Social status is defined as the relative position of individuals in a stratified society as characterized by economic capital and human capital (Grusky et al. 2008). Economic capital refers to one’s material resources like income and assets as means to procure further social status (Bourdieu 1986; Oakes and Rossi 2003). Human capital captures one’s knowledge and skills as means to gain social status (Becker 1993; Oakes and Rossi 2003). Status attainment refers to the processes by which initial social status or “origin” (e.g., parents’ social status) subsequently becomes social status over time or “destination” (Schoon 2008). Blau and Duncan’s (1967) original status attainment model examined father-to-son occupational mobility in a variable-oriented approach (e.g., path analysis). The Wisconsin model of status attainment highlighted key social, psychological, and institutional factors in adolescence that mediate the effects of social origins on adult education, occupation and income (Sewell and Hauser, 1975). However, variable-oriented approaches to study these processes are often unable to detect subgroups with different status attainment processes, which are better captured using a person-oriented approach (Bergman and Magnusson 1997).

Status attainment involves intergenerational (social status transferred across generations) and intragenerational (social status acquired within one’s lifetime) processes that trace patterns of social stability and mobility in society (Grusky and Ku 2008). In the early 20^th century, intergenerational processes played a large part in determining one’s social status, such that individuals closely resembled their parents. This “stickiness” of social status can limit opportunities for movement up or down the social ladder (Grusky and Ku 2008). One’s own achievements, however, provide an additional pathway for upward or downward mobility, and both intergenerational and intragenerational processes are often in dynamic tension. For instance, although education has traditionally played an integral role in upward mobility, intergenerational effects of increasing economic inequality have in recent decades led to more stickiness in educational attainment (Campbell et al. 2005). Since the transition to adulthood is a time when social status is in flux, an individual’s experiences of accruing status during this transition should be considered in conjunction with the potential advantages or disadvantages bestowed by the previous generation.

In addition, previous studies often restrict status attainment models to a single dimension of social status and overlook the conceptual differences between types of capitals and nuances within capitals that contribute to one’s social status (Krieger et al. 1997; Oakes and Rossi 2003). Specifically, economic capital plays a different role in determining one’s social status than human capital (Bourdieu 1986; Braveman et al. 2005). Material resources such as income can elevate one’s living conditions and ability to purchase goods that symbolize higher social status. Furthermore, experiences of poverty and wealth signal accumulated disadvantages and advantages that are not captured by a single measure of income. In contrast, human capital, such as education and occupation, represent one’s skills, ability, and intellect that are associated with a particular lifestyle, social network, or environment of higher or lower social status (Bourdieu 1986; Becker 1993). Characteristics such as vocational training or time spent in school or work provide a more complete assessment of the human capital experiences during the transition to adulthood. These distinctions highlight the importance of investigating both economic and human capital, but draw attention to the markers that go beyond income (for economic capital) and education (for human capital) to capture diverse aspects of social status that may be especially important during the transition to adulthood.

The life-course perspective provides an optimal framework for studying status attainment over the transition to adulthood. Its themes of time and timing, trajectories, linked lives, and human agency serve as guiding principles for this study (Elder et al. 2004). Accruing, maintaining, and losing social status are viewed as key developmental processes that fit within the life-course framework. Applying a life-course perspective identifies several limitations of previous research about the evolution of status attainment. First, conventional markers of social status —e.g., education, occupation, income— are sensitive to life stage, making it essential to study time and timing. For example, equating initial social status with parents’ education or income may be appropriate for adolescence, but not young adults, and income among young adults may fluctuate considerably, especially when individuals are still in school, making the timing of educational attainment a key consideration (Furstenberg 2008; Do 2009; Scharoun-Lee et al. 2011). To assess intergenerational (from parents) and intragenerational (from self) mobility, these traditional markers should be captured over repeated time points. In addition, non-traditional markers (e.g., formal government resources, informal family resources) provide a broader assessment of status that may be especially relevant for some subgroups of the population, such as low income families. Furthermore, involvement in school and at work should be considered during this time given that individuals are accruing status at varying rates (e.g., time spent on education may exceed time spent working, or vice versa). For these reasons, the current study uses longitudinal data with a diverse set of social status measures to provide a deeper assessment of the status attainment process during the transition to adulthood.

Second, individual decisions and the timing of early life events can impact future status attainment by forming status attainment trajectories leading to the accumulation of advantages or disadvantages. This embodies the concept of human agency in decision making, which refers to the choices individuals make given the opportunities and constraints in their world (Elder et al. 2004). For example, early entry into adult social roles (e.g., becoming a teen parent) has long-term social status effects and creates a build-up of subsequent transitions, such as early exit from school or early entry into work (Elder et al. 2004). These early transitions have been linked to relatively low subsequent social status (Foster et al. 2008; Johnson and Mollborn 2009). Delayed entry into adulthood (e.g., continued schooling) also has long-term social status effects and leads to a build-up of transitions, such as further dependence on parents, delayed entry into work, postponement of marriage and parenthood (Arnett 2000; Furstenberg et al. 2005). These delayed transitions have been linked to relatively high subsequent social status (Furstenberg et al. 2005). This social bifurcation of the transition to adulthood demonstrates that although individuals are active agents in the status attainment process, the consequences of early decisions have compound effects on future trajectories.

Third, the role of parents is an application of the life-course theme of linked lives that is critical to the transfer of advantages and disadvantages from one generation to the next. The social status and resources of one’s family of origin can aid or deter status attainment. Compared to individuals from advantaged families, individuals from disadvantaged families have fewer resources and capacities to navigate the various transitions of obtaining secondary education, entering the work force, and forming families (Furstenberg 2008). Furthermore, some individuals remain financially dependent on parents while others financially contribute to parents (Settersten et al. 2005). These patterns are evident for both economic and human capital, but with varying returns on one’s social status. For instance, economic capital transfers may provide more immediate effects on status (e.g., money for living), whereas human capital transfers may be more intangible and long-term (e.g., parents’ higher education provides exposure to higher status cultural preferences, dress, speech). Finally, social and economic inequality across parents creates different starting positions for status attainment among offspring. In this manner, the social status of parents impacts that of their sons and daughters, as do the choices these young people make with regard to transitions into adult roles and trajectories of economic and social capital.

Add Health is an ongoing study of a nationally representative sample of 7^th–12^th grade U.S. youth during the 1994–95 school year who were followed into adulthood (Harris et al. 2009). A multistage stratified (e.g., stratified by region, race/ethnicity) cluster design was used to select 80 high schools and 52 feeder schools (had 7^th grade and sent graduates to sampled high school). More than 90,000 students participated in an initial school-based survey (response rate [RR]: 79%). A core in-home sample (n=20,745) was drawn from the school-based sample with oversamples based on ethnicity (Blacks with a college-educated parent, Chinese, Cubans, and Puerto Ricans), genetic (siblings and twins), adoption status and disability (Harris 2011). This core sample then completed a 1–2 hour survey at home (W1; 78.9% RR), and were surveyed again in 1996 (W2; 88.2% RR), 2001–02 (W3; 77.4% RR) and 2008–09 (W4; 80.3% RR). Parents (primarily mothers; 85.4% RR) were interviewed at W1 (Harris et al. 2009).

The current study’s analytic sample is restricted to respondents with W1, W3, and W4 in-home and W1 parent interviews (n=13,034). W2 data are not used given the close proximity to W1 and lack of follow-up with W1 high school seniors. Members of small racial/ethnic groups (n=515) were excluded because there are too few for analysis and the category is too heterogeneous to be meaningful. Cases without a sample weight (i.e., selected outside the sampling frame) were excluded. The final analytic sample is 8,977. Data are weighted and standard errors are adjusted for the complex sample design.

Weighted sample characteristics are presented in Table 1. Gender composition is fairly equal and the majority of respondents self-identified as Whites, with far fewer Blacks and Hispanics, and even fewer Asians. The mean age in adolescence is 15 (range of 12–18), young adulthood is 21 (range of 19–25), and adulthood is 28 (range of 25–31). Due to higher attrition among respondents with lower income and education, the results may be biased towards higher income and education although sample weights adjust for this factor.

This study was approved by the University of California, Los Angeles Human Subjects Protection Committee (IRB #10-001106).

This study applies a person-oriented approach of latent class analysis (LCA) for analyzing longitudinal data. LCA is a technique to identify substantively meaningful subgroups within the larger population (Nyland et al. 2007; Collins and Lanza 2010). This approach allows researchers to go beyond identifying the effect of a single variable on an outcome to understand patterns of a set of variables as a whole on an outcome (Bergman and Magnusson 1997). It is especially beneficial for studying complex constructs such as social status. LCA is a non-parametric statistical technique that assumes that patterns among a set of observed variables are explained by an unmeasured latent variable with discrete classes where classes are groups with similar values on the variable set (Lazarsfeld and Henry 1968; McCutcheon 1987; Collins and Lanza 2010). LCA is similar to factor analysis except that the resulting latent variable is categorical with a multinomial distribution (Collins and Lanza 2010). Observed variables are assumed to be ‘locally independent’ within each class such that any intraclass correlations or multicollinearity between variables are not an issue in LCA whereas they are common and problematic in variable-oriented frameworks (Clogg 1995, Hagenaars and McCutcheon 2002). Respondents are assumed to belong to only one class (Lanza et al. 2007). In this study, LCA was used to identify (1) the optimal number of latent classes, and (2) class size/characteristics.

A series of LCA models was tested specifying 1 to 6 classes. Model selection was based on model fit statistics (e.g., Akaike Information Criteria [AIC] and Bayesian Information Criterion [BIC], and sample size adjusted Lo-Mendell-Rubin [LMR] likelihood ratio test (Schwarz 1978; Sclove 1987; Lo et al. 2001; Nylund et al. 2007). Other model fit criteria were high class homogeneity (degree that individuals in one class follow the same observed response pattern, implying that the pattern is highly characteristic of the class) and high class separation (degree of distinction between latent classes on item-response patterns) (Collins and Lanza 2010).

Once the best-fit model was identified, additional LCA models were estimated adding covariates of demographic characteristics and transition to adulthood markers. The statistical significance of the association between class membership and each covariate was assessed with a multinomial logistic regression models; each covariate was statistically significant at a p-value of less than 0.05 using log-likelihood ratio tests. In sensitivity analyses, respondents were assigned to a class based on their maximum predicted probability of class membership, and covariates were compared across groups using conventional bivariate statistical techniques, specifically chi-square tests for categorical covariates and F-tests for continuous covariates. These findings aligned well with the LCA results. For ease of interpretation, the bivariate relationships are presented as proportions and means using these conventional methods.

Descriptive statistics were conducted in Stata version 12.0 (StataCorp 2011), and LCA was conducted in Mplus version 6.11 (Muthen and Muthen 1998–2011). Survey procedures in Stata and Mplus corrected for unequal probability of selection, attrition, and the complex sample design (Harris et al. 2009). Missing data in social status measures were handled using full information maximum likelihood (FIML) estimation. FIML assumes that data are missing at random (MAR); however, even if the MAR condition is not completely satisfied, FIML estimation can reduce bias while maximizing the number of observations (Arbuckle 1996; Wothke 2000).

With each additional class, the log-likelihood, AIC and BIC values first decreased and then leveled off between 3- and 6-class solutions [Appendix A]. LMR test statistics also supported 3- to 6- class solutions. However, there were trivial class sizes (<5%) in the 5- and 6-class solutions, and item-response probabilities showed poor class homogeneity (where a dominant pattern was not apparent within the 5- or 6-class solution) and poor separation (in which these small classes did not differ substantially from the other larger classes). These considerations point to a 4-class LCA model as having the most distinct groupings of economic capital.

Table 2 presents the latent class prevalences for the 4-class model, and the conditional response probabilities and means for each observed economic capital indicator by class. The table also shows the mean values and proportions for the total sample. Profiles of life-course economic capital are characterized as Class 1— Persistently disadvantaged (18.1%); Class 2— Upwardly mobile (21.1%); Class 3 —Downwardly mobile (27.8%); and Class 4— Persistently advantaged (33.0%). Stability in economic capital is about as common as mobility. More than half of respondents were in groups with high economic capital in adolescence; however, only one in two of these persons maintained that high level in adulthood.

The persistently disadvantaged group was characterized by accumulated disadvantages over the life course. In adolescence, the mean household income was substantially lower than the total sample. Personal income grew slightly from young adulthood to adulthood. Public assistance and not having health insurance were consistently high at each wave, and home ownership was low relative to the sample overall. Despite receiving family financial support, this group was the second highest in giving financial help to their families in adulthood.

The upwardly mobile group possessed characteristics of increasing economic capital from adolescence to adulthood. In adolescence, this group had the second lowest household income among the four classes. By young adulthood, this group represented the highest mean income, and by adulthood, the second highest. Financial strain, public assistance, and not having health insurance declined over time. By adulthood, almost half of respondents owned a home. This group received little family financial support, and was most likely to provide financial support to family in adulthood.

The downwardly mobile group had the second highest mean household income in adolescence. However, by young adulthood and adulthood, personal incomes were the second lowest of all classes, on average. Economic hardships gradually increased over time and only one-quarter owned a home in adulthood. A majority received family financial support in young adulthood and adulthood with little return to family in adulthood.

The persistently advantaged group was characterized by an economic environment of high incomes and little economic hardship over this period of the life course. In adolescence, the mean household income was the highest of any group and personal income grew substantially from young adulthood to adulthood. Over half owned a home in adulthood. Family financial support in young adulthood was highest across all groups, but was lowest of all groups by adulthood. This group was also likely to receive financial support for their home purchase, but provided little support back to family in adulthood.

Table 3 presents demographic characteristics for each economic capital group as defined by the highest predicted probabilities of class membership. The persistently disadvantaged had more females, Blacks, and Hispanics than the persistently advantaged, which had more males, Whites, and Asians. Males and Hispanics made up higher proportions of the upwardly mobile group than the downwardly mobile group, which had more Whites and slightly younger persons. The downwardly mobile had the youngest mean age in adulthood. However, the mean age for all groups is 27.9, which signals a young sample.

Turning to the two types of capital, Table 3 also presents education as a representative human capital indicator for each economic capital group. Parents’ and respondent’s education were very similar within the persistently advantaged and disadvantaged groups. When compared to parents, respondent’s education levels were lower for the downwardly mobile group but higher for the upwardly mobile group. When examining across economic capital groups, the downwardly mobile had higher education levels than the upwardly mobile.

For the transition to adulthood markers, also shown in Table 3, the persistently disadvantaged had early adoption of adult social roles, being most likely to have left the parental home and school, and second most likely to be married by young adulthood. Having the highest proportion of those who have children in young adulthood and adulthood suggests early childbearing. In contrast, the persistently advantaged had patterns of delayed entry into adult roles; in young adulthood, a majority was still in school, and few had married or had children. By adulthood, however, the persistently advantaged had higher proportions of marriage than the persistently disadvantaged. The upwardly mobile resembled the persistently disadvantaged with respect to early adoption of adult roles, in particular full-time work status; marriage in this group, however, remained high in adulthood. The downwardly mobile group was similar to the persistently advantaged except with higher proportions living at home in young adulthood and being in school in adulthood. While they had a high proportion of never married compared to the persistently advantaged, the downwardly mobile also had a higher proportion of being a parent.

The log-likelihood, AIC and BIC values, and LMR statistic supported four or higher class solutions for LCA models [Appendix B]. However, the 4-class and 5-class solutions had large clusters in one class, suggesting that a large proportion of respondents shared similar human capital characteristics. Although class prevalences were equally distributed in a 6-class solution, class separation was poor in that patterns were similar from one class to another class. In the 5-class solution, class separation and homogeneity were good. Therefore, a 5-class model was selected as the best-fitting model. Table 4 shows the class prevalences and probabilities for this model: Class 1 —Persistently low (10.6%); Class 2— Upward mobility with early entry to work (42.1%); Class 3 —Upward mobility with continued schooling (15.0%); Class 4— Downward mobility with early entry to work (16.4%); and Class 5 —Persistently high (16.0%). Similar to economic capital, the persistently low made up the smallest human capital group. A plurality of respondents falls into the upward mobility with early entry into work group, while the remaining three groups are similar in size.

Class 1, the persistently low, accumulated disadvantages in human capital over this period of the life course. In young adulthood, a majority had not completed high school, and very few were in school, suggesting that most of this group had attained their maximum education. Adult education ranged from less than high school to high school degree or GED. Among parents who worked, occupations were in sales/service (for mothers) and manual (for fathers). Typical adult occupations were manual or sales/service.

Class 2 was similar to Class 1 in adolescence; however, several key differences signal upward mobility during young adulthood. First, 95.8% of respondents in this class had a high school degree in young adulthood, compared to 27.7% in Class 1. Adult education levels were between a high school degree and some college, which was higher than their parents but lower than other groups, suggesting relatively early exit from school and entry into work. Second, this group worked more in adolescence and young adulthood compared to all other groups. This group had the highest proportion who received vocational training. The modal adult occupations include sales/service and manual.

Class 3 also had an upward trajectory; the key difference with Class 2 is that a majority continued schooling during young adulthood. Intergenerational gains are evident when comparing parents’ education (between high school degree and some college) and respondent’s education (between college degree and graduate school). An upward path is also evident with occupation. Mothers worked primarily in sales/services and fathers in manual sectors. Respondents, however, worked in professional or managerial, other professional, and sales/service sectors.

In contrast, Class 4 shows downward mobility. Their parents had the second highest education levels of some college. The majority of respondents, meanwhile, had a high school degree by young adulthood, but by adulthood, the mean education level was less than both parents. In addition to early work in young adulthood, there was also continued schooling in young adulthood and adulthood. Therefore, the downward pattern may have the potential to reverse later in life, but the combination of working may have made educational attainment slow for this group. Mother’s occupation was high while father’s occupation had a wide variation. The modal respondent occupation is sales/service. \

Class 5 represents accumulated advantages in human capital. The majority had completed high school by young adulthood, and a large proportion continued schooling in young adulthood. The average adult education was between a college degree and graduate school. Average work hours were the lowest in young adulthood, but highest by adulthood. Parents’ and respondent’s occupation were high at a professional or managerial level.

Table 5 shows the relationships between demographics and human capital classes. Similar to economic capital, the persistently low had more Blacks and Hispanics than the persistently high, which had more Whites and Asians. Gender patterns were reversed for human capital; the persistently low had more males while the persistently high had more females. Upward with continued schooling group comprised of more females than males. The upward with early work group resembled the persistently low with more Blacks and Hispanics than Whites, but with equal gender composition. The downward with early work group included more males and Whites than females and non-Whites. Groups that continued schooling (i.e., upward and persistently high groups) in young adulthood were characterized by younger respondents, whereas the groups that entered work early were older respondents.

To compare the two forms of capital, Table 5 presents income as an indicator of economic capital across the three life stages. The correspondence between economic and human capitals is most evident for the two extremes: the persistently low group has the lowest incomes at all three times and the persistently high have the highest incomes except for young adulthood when they were in school. Income in young adulthood is highest instead among the two groups who entered the workforce early. Furthermore, by adulthood, this pattern reverses, and a positive association of higher human capital and income is evident among the upward with continued school and persistently high groups. Thus the notion of the positive association between economic and human capitals is clearly challenged by the time-dependent effects of continued school on income during this period.

For the transition to adulthood markers, human capital tends to correspond to trends of being in school, marriage and parenthood, but not with living with parents. The persistently low human capital had patterns of earlier adult roles (e.g., not in school, full-time work, marriage in young adulthood, parental status by young adulthood and adulthood) relative to the persistently high human capital (e.g., currently in school, fewer with children). Yet a higher proportion of persistently low lived with parents than the persistently high. The upward with early work and downward with early work had patterns similar to the persistently low, which signal earlier adult roles. The upward with continued schooling exhibited characteristics of delayed adult roles that were comparable to the persistently high.