample, are interpreted as early fatherhood pathways. Serving as a comparison group is a latent class termed here as the On-Time On-Sequence Fathers pathway (12.8% of the sample).

In addition to the five classes of early fathers and the class of On-Time On-Sequence Fathers analyzed in this article, five other classes represent men who do not have children by age 37 (referred to as “foregoers”; Dariotis et al. 2009). Two classes include men who are, on average, older than the cohort median when they have their first child (referred to as “older fathers,” Dariotis et al., 2009), one additional class that is composed of half those men who become fathers late and half who do not become fathers by age 37 (a class that cannot be characterized simply as either foregoers or older fathers), and one class includes men whose first fatherhood trajectory is nearly identical to the reference group of On-Time On-Sequence Fathers (starting at age 22 or 24 and all having children by age 32) but among whom the proportion of full-time employment never reached 40% (a class that cannot be characterized as early fathers, older fathers, or foregoers).

For each early fatherhood pathway, as well as the On-Time On-Sequence Fathers pathway, the proportions attaining first fatherhood and first marriage and holding full-time employment at each age observation are reported in Table 2 and graphically depicted in Fig. 2. Because On-Time On-Sequence Fathers serve as a point of comparison for the early fatherhood pathways, we discuss them first.

The On-Time On-Sequence Fathers Pathway

This pathway, representing 12.8% of the full sample, evinces median ages of first fatherhood, first marriage, and rates of full-time work that are close to the medians observed in the NSFG 2002 for men aged 38 to 42 (same cohort as NLSY79 men; authors’ calculations). These men are normative with respect to both the timing and the sequencing of work, marriage, and fatherhood. The age by which one-half of these men are currently employed full-time is slightly older than age 20; the age by which one-half have married for the first time is nearly 23; and the age by which one-half have become fathers is approximately 27.

The names assigned to the early fatherhood pathways presented in this article distinguish whether first births occurred predominantly younger than age 20 (“teen”), or occurred either relatively equally before and after age 20 or predominantly after age 20 (“young”). If a low proportion of pathways members were fully employed at the time of the birth, the pathway is characterized as “underemployed”; use of this term connotes that a minority of men in the pathway were fully employed, not that individual men in the pathway were underemployed. Pathways whose names are unmarked for employment were predominantly fully employed at the birth.

The Young Married Fathers Pathway

One-half the men who take this pathway are fathers by age 23, about 4 years earlier than the median for On-Time On-Sequence Fathers. Fatherhood is preceded by employment and marriage, in that order. One-half the men in this pathway work full time from age 19.5 onward. The median age at first marriage for men of this pathway is 22.1 years. These men differ from On-Time On-Sequence Fathers in that they start their role transitions earlier and also proceed from one to the next more quickly. This class comprises 13.2% of the full sample and represents 40.9% of all early fathers.

The Teen Married Fathers Pathway

Seventy percent of this group fathered a child by their 20th birthday. These men typically marry prior to the first birth, but the interval between marriage and birth is quite short (median ages differing by 0.6). In addition, these fathers engage in full-time employment coincident with first birth, rather than prior to it, as Young Married Fathers do. This class comprises 8.5% of the full sample and 26.2% of early fathers.

The Young Underemployed Married Fathers Pathway

Men in this pathway marry early and simultaneously with their first birth; for many, marriage might be triggered by the pregnancy. The distinctive feature of this pathway is its low rates of full-time employment through the life course, not rising above 40% at any age; the rate fluctuates between 18% and 35% full-time employment between age 18 and 37. We characterize the pathway as “underemployed” to connote that at any given year, only a minority of class members are employed full time—not that individual men in the class are underemployed. This class comprises 4.5% of the full sample and 13.8% of early fathers.

The Young Underemployed Single Fathers Pathway

Men in this pathway are similar to Young Underemployed Married Fathers in two respects. Men who take this pathway likewise straddle 20 as the age at which half report a first birth, and they do not reach full-time employment at any age. The proportion in full-time employment increases through the early 20s and peaks at just less than 50% from age 24 to age 28, followed by a slow decline until age 37. However, this pathway diverges markedly from the Young Married Underemployed Fathers in its very low rate of marriage, the lowest of any early fatherhood pathway. No men in the Young Underemployed Married Fathers pathway report a marriage until after age 35, and fewer than 5% marry by age 37. It is possible that many of these marriages are to a woman other than the mother of their oldest child. This class comprises 4.0% of the full sample and 12.3% of early fathers.

The Young Later-Marrying Fathers Pathway

Although all men in this pathway marry by age 35, all become fathers while they are unmarried, one-half doing so by age 22. In contrast with Young Underemployed Single Fathers, full-time employment in this pathways rises in tandem with becoming a father until age 24, after which age full-time employment continues to increase slowly, reaching nearly 90% by age 37. This class comprises 2.2% of the full sample and 6.9% of early fathers.

Demographic Background Characteristics Associated with Early Fatherhood Pathways

Table 3 shows bivariate associations between early fatherhood pathway and four sociodemographic background characteristics: race/ethnicity, youth poverty status, family structure at age 14, and mother’s education.

All Early Fatherhood Pathways Versus On-Time On-Sequence Fathers

When the men in all early fatherhood pathways are pooled (Table 3, top panel, row 3) and contrasted with On-Time On-Sequence Fathers (row 2), they are significantly more likely to be ethnic minorities. These men are also more likely to have disadvantaged backgrounds in terms of youth poverty, family structure, and maternal education.

Heterogeneity Among Early Fatherhood Pathways

Table 3 indicates considerable heterogeneity among early fatherhood pathways. The five pathways differ from each other in racial/ethnic compositions in complex ways. Young Married Fathers and Teen Married Fathers are significantly more likely to be white than the other three early fatherhood pathways. Within the three other pathways, Young Married Underemployed Fathers are more likely to be white than the other two. All early fatherhood pathways except Young Married Fathers include lower proportions of whites than On-Time On-Sequence Fathers.

Young Underemployed Single Fathers show a significantly higher proportion of youth poverty than do the other early father pathways, which otherwise do not differ from one another. Young Underemployed Married Fathers and Young Underemployed Single Fathers exhibit higher proportions of youth poverty than On-Time On-Sequence Fathers. Young Underemployed Married Fathers and Young Underemployed Single Fathers have a lower proportion living with both parents at age 14 than do Young Married Fathers and Teen Married Fathers, and the former two groups are the only pathways with significantly lower proportions of youth poverty than On-Time On-Sequence Fathers. Young Married Fathers, Teen Married Fathers, and Young Later-Marrying Fathers do not differ from one another with respect to family structure at age 14. For mother’s years of education, none of the five early fatherhood pathways significantly differ, but all five are significantly lower than On-Time On-Sequence Fathers.

In summary, all early fatherhood pathways pooled together show more disadvantaged demographic backgrounds than On-Time On-Sequence Fathers. In addition, the five early fatherhood pathways differ from one another in complex ways. A general pattern is that the two pathways with lower proportions of full-time employment (Young Underemployed Married Fathers and Young Underemployed Single Fathers) tend to have more disadvantaged backgrounds than the pathways with high proportions of fully employed men (Young Married Fathers, Teen Married Fathers, and Young Later-Marrying Fathers).

How Are Early Fatherhood Pathways Associated With Later-Life Circumstances?

For Hypothesis 1 (early fatherhood is associated with later-life disadvantage), we compare each early fatherhood pathway with On-Time On-Sequence Fathers. We use this approach (rather than compare the pool of all five early fatherhood pathways with On-Time On-Sequence Fathers) because of the heterogeneity observed among the early fatherhood pathways.

Table 4 summarizes the comparisons among early fatherhood pathways involved in testing Hypotheses 2–4. For Hypothesis 2 (age context), we compare Teen Married Fathers with Young Married Fathers, given that these pathways differ only in the timing of fatherhood and not in employment and marital status at or around the time of the birth.

For Hypothesis 3 (marital context), we make two comparisons. We first contrast Young Married Fathers with Young Later-Marrying Fathers, who have similar age and employment status at the time of the birth but differ in marital status at that time. We then compare Young Underemployed Married Fathers and Young Underemployed Single Fathers, given that the timing and employment context of their first birth are similar but the marital context differs. The first comparison indicates how the linkage between early fatherhood and later-life circumstances varies by martial context at birth among men in high-employed pathways, while the second contrast does so among men in low-employment pathways.

For Hypothesis 4 (employment context), we compare Young Married Fathers, Teen Married Fathers, and Young Later-Marrying Fathers (pathways in which the majority of members are fully employed at the time of birth) with Young Underemployed Married Fathers and Young Underemployed Single Fathers (pathways in which only a minority of members are fully employed). In making these comparisons, we take into account that the first set of pathways differs from the second set in other respects besides proportions employed at the time of birth.

Finally, for Hypothesis 5 (disadvantage decreases with age), we compare results for the subset of later-life correlates that are available at both age 26 and age 37.

Tables 5, 6, 7, and 8 report linear regression analyses and logistic regression analyses of later-life circumstances as a function of early fatherhood pathway, with On-Time On-Sequence Fathers as the reference category. The social background differences we observed between early fatherhood pathways combined and On-Time On-Sequence Fathers, and among early fatherhood pathways, suggest selection processes that need to be controlled in all analyses. Thus, we report three regression models. Model 1 analyzes the relationship between later-life correlates and each early fatherhood pathway (On-Time On-Sequence Fathers are the reference category) with no other independent variables. In Model 2, only demographic background variables and current covariates (referred to collectively as controls) are used as predictors. Both early fatherhood pathways and controls are predictors in Model 3. Table 9 provides means and proportions for the later-life correlates adjusted for the controls, with significance tests comparing early fatherhood pathways with one another as well as with On-Time On-Sequence Fathers.

Early Fatherhood Pathways Versus On-Time On-Sequence Fathers (Hypothesis 1)

According to Hypothesis 1, men in early fatherhood pathways will evidence disadvantage on later-life correlates compared with On-Time On-Sequence Fathers. Results for earnings at ages 26 and 37, earnings per child at ages 26 and 37,⁵ and highest educational attainment at age 37 are reported in Tables 5, 6, and 9. Young Underemployed Married Fathers, Young Underemployed Single Fathers, and Young Later-Marrying Fathers report significantly lower earnings at age 26 than do On-Time On-Sequence Fathers. These three pathways as well as Teen Married Fathers also have lower earnings at age 37 than On-Time On-Sequence Fathers. For earnings per child at both ages 26 and 37, all five early father pathways earn significantly lower amounts than On-Time On-Sequence Fathers. For educational attainment by age 37, all five pathways also report significantly fewer years relative to On-Time On-Sequence Fathers. For all five variables, the increments in explained variance in Model 3 over Model 2 are significant.

Tables 7, 8, and 9 provide results for marital status and incarceration. For current marital status at age 26, due to the lack of variance in Young Underemployed Single Fathers and Young Later-Marrying Fathers (none were married at age 26), we are not able to statistically compare them with other groups in testing any of the hypotheses. Young Underemployed Married Fathers have significantly lower odds of being currently married at age 26 than do On-Time On-Sequence Fathers. All young father pathways except Young Married Fathers and Young Later-Marrying Fathers have significantly lower odds of being currently married at age 37 than On-Time On-Sequence Fathers. For incarceration by age 26, men in three of the five early fatherhood pathways are significantly more likely to be incarcerated than On-Time On-Sequence Fathers. Young Married Fathers, by contrast, have significantly lower odds of incarceration than On-Time On-Sequence Fathers, however.

Thus, of 38 comparisons (comparing five early father pathways with On-Time On-Sequence Fathers for seven measures and three pathways for one measure, marriage at age 26), significant differences between early fathers and On-Time On-Sequence Fathers occur in the expected direction for 30. Later-life disadvantages relative to On-Time On-Sequence Fathers are evident for every later-life correlate studied here for the three pathways in which fathers were predominantly not married and/or not fully employed at the birth (Young Underemployed Married Fathers, Young Underemployed Single Fathers, and Young Later-Marrying Fathers). Relative disadvantage occurred less consistently in the two pathways in which fathers were married and fully employed at the birth (Young Married Fathers and Teen Married Fathers), but disadvantage is nonetheless evident even in those pathways for earnings per child at age 26 and age 37 and for educational attainment.

Differences in social background, as well as in work limitation and region of residence in later life, account for some of the later-life disadvantage associated with young fatherhood. Background and covariate differences, however, do not account for all the disadvantage observed. The first point is shown by comparing the standardized regression coefficients and odds ratios in Model 1 (only early fatherhood pathways as independent variables) with those in Model 3 (early fatherhood pathways and controls). Compared with values in Model 1, Model 3 regression coefficients become closer to zero, and odds ratios become to closer to 1.0; also, some significant values become nonsignificant. The second point is evidenced by the significant coefficients for early fatherhood pathways in Model 3 in which social background and current covariates are controlled. In addition, for six of the eight later-life correlates, Model 3 has better fit than Model 2.

Age at First Birth (Hypothesis 2)

We hypothesize that among early fathers, becoming a father as a teen is associated with greater disadvantage in later life than becoming a father in the early 20s. To test this hypothesis, we compare Young Married Fathers and Teen Married Fathers, pathways differing on the degree of “earliness” of the early transition into fatherhood (in the teens vs. in the early 20s) while being similar in the marital and employment context of the birth. Significant differences between these two pathways are evident for five of the eight later-life correlates after adjusting for controls (Table 9). Compared with Young Married Fathers, Teen Married Fathers report lower earnings at age 37 ($6,574 less, comparing adjusted means in Table 9, p < .05), and lower earnings per child at both age 26 ($2,862 less) and at age 37 ($3,033 less). In addition, Teen Married Fathers report lower educational attainment at age 37 (0.80 years less). Teen Married Fathers had lower odds of being currently married at age 37 than Young Married Fathers: .71 compared with .79 (Table 9). No significant differences were evident, however, for three other later-life correlates: earnings at age 26, currently married at age 26, and incarceration by age 26.

Thus, the hypothesis that among young fathers, relatively earlier age at first fatherhood is associated with greater later-life disadvantage is partially confirmed. Disadvantage is more often confirmed on the later-life correlates assessed at age 37 than on correlates assessed at age 26. See Hypothesis 5 for further analysis of trends in disadvantage over the life course.

Marital Context of First Birth (Hypothesis 3)

To test the hypothesis that being unmarried at the first birth exacerbates later disadvantage, we first compare Young Married Fathers and the Young Later-Marrying Fathers, pathways in which men were of comparable age and were predominantly fully employed at the time of the birth, but differed in marital status at that time. Young Married Fathers have significantly higher earnings at age 26 and age 37, and have lower rates of incarceration by age 26 than Young Later-Marrying Fathers (Table 9).

We then compare Young Underemployed Married Fathers and Young Underemployed Single Fathers because they differ in the marital context of first birth while being similar in other respects. The two underemployed pathways do not differ significantly on either earning or earnings per child at ages 26 and 37, reflecting their similarity in their low proportions of fully employed men over the part of the life course examined here (Table 2 and Fig. 2). Those unmarried at the birth, however, do exhibit significantly fewer years of completed education (0.85 years less). Young Underemployed Single Fathers have substantially lower odds of being currently married at age 37, however. Finally, Young Underemployed Single Fathers are twice as likely as the contrast group to be incarcerated by age 26. The overall pattern of results yields support for the hypothesis in somewhat less than one-half the contrasts examined.

The Employment Context of First Birth (Hypothesis 4)

We hypothesize that men who work less than full time concurrently with or soon after the birth exhibit greater lifetime disadvantage relative to men who work full time. We compare the two pathways without full-time employment at the time of the birth (Young Underemployed Married Fathers and Young Underemployed Single Fathers) with the three pathways with full-time employment (Young Married Fathers, Teen Married Fathers, and Young Later-Marrying Fathers); thus, for each variable considered, comparing each of the first two groups with each of the latter three groups involves six comparisons. Both Young Underemployed Married Fathers and Young Underemployed Single Fathers report significantly lower earnings at ages 26 and 37 than each of the other three groups, as well as significantly lower earnings per child at both ages (all six comparisons supporting the hypothesis for each variable). On educational attainment at age 37, Young Underemployed Single Fathers report fewer years than two of the three fully employed pathways: Young Married Fathers and Young Later-Marrying Fathers. Young Underemployed Married Fathers, however, did not differ from the employed pathways, thus yielding support for the hypothesis in two of six comparisons. On current marital status at age 26, Young Underemployed Married Fathers show a lower proportion than Young Married Fathers and Teen Married Fathers, with both of the two possible contrasts providing support (Table 9). On current marital status at 37, Young Underemployed Single Fathers have a lower proportion of being currently married than the three fully employed pathways, and Young Underemployed Married Fathers have a lower proportion than one fully employed pathway, Young Married Fathers; thus, four of six comparisons support the hypothesis. For incarceration by age 26, Young Underemployed Married Fathers and Young Later-Marrying Fathers both have a higher proportion of incarceration than either Young Married Fathers or Teen Unmarried Fathers, and Young Underemployed Single Fathers have a higher proportion of incarceration than all four comparison pathways, with five of six contrasts supporting the hypothesis.

In summary, the hypothesis that early fatherhood pathways in which most men were not fully employed around the time of the birth are associated with later-life disadvantage is consistently confirmed for earnings at both ages, for earnings per child at both ages, and for current marital status at age 26. The hypothesis receives only limited support for educational attainment at age 37, but stronger support for current marital status at age 37 and for incarceration by age 26.

Decreasing Disadvantage Associated With Early Fatherhood Over the Life Course (Hypothesis 5)

We hypothesize that the later-life disadvantage associated with early fatherhood decreases over the life course, as it does for early motherhood. We evaluate this hypothesis by comparing results for the subset of later-life correlates that are available at both age 26 and age 37: earnings, earnings per child, and current marital status. For earnings, the disparities between each early fatherhood pathway and On-Time On-Sequence Fathers (calculated from Table 9) are greater at age 37 than at age 26: $2,937 vs. $1,041 for Young Married Fathers; $9,511 vs. $1,468 for Teen Married Fathers; $29,465 vs. $12,139 for Young Underemployed Married Fathers; $31,220 vs. $10,296 for Young Underemployed Single Fathers; and $16,184 vs. $6,845 for Young Later-Marrying Fathers. The same holds true for earnings per child: $4,091 vs. $2,838 for Young Married Fathers; $7,123 vs. $5,700 for Teen Married Fathers; $16,060 vs. $12,520 for Young Underemployed Married Fathers; $15,176 vs. $10,080 for Young Underemployed Single Fathers; and $7,941 vs. $5,135 for Young Later-Marrying Fathers.⁶

Compared with On-Time On-Sequence Fathers at ages 26 and 37, the odds of being currently married at ages 26 and 37 does not differ markedly for Young Married Fathers, Teen Married Fathers, and Young Underemployed Married Fathers. The difference in odds of current marriage at ages 26 and 37 are, respectively, 0.00 versus −0.02 for Young Married Fathers, –0.20 versus −0.20 for Teen Married Fathers, and −0.43 versus −0.27 for Young Underemployed Married Fathers.

Overall, for earnings and earnings per child, the hypothesis that the later-life disadvantage associated with early parenthood declines over the life course for men as it does for women is not only disconfirmed, but relative disadvantage actually increases. The hypothesis is disconfirmed for current marital status.

Discussion

Men take varying pathways to adulthood characterized by the timing and sequencing of fatherhood, marital, and employment transitions. For NLSY79 men, this study empirically identified five pathways to adulthood that include early fatherhood, pathways varying in their timing (teens vs. early 20s) as well as in their marital and employment context. In analyzing correlates of these early fatherhood pathways, we utilized a life course conceptual model in which varying pathways of transition to adulthood are linked to social origins and influence later-life circumstances net of social origins. In addition, the analysis draws on Arnett’s (2000) notion that early transition to parenthood forecloses the opportunity to complete the developmental tasks of emerging adulthood.

Results suggest that for nearly one in ten NLSY79 men, the developmental tasks of emerging adulthood may be precluded by teen parenthood (Teen Married Fathers), and another one-quarter may experience emerging adulthood in only truncated form because of early 20s parenthood (the remaining four early fatherhood pathways). Consistent with Bynner’s (2005) suggestion, our results indicate that the exclusion or foreshortening of emerging adulthood is associated with less privileged social backgrounds.

The analysis also provides support for Arnett’s argument that the foreshortening of emerging adulthood associated with early parenthood has negative sequelae (Hypothesis 1): disadvantage relative to On-Time On-Sequence Fathers on at least some later-life circumstances through age 37 is evident for all early fatherhood pathways. Further, for the three pathways in which early fathers were predominantly not married and/or not fully employed at the birth (Young Underemployed Married Fathers, Young Underemployed Single Fathers, and Young Later-Marrying Fathers), disadvantage is manifest on every later-life measure examined here. Although early fatherhood is associated with less privileged backgrounds, these backgrounds are not sufficient to account for the disadvantages shown in later life by early fathers.

We hypothesized that fatherhood as a teen is associated with greater later disadvantage than fatherhood during the early 20s, on the grounds that the earlier fatherhood occurs, the greater the truncation of emerging adulthood (Hypothesis 2). The comparison of pathways characterized by marriage and full-time employment at the birth, and differing only on age at fatherhood (Teen Married Fathers and Young Married Fathers), provided a direct test. Greater relative disadvantage accrues to teen fathers than early 20s fathers for earnings at age 37, earnings per child at ages 26 and 37, educational attainment, and marital status at age 37. In past research, Nock (1998) found more negative associations with later-life circumstances for first births occurring at ages 20–25 than at ages 12–19. However, Nock’s analysis concerned how later-life correlates varied by age at birth for births outside marriage, whereas our results pertain to how these correlates vary by age at birth for births within marriage and full-time employment.

Concerning the role of marriage itself, our results partially confirm the hypothesis that early births are associated with less negative outcomes when they occur in marital rather than nonmarital contexts (Hypothesis 3). Contrasting the two pathways with high rates of full-time employment and similar age at birth, the pathway “married at birth” exhibits higher earnings at ages 26 and 37 and less incarceration history than the pathway “unmarried at birth.” When the two pathways with low rates of employment and comparable age at birth are compared, early fathers in the pathway including married at birth complete more education, are more often married at age 37, and less often have incarceration history than those in the pathway unmarried at birth. These results on the protective effects of marriage are stronger than those found in Sigle-Rushton’s (2003) research.

Our and Sigle-Rushton’s findings that early marriage has any protective effect might seem discrepant from prior literature suggesting that early marriages are less stable than later ones (Moore and Waite 1981). Two points can be made, however. First, in our analyses, early marriage is associated with greater marital stability, particularly when comparing early fatherhood pathways with low rates of full-time employment at birth but not when comparing pathways with high rates. In addition, marital stability is operationalized here by current marital status at age 37, not the stability of first marriage.

Second and more generally, we made a different comparison than previous research on early marriage did. We contrasted the earnings, education, and marital outcomes for men associated with the joint pattern of early marriage and fatherhood, with the outcomes associated with early nonmarital unions and fatherhood. Prior research, however, contrasted the later-life circumstances linked to early marriage with those linked to later marriage. Put succinctly, relative to later marriage, early marriage may pose risks. Compared with early nonmarital fatherhood, however, the potential negative effects of early fatherhood on men’s later outcomes may be buffered to some degree. One reason is that in the context of early parenthood, early marriage may lead to increased social support from the partner, kin, and others.

In an additional comparison with past research on the role of marriage in the later consequences of early fatherhood, our results are consistent with Nock’s (1998) proposition that the negative effects of nonmarital early fatherhood are partially mediated by its making subsequent marriage less likely. In our analysis, we compared two early fatherhood pathways characterized by nonmarital births: one in which later marriage occurred rarely, and the other in which later marriage was common. Although these two pathways differ somewhat in social background and other characteristics at age 26, comparing their later-life circumstances provides some support for the mediating role of subsequent marriage: the later-marrying pathway showed significantly higher earnings and earnings per child at age 37.

We hypothesized that being employed around the time of the birth is protective (Hypothesis 4). Strong support is provided by results for earnings and earnings per child at both ages 26 and 37 and for marriage at age 26, and moderate support is provided for marriage at age 37 and incarceration. The role of employment status at birth on the later-life circumstances of early fathers has not heretofore been investigated, and our findings indicate a new direction for future research.

Finally, based on prior findings in women, we hypothesized that the disadvantage associated with young fatherhood decreases over the life course. Disadvantage in marital status changed little, and results were actually opposite to those predicted for earnings measures. Not only men do not recover over time, but the economic disadvantages associated with early fatherhood accumulate. This result poses an interpretive challenge. In our view, there are important differences in the way that parenthood affects labor force attachment for men compared with women. Women of all ages are likely to reduce their work effort at first birth and while their children are very young, and then gradually increase their work effort as their children grow older (Glauber 2007). If women in their 30s who were young mothers and on-time/late mothers are compared, for example, it is possible that differences in economic later-life correlates (labor supply, wages, work effort) will be quite small. This is because the children of young mothers are, on average, older than the children of on-time/late mothers and thus create less conflict between women’s work and family roles.⁷ For men, however, parenthood typically intensifies work effort (Glauber 2008). It is possible that early fatherhood limits the acquisition of human capital either by interrupting education or by preventing fathers from putting in extra effort (overtime, residential moves) to acquire human capital at work. The restriction of early human capital acquisition may limit opportunities for later human capital acquisition such that disadvantage accumulates over the life course.

The study has three principal limitations. First, the associations we observed between early fatherhood pathways and later-life circumstances cannot be interpreted as causal. Although differential selection into early fatherhood pathways is addressed to some degree by controlling for demographic background characteristics, unmeasured selection factors may nonetheless account for the associations found. Future research is needed, using additional procedures to address selection and to justify causal inference. Second, our results are specific to the particular LCGA model specification we employed. Some potentially important status variables are not available in NLSY79 (such as cohabitation), and LCGA is limited in any event in the number of role status variables that it can employ to derive classes. LCGA modeling using other variables, such as cohabitation or current residence with a child, might yield somewhat different pathways. Third, our results concern relationships between early fatherhood and later-life circumstances in the 1958–1965 birth cohort included in NLSY79. In light of the changing historical context of fertility, the associations observed in that cohort cannot necessarily be generalized to more recent cohorts.

These limitations notwithstanding, the study contributes to research on emerging adulthood. With early parenthood interpreted as precluding or foreclosing this life stage, analyses assessed the frequency and concomitants of compromised emerging adulthood. It also adds to the literature on early fatherhood per se by empirically deriving multiple pathways of early fatherhood and investigating their associations with subsequent well-being. In the latter area, our results can be interpreted in two ways. The more optimistic reading is that although some later-life disadvantage is associated with fatherhood occurring in the early 20s, in marriage and in the context of full-time employment, this disadvantage is less than when the father is a teen, unmarried, or not fully employed at the time of birth. The more pessimistic view simply reverses the order of these statements: although early fathers who are not teens and who are married and fully employed at the time of birth show less decrement in well-being later in life, some degree of disadvantage remains. And in all early fatherhood trajectories, relative disadvantage increases over the life course, rather than lessening, as is the case for women.

Acknowledgments

This research was supported by the National Institutes of Health – NICHD: PO1 HDD45610-03. Part of the work reported here was also supported by the Cooperative State Research, Education and Extension Service, U.S. Department of Agriculture, under Project No. ILLU-45-0366 to Joseph H. Pleck. The authors pay special thanks to Stephanie Lanza for her methodological wisdom and invaluable feedback on earlier drafts of the paper.

Notes

References