Growing up Faster, Feeling Older: Hardship in Childhood and Adolescence

Abstract

We examine whether hardship while growing up shapes subjective age identity, as well as three types of experiences through which it may occur. Drawing on data from the National Longitudinal Study of Adolescent Health, we find that hardship in several domains during childhood and adolescence is associated with feeling relatively older and self-identifying as an adult in the late teens and twenties. Specifically, young people who as adolescents felt unsafe in their schools or neighborhoods, witnessed or were victims of violence, had fewer economic resources in the household, and lived in certain family structures, reported older subjective ages (by one or both measures). We find no evidence that hardship’s association with subjective age is mediated by work responsibilities in adolescence or by anticipating a very curtailed life span, but entering adult roles earlier mediates or partially mediates many of these relationships.

Sociologists recognize age as one of the most fundamental categories organizing social life (Riley, Foner, and Waring 1988; Settersten and Mayer 1997). However subjective age is equally or more important than chronological age in many processes (e.g., Bowling et. al 2005; Montepare and Lachman 1989; Neugarten and Hagestad 1976). Subjective age captures self-perceptions of one’s age, often expressed in terms of relative age, such as how old one feels compared to others of the same chronological age, or the age group with which one identifies (Settersten and Mayer 1997). Social science has recently witnessed a growing interest in the subjective side of aging during the early life course, including pseudomaturity in adolescence (e.g., Galambos et al. 1999; Greenberger and Steinberg 1986) and young people’s understanding of what it means to grow up and be an adult (e.g., Arnett 2000; Shanahan et al. 2005; Macmillan 2007).

In addition to interest in the cultural meaning attached to maturity and adult status, some of this newer research points to important variation in the pace of growing up. The idea that some children or adolescents grow up more quickly than others is not entirely new, however. The literatures on divorce and economic disadvantage both reveal themes of accelerated young lives. For example, Weiss (1979) suggested in the 1970s that divorce makes children “grow up a little faster,” and consistently, recent research indicates that young people in their late teens and twenties from married biological-parent families feel younger and are less likely to consider themselves adults than young people with other family structures (Benson and Furstenberg 2007; Johnson, Berg, and Sirotizki 2007a, b). Similarly, consistent with Elder’s (1974) characterization of adolescents’ lives in economically pressed families during the Great Depression, recent ethnographic accounts of poor inner-city youth suggest children and teenagers age into adulthood very quickly in these settings (Burton, Obeidallah, and Allison 1996; Kotlowitz 1991). Family disruption and poverty, along with other conditions of deprivation or stress like feeling unsafe in daily life, represent key hardships implicated in subjective aging during the early life course.

We build on the foundation provided by these studies to better understand who grows up more quickly and why. We argue that hardships are linked to older subjective ages because they affect three bases of age identities, all of which are grounded in age-normative understandings of the life course: earlier timing of entry into age-graded roles, earlier assumption of responsibilities, and expectations for a shorter life span. We draw on nationally representative data from eighteen to twenty-six year olds in the National Longitudinal Study of Adolescent Health to examine whether hardship experienced in several domains during childhood and adolescence is related to the sense of being older than one’s age peers and, for the youngest in our sample, identifying as an adult. We also examine whether these relationships operate through three types of mediating experiences consistent with our argument. Because identity shapes behavioral choice (Stryker and Burke 2000), and this stage in the life course involves critical decisions affecting life trajectories of education, work, and family, it is important to understand influences on age identity in young people. Our research helps build an understanding of how the social contexts in which individuals grow up shape their identity. Our findings also contribute to important debates in the life-course tradition involving the import of adult role transitions and the nature of age-norms in contemporary society.

THE LIFE COURSE AND AGE IDENTITIES

As other life-course scholars have also done, we view subjective age identity, and identities more broadly, as part of the self-concept (e.g., George, Mutran, and Pennybacker 1980; Kaufman and Elder 2003). As such, it is part of one’s perception of oneself as an object, in this case a subjective assessment of one’s chronological age. The self-concept develops in interactions with others, particularly significant others, and is influenced by cultural, social structural, and historical forces (Gecas and Burke 1995). Roles and other structural features of societies provide expectations that guide behavior in interaction and provide external, socially agreed-upon standards for comparison and evaluation.

Identity theory, arguably the dominant perspective on self and identity in sociological social psychology, specifically links the self to social structure via roles. Indeed, the theory defines identities as internalized role expectations, or, alternatively, the meanings people attach to the many roles they play (Stryker and Burke 2000). MacMillan (2007) notes that this dominant conceptualization, and subsequent research foci, have left several gaps in knowledge of identities to date. First, he notes that many of the roles in which identities are based are age-graded, and that normative timetables for the life course indicate the appropriate timing of major role transitions. These social aspects of aging are a product of culture, varying historically and across subgroups, making up an important part of the “meanings” people attach to roles. As Settersten and Mayer (1997) observe, age, along with sex, is the primary source of variation in social and cultural expectations for life experiences and roles. Societies are age-structured, and Macmillan’s critique reflects the underacknowledged role of age-related expectations and meanings involved in identity processes. In addition, Macmillan (2007) points out that roles “aggregate to define particular life stages and serve in an almost taken for granted manner as the master statuses of one’s identity” (2007:15). Internalized expectations for what it means to be a child, adolescent, or adult are therefore crucial to identity.

Among these expectations are age norms that prescribe appropriate behavior for someone in a particular life stage. The contemporary Western view of childhood has been called a “sentimental perspective” for its understanding of children as priceless, innocent, and in need of protection (Best 1990). This view idealizes happy, carefree childhoods. Cultural constructions of childhood thus entail expectations of both innocence and freedom from adult responsibilities, roles, or burdens, and these notions take the form of age norms. Childhood, and to a good extent adolescence, are also organized around a norm of dependence (Macmillan 2007). Again, this limits adolescents’ responsibilities and puts parameters on adolescents’ relationships to parents or those on whom they depend.

Although movement in and out of roles across the life course has been acknowledged as important (e.g., Wells and Stryker 1988), very little work (inside or outside of identity theory) takes on these life-course themes and their implications for identity. We gain leverage in understanding age identities by taking the role-based conceptualization of identities from identity theory, but also by adding consideration of these internalized age-related cultural expectations.¹ Specifically, we argue that violating age norms for childhood and adolescence challenges the child or adolescent identity and accelerates subjective aging. A young person’s understanding of herself as relatively older (or younger) emerges as she perceives that her experience violates age norms for someone her age. We focus on three (non-exhaustive) types of age-norm violations through which young people develop older age identities: assuming adult social roles, accumulating the responsibilities and performing the tasks of an adult, and anticipating a highly curtailed life span (i.e., an early death). We begin by discussing these three types of experiences and then complete the circle by returning to articulate how specific forms of hardship accelerate subjective aging through them.

Behavior and age identity are likely reciprocally related across the full life course. While we focus on the effects of hardship on age identity through a process in which engaging in non-normatively older behaviors lead to older age identities, it is possible that there is actually a bidirectional relationship involved whereby those with older age identities also engage in behaviors considered normative for older ages. Because our primary interest is in building an understanding of how hardship fosters growing up faster, we focus on the ways behavior and experience can affect identity in this study.

Earlier Assumption of Age-graded Roles

Consistent with the conceptualization of identities in identity theory, we expect that identities are based in social roles. Importantly, many roles are age-graded, and studies confirm that age identities are grounded in the age-graded social roles that make up the life course, including parent, grandparent, and retiree (George 1990; Logan, Ward, and Spitze 1992; Neugarten 1977). Age-related cultural meaning attached to roles provides a reference point in understanding one’s age. Being a grandparent is something “older” people do. In fact, transitioning to roles such as grandparent earlier than others is associated with older subjective ages (Kaufman and Elder 2003).

With respect to the adolescent and young adult life course, earlier entry into the “adult” roles of marriage, parenthood, full-time work, and living independently from parents should, accordingly, foster older subjective ages. Compared to others of one’s chronological age, those who have transitioned into these roles should feel older, and recent studies confirm this expectation (Benson and Furstenberg 2007; Johnson et al. 2007a; Foster, Hagan, and Brooks-Gunn 2008). Occupying these roles is often inconsistent with an understanding of oneself as a child or adolescent.

The meaning of adult social roles in the life course is subject to much debate, however, particularly given growing variability in the timing and configuration of these roles (Fussell and Furstenberg 2005). Normative timetables for adult role transitions are historically variable and may also be weaker now than they have been for some time. Many scholars argue that the life course has become more individualized (Buchmann 1989), with fewer normative and institutional constraints and more opportunities to chart one’s own path. Overall delays in family formation have been combined with frequent movement back and forth between school and work and between living independently and in the parental home (Shanahan 2000).

In this context, Arnett (2000) has argued that role transitions are no longer used as criteria for defining adulthood, and his studies show that very few people of any age indicate that traditional role transitions like marriage and parenthood are necessary for young people to achieve adult status. Instead, he argues, it is the development of adult personal qualities like responsibility that now define adulthood in contemporary Western societies. Yet, research by Furstenberg et al. (2003) indicates that key adult roles (e.g., completing school, achieving financial independence, working full-time, and establishing an independent household) are viewed as important markers in considering someone an adult. Additionally, evidence of age-based expectations for transitions to adulthood exists in violations that occur at very late or very early ages, as demonstrated by Settersten (1998) in failing to leave home and Mollborn (2007) in teenage pregnancy. Thus we expect that young people who have made transitions to adult social roles will report older subjective age identities.

Earlier Assumption of Responsibilities

Those who have argued for the continued relevance of adult role transitions in marking adulthood have also endorsed the idea that adulthood is marked by qualities of character such as responsibility (Shanahan et al. 2005; Johnson et al. 2007a). Indeed, responsibility for self (Arnett 2000), or self and others (Pallas 2007; Andrew et al. 2007; Macmillan 2007), is emerging as one of the key themes in current research on what it means to grow up and become an adult.

When children and adolescents take on a level of responsibility that is understood as more typical for those at older ages, they are likely to see themselves as older than their age peers. And in taking on adult responsibilities, young people begin to feel like adults. Certain responsibilities during childhood and adolescence can be expected to challenge cultural understandings of these life stages as relatively carefree and dependent upon others. These could include taking on significant responsibility in child or elder care or in household management and housework. It could also include taking on adult-like paid work responsibilities, becoming financially independent, or making financial contributions to support others. Though there is little extant research on the subject, Benson and Furstenberg (2007) show levels of both household and financial responsibilities in late adolescence are associated with identifying as an adult at age 21 in a predominantly working poor and working-class sample. Thus we expect that young people who held greater responsibilities as adolescents will report older subjective age identities.

Perceived Location in the Life Span

Another basis for age identities may lie in perceptions of where one is located in the life span. Research finds that subjective aging during midlife involves facing one’s own mortality and a reorientation toward time in which one begins to think of one’s life as time left rather than time since birth (Neugarten 1968). This suggests that perhaps subjective age can become a measure of nearness to the end of life rather than time since it began. Those who look toward the future and see a sharply curtailed life course may feel older than others their age.

Most research on this process has focused on midlife and beyond. Yet any significant sense of being near the end of one’s life during childhood and adolescence violates dominant normative timetables that assume living into old age and should also accelerate subjective aging. Those facing life-threatening risks in childhood and adolescence do report feeling older and that they grew up quickly (Burton et al. 1996; Parry and Chesler 2005). Thus we expect that young people who anticipate a sharply curtailed life span will report older subjective age identities.

HARDSHIP AND SUBJECTIVE AGE

Having identified several bases for understanding accelerated subjective aging during childhood and adolescence, we now return to the question of why hardship during this period in the life course might be linked to growing up faster in terms of subjective age. We argue that hardships foster older age identities because they tend to propel children and adolescents toward experiences that challenge cultural norms about childhood and adolescence. In this section we discuss three forms of hardship during the early life course with a focus on how older subjective ages develop through the processes we have identified above.

By considering multiple forms of hardship, we can assess common and disparate experiences to the benefit of theory development. The three hardships on which we focus, family disruption, poverty, and living with violence and fear for one’s safety, are also interrelated. The highest rate of poverty, for example, occurs among single-mother families (U.S. Census Bureau 2007), and rates of witnessing or experiencing violence during adolescence are inversely related to household income (Crouch et al. 2000). Our simultaneous consideration allows us to evaluate whether they also have independent effects.

Each hardship can be understood in a rich literature to which we cannot do justice in this brief discussion. For example, Amato’s (2000) review of research on divorce shows how both the severity and duration of consequences for children are affected by children’s own coping skills, the degree of social support they receive, parental conflict and adjustment, as well as factors of gender, age, and custody arrangements. Similarly, what poverty means for children’s lives depends on its duration and timing in the life course as well as parenting strategies and families’ ecological locations with respect to services and the resources of schools and neighborhoods (Seccombe 2000). Although mindful of this complexity, we focus our discussion rather narrowly on how these hardships can be related to the earlier acquisition of responsibilities, earlier transitions to adult roles, and anticipating a sharply curtailed life span.

Family Structure

Family structure in childhood and adolescence may affect subjective aging because it is linked to responsibilities in the family as well as the timing of entry into adult roles. In his classic piece that first suggested that some youth grow up faster, Weiss (1979:101) argued that children in single-parent homes grow up faster because they often became “junior partners” in household management. This often meant doing more housework and taking on higher levels of responsibility compared to children in married-parent families. Consistent with this argument, studies have shown that children and adolescents in single-parent families do more housework on average than those in married biological-parent families, with those in stepfamilies falling in between (Cooney and Mortimer 2000; Goldscheider and Waite 1991). If children and adolescents in certain family structures perceive that they do housework and have other responsibilities that are more normative for older ages, they may develop older age identities.

Family structure is also related to the average timing of certain adult role transitions, including a tendency to leave home earlier among children in a variety of family patterns other than those headed by married biological parents (e.g., Cooney and Mortimer 2000; Aquilino 1991). On average, young people from single-parent and step-parent families leave school earlier, experience their first sex earlier, are more likely to experience a pre-marital birth, and form unions earlier than those from intact married families (e.g., Musick and Bumpass 1999). Role performance in “adult” statuses is inconsistent with cultural expectations for childhood and adolescence. Through participating in adult role relationships and being treated as an adult, the young person begins to feel more like an adult.

Poverty and Low Socioeconomic Status

Economic hardship is also likely to accelerate subjective aging through the earlier assumption of responsibilities and earlier movement into adult roles. For many youth, economic hardship involves what Elder ([1974]1999) calls a “downward extension of adultlike experience,” including increased contributions to the family via household or paid labor. Young people from lower income families also take on heavier work loads outside the home, working longer hours at paid jobs (Mortimer 2003). Middle-class youth are also often employed, but they are less likely to work at high intensities of 20 hours per week or more and less likely to work in higher earning, more stressful, adult-like jobs (Mortimer 2003). Poor youth are also more likely to give part of their wages to family members or pay for expenses that parents would cover under better economic circumstances (Newman 1999). Shouldering a level of financial responsibility that is more normative for older ages, and more consistent with conceptions of adulthood, may foster a sense of being older, or grown up, at younger ages.

Burton and colleagues’ (1996) research on poor, inner-city, minority families illustrates this process well. They note that at young ages, children and adolescents are often treated as adults in their families. One young man from this study, clearly conveying the adult-like level of responsibility he shoulders, explains how inconsistent the role expectations he faces in school are from his family experience:

Sometimes I just don’t believe how this school operates and thinks about us. Here I am a grown man. I take care of my mother and have raised my sisters. Then I come here and this know-nothing teacher treats me like I’m some dumb kid with no responsibilities. I am so frustrated. They are trying to make me something that I am not. Don’t they understand I am a man and I been a man longer than they been a woman?

Burton (1990) argues that among very poor minorities, an accelerated timetable for the life course has become normative in response to environmental constraints. This suggests that these young people age faster, and feel adult earlier, than other young people. Yet, as evident in this young man’s frustration, they still bump up against society’s dominant age norms in their daily life. Larger cultural conceptions of childhood and adolescence are embedded into relationships at schools, are portrayed in mass media, and so on. Any disjuncture between local and dominant timetables could heighten the feeling of being older.

In addition to their higher level of responsibilities in their natal families, youth from lower socioeconomic backgrounds also experience more accelerated transitions into adult roles (Buchmann 1989; Goldscheider and Goldscheider 1999). Normative timetables seem to reflect this behavioral pattern. Less educated and less affluent people expect that individuals should move through the sociodemographic transitions of adulthood, especially marriage and parenthood, at earlier ages than do others (Furstenberg et al. 2003; Settersten 2003). This may occur in part because socioeconomic status is positively associated with life expectancy. Geronimus (1996), for example, argues that low-income black women face accelerated physical aging and shorter life spans compared to middle-class and affluent women, even during the young adult years, such that infant health outcomes are better when mothers give birth in their teens compared to older ages. But we do not expect that low socioeconomic status affects subjective aging through anticipating a highly curtailed life span (i.e., expecting death in adolescence or young adulthood) in the way we do for the final hardship we consider, experiencing violence and lacking safety.

Safety and Violence

Violence and lack of safety in one’s daily life can accelerate subjective aging by altering perceptions of where one is in the life span and by pushing young people into more adult-like responsibilities. When young people witness or experience violence and fear for their safety in their daily lives, it can raise the possibility (and sometimes the reality) of the end of one’s life being nearer. One seventeen-year-old high-school senior in Burton et al.’s (1996) study balked, “Me, a teenager? Be for real, lady. Who’s got time for that? I’m a man. I’d better be one before I lose my life out on these streets.” Kotlowitz’s (1991) interviews and observations in Chicago housing projects present similar themes. Youthful aspirations were couched in terms of “if I grow up” rather than “when.”

The harshness of these experiences may also contribute to adult-like decision-making, responsibilities, and stresses. Kotlowitz (1991:xi) writes:

By the time they enter adolescence, they have contended with more terror than most of us confront in a lifetime. They have had to make choices that most experienced and educated adults would find difficult. They have lived with fear and witnessed death.

Confronting death and living in fear challenge the very notion of childhood (Kotlowitz 1991), and thus those who feel unsafe and who experience non-normative levels of violence growing up should age more quickly in subjective terms.

METHODS

Data and Sample

This research uses data drawn from the National Longitudinal Study of Adolescent Health (see SPQ website for additional information on Add Health, www.asanet.org/spq). Add Health began as a nationally representative study of American adolescents in grades 7–12 from 134 middle and high schools in 80 communities in 1994–1995. An in-school questionnaire was followed by in-home interviews with a subsample of students and one parent or parent-like figure in 1995 (Wave I) and with the adolescents again in 1996 (Wave II) and 2001–2002 (Wave III). The original sampling was conducted with a stratified design in which schools were selected from a complete list of American high schools based on their region, urbanicity, school type (public vs. private), racial composition, and size. Each of the selected high schools was matched to a school that fed into it, with the probability of the feeder school being selected proportional to its contribution to the high school’s student body. We weighted all analyses and used survey analysis techniques to adjust for the complex sample design (see Chantala and Tabor 1999).

We use data collected in Waves I and III and restrict our analysis to those respondents participating in both waves who were assigned a sampling weight. We further restrict our analyses to those who were under age 19 at the first wave to exclude students with non-normative ages for high school, and those who reported dramatically inconsistent ages at Wave I and III, resulting in 13,868 eligible cases. Of these cases, 4.6 percent were omitted because of missing data, resulting in an analysis sample of 13,230. For reasons described below, part of our analyses draw only on data from respondents ages 22 and under at Wave III.

Respondents were asked questions on their subjective age identity for the first time in Wave III (ages 18–26), so our dependent variables are taken then. Our measures of hardship were all taken from Wave I, and our measures of mediating experiences come from both Waves I and III. Older age identities may emerge close in time to when hardship is experienced and persist into young adulthood. When the subjective aging process speeds up or skips ahead, young people are likely to continue to feel older even if the pace of aging returns to that comparable with age peers after hardship abates. Young people may also continue to feel older relative to peers because circumstances surrounding the hardship persist, or because they continue to draw on a repertoire of adult-like behaviors learned under hardship conditions when those conditions change. Precedent for similar long-term social psychological effects exists; Mirowsky and Ross (2000) found that economic hardship reduced subjective life expectancy by several years even if decades had passed since the tough times. Although we assume that older subjective ages among those experiencing hardship in childhood and adolescence may be evident at earlier ages, we cannot test this assumption. We evaluate whether such young people have older subjective ages several years later, at a time when they are making the transition to adulthood.

Measures

Subjective age

Our measurement strategy follows closely from our theoretical concerns. Our interest is in feeling relatively older rather than in the full range of subjective ages. What might drive a distinction between feeling younger or about the same as one’s age peers would not necessarily parallel our discussion of how hardship promotes growing up faster. We therefore dichotomized responses to a question in Wave III which asked respondents, “In general, how old do you feel compared with others your age?” to contrast those who felt “older all of the time” (17 percent) with those who did not (83 percent; four original categories ranging from “older most of the time” to “younger all of the time”).²

We also examine how often young people feel like an adult as a second indicator of subjective age. Unlike our first measure, feeling like an adult could mean feeling older or not, depending on the age one is. We therefore examine feeling adult only for our younger respondents, those eighteen to twenty-two years old. Thus, this dependent variable captures feeling adult at reasonably young ages. Any specific age cutoff would be somewhat arbitrary, but we selected age 22 to maintain a sufficient sample size and because it approximates a culturally ideal end to the “college-age” years, even among those young people who do not attend college. As we do for relative age, we dichotomize this measure, contrasting those who felt like an adult “all of the time” (37 percent of those aged 18–22) with those who felt like an adult less often (63 percent; four original categories ranging from “never” to “most of the time”).

Although both these indicators of subjective age are single-item measures and limited as a result, previous research with this data shows that both are related in predictable and meaningful ways to measures of independence, maturity, and role behaviors, such as financial independence, leaving school, starting full-time work, and family formation (Johnson et al. 2007a, b; see also Benson and Furstenberg 2007). And our confidence is bolstered by examining two aspects of subjective age—relative age and self-perceived adulthood.

Our dependent variables are conceptualized as dichotomous, but the measures on which we drew were ordinal. Supplemental models using ordinal logit might have been appropriate for purposes of comparison, but the proportional odds assumption was not met for either ordinal dependent variable. For comparison purposes we instead estimated binary logit models with different cutoff points. Instead of comparing “all of the time” to all other responses for the measures of feeling older and feeling adult, we compared “all” and “most of the time” to other responses. Out of the respondents, 59 percent felt older most or all of the time; 71 percent of those aged 18–22 felt adult most or all of the time. We discuss the differences that arise under these specifications after we report the findings with our original cutoffs.

Hardship

Similar to the dependent variables, the very notion of hardship suggests one end, rather than a continuous distribution, of experience. As such, we measured hardship with dichotomous or categorical indicators to capture this distributional end. We examined family structure in adolescence using Harris’s (1999) coding scheme, measured in five categories, distinguishing married biological-parent families, single-mother families, single-father families, other two-parent families (largely stepfamilies), and other family structures. Based on prior research primarily in the area of household and family responsibilities, we expect subjective age to be highest in single-parent families, followed by stepfamilies, with those in married biological-parent families having the youngest subjective ages. Because the “other family structure” category includes a diversity of structures, we do not have a specific hypothesis about its effects on subjective age.

We measured economic hardship through a series of dummy variables that calculate household income during adolescence as a percentage of 1994 federal poverty thresholds (0–100 percent, 101–200 percent, 201–300 percent, 301–400 percent, and greater than 400 percent), which account for the number of people in the household. An additional variable indicates respondents who were missing income information.

To measure perceived lack of safety, we combined adolescent reports of whether they usually feel safe in their neighborhood (yes or no) and at their school (strongly agree to strongly disagree). Adolescents who answered “no” to feeling safe in their neighborhood and/or strongly disagreed that they felt safe at school were coded as feeling unsafe. To measure violence exposure, we combined reports of whether adolescents witnessed or experienced violence. Adolescents were asked how many times during the past 12 months a series of things happened (“you saw someone shoot or stab another person,” “someone pulled a knife or gun on you,” “someone shot you,” “someone cut or stabbed you,” and “you were jumped”). The original responses included 0, 1, or 2 (for more than once). We summed responses across the five items and distinguished adolescents with minimal exposure to violence (0 – 1) from those with greater exposure (2 – 10). We selected this threshold based on the severity and infrequency of the violence referenced in these items.

Mediators

We expected hardship to operate through three types of mediating experiences, including transitioning into adult roles earlier than one’s age peers, having had a lot of early responsibilities, and perceiving one’s life expectancy as severely curtailed. Measures of adult role transitions were obtained from Wave III. The transition markers included whether the respondent was out of school (1 = not in school; 0 = currently attending school), working full-time (1 = working 35 or more hours per week; 0 = working fewer than 35 hours per week), living away from the familial home (1 = not living with parents or relatives; 0 = living with parents or relatives), married or cohabiting (1 = currently married or cohabiting; 0 = not married or cohabiting), and had one or more children (1 = parent; 0 = not a parent).³

With respect to heavy responsibilities, we were able to tap into the specific forms of household and paid-work responsibilities. We believe there are other relevant responsibilities (e.g., caretaking, financial responsibility) to which we return in the discussion. We dichotomized a measure of adolescents’ Wave I reports of their involvement in housework in the past week to distinguish those who were involved almost daily (five or more times, the top response category) with other adolescents. While “daily” housework may not be statistically or socially non-normative, more frequent or heavier involvement in housework was not captured in the measure we have available. As a result, we may underestimate the effects of heavy housework. Because research has shown that 20 or more hours of paid work during adolescence is an important approximate threshold for excessive work involvement (National Research Council 1998), we distinguished adolescents who reported in Wave I typically working 20 or more hours per week during the school year from those working fewer hours or not at all. Employment at this intensity is more adult-like and we expect it captures heavier responsibility. We also examined whether the effect of housework and paid work on subjective age was stronger for younger respondents based on the assumption that our measures would better capture violations of age norms at Wave I for the younger respondents.

Our measures of anticipating a shorter life span were based on responses to two questions in Wave I: whether adolescents expected to be living at age 35 and whether they expected to be killed by 21. Because they represent conceptually distinct processes, we kept these as separate indicators in our models and in each case distinguished those who expected death (a 50–50 chance or less of being alive at age 35 and a 50–50 chance or greater of being killed by 21).⁴

Controls

We also controlled for chronological age, gender, race/ethnicity (Hispanic/Latino, non-Latino white, non-Latino Black/African American, Asian/Pacific Islander, Native American/American Indian, and “other race,” with respondents identifying as multiracial asked to choose a primary racial identity), nativity, parents’ education level, and residential mobility, all of which have been linked to subjective age as well as to our independent variables of interest (Benson and Furstenberg 2007; Galambos, Turner, and Tilton-Weaver 2005; Johnson et al. 2007a, b; Montepare and Lachman 1989; Shanahan et al. 2005). With the exception of chronological age, which reflected age reported at Wave III, these variables were measured at Wave I. Each was reported by the adolescent, except for nativity and parents’ education, which were based on adolescent reports only when parent reports were missing. Parents’ education was measured by the average years of schooling of residential parent(s). We converted the original ordinal scale to years of education (no school = 0, 8th grade or less = 8, some high school = 10, trade/vocational or business school instead of high school = 11, graduated high school = 12, trade/vocational or business school after high school = 13, some college = 14, college degree = 16, and graduate or professional training = 18). We measured residential stability in Wave I as the duration since last move in years.

FINDINGS

Descriptive statistics on the study measures appear in Table 1. Consistent with our expectations, each type of hardship was significantly associated with both feeling relatively older and, for those aged 18–22, feeling adult. Those with older subjective ages were less likely to be living with both biological parents in adolescence and were disproportionately from poor or near-poor families. Those with older subjective ages also disproportionately felt unsafe and experienced higher levels of violence in adolescence.

Table 1.

Weighted Means and Proportions for Variables used in Analyzing Older Subjective Age, Overall and by Values of the Dependent Variables

	Overallb	Feels Older All the Time		Feels Adult All the Time
		No	Yes	No	Yes
Dependent variables:
Feels older all the timea	0.17
Feels adult all the timea	0.37
Control variables:
Age at Wave III (18–26 years)	21.71	21.69	21.83	20.50***	20.75
Femalea	0.50	0.50	0.51	0.49***	0.54
Race/ethnicity
Latinoa	0.11	0.11	0.12	0.12	0.10
Non-Latino Whitea	0.68	0.69***	0.63	0.71*	0.66
Non-Latino Blacka	0.15	0.14***	0.21	0.11***	0.20
Native Americana	0.01	0.01	0.01	0.01	0.01
Asiana	0.04	0.04*	0.03	0.04**	0.02
Other race/ethnicitya	0.01	0.01	0.01	0.01	0.01
Parents’ educ. attainment (0–18 years)	13.16	13.23***	12.83	13.45***	12.76
Nativity (U.S. born)a	0.94	0.94	0.95	0.94***	0.97
Years in residence at Wave I (0–19)	7.75	7.91***	7.01	7.42**	6.94
Hardship variables:
Family structure at Wave Ia
Two biological parents	0.57	0.59***	0.47	0.63***	0.51
Other two-parent family	0.17	0.16	0.18	0.16***	0.20
Single-mother family	0.19	0.18***	0.26	0.18***	0.22
Single-father family	0.03	0.03	0.03	0.02	0.03
Other family structure	0.04	0.04***	0.06	0.02***	0.04
Household poverty statusa
Below poverty line	0.14	0.13***	0.18	0.13***	0.18
100–200% of poverty line	0.18	0.17**	0.21	0.17***	0.21
200–300% of poverty line	0.18	0.18	0.16	0.19	0.18
300–400% of poverty line	0.12	0.13	0.11	0.14***	0.10
> 400% of poverty line	0.18	0.19**	0.15	0.20***	0.15
Family income missing	0.20	0.20	0.19	0.17	0.19
Feels unsafea	0.12	0.10***	0.18	0.11**	0.14
Victim or witness of violencea	0.11	0.10***	0.17	0.08***	0.13
Potential mediators:
Housework every day at Wave Ia	0.37	0.36***	0.43	0.37***	0.43
Works 20+ hours/week at Wave Ia	0.18	0.17***	0.23	0.06***	0.09
Expects to be alive at 35a	0.87	0.88**	0.84	0.89***	0.85
Expects to be killed by 21a	0.14	0.14***	0.18	0.12**	0.15
Has a child at Wave IIIa	0.19	0.17***	0.28	0.09***	0.26
Lives on her/his own at Wave IIIa	0.57	0.55***	0.64	0.47***	0.56
Not a student at Wave IIIa	0.63	0.61***	0.72	0.48***	0.66
Married/cohabiting at Wave IIIa	0.49	0.46***	0.64	0.33***	0.56
Full-time worker at Wave IIIa	0.52	0.51**	0.57	0.40***	0.51

Source: National Longitudinal Study of Adolescent Health (1995). N = 13,230 for feeling older; 7,863 for feeling adult.

Note:

^a1 = yes.

^bFor feeling older analysis sample (N = 13,230) except on feeling adult all the time item (N = 7,863).

^*p < .05

^**p < .01

^***p <. 001 design-based F-tests comparing older to younger subjective age.

Analyses correct for complex survey design (stratification and clustering).

The measures of potential mediating experiences were also related to feeling relatively older and feeling adult as expected. Those with older subjective ages as young adults had been engaged in more frequent housework and more paid work than other adolescents. Those with older subjective ages were also less likely as adolescents to have expected to live to 35 and were more likely to have expected to be killed by 21. Finally, by Wave III they were more likely to have made each of the adult role transitions (parenthood, living on one’s own, not being a student, marriage/cohabitation, and full-time work).

Several sociodemographic variables were also related to feeling relatively older and feeling adult. Those who felt older and adult all of the time were less likely to be white or Asian and more likely to be black. They were also disproportionately from families with less-educated parents and who had moved more recently. Chronological age, gender, and nativity were not associated with feeling older, but older respondents, women, and native-born respondents were more likely to report feeling adult.

We examined our hypotheses about hardship and the processes through which it shapes subjective age in a series of logistic regression models shown in Table 2 for both dependent variables. Model 1 includes only the control variables. Transforming the regression coefficients to odds ratios (e^b) indicates that blacks were 51 percent more likely to feel older (e^.41 = 1.51) and (among eighteen to twenty-two year olds) 79 percent more likely to feel adult (e^.58=1.79) compared with whites. Older respondents, those from families with less-educated parents, and those who had experienced more recent residential moves were also more likely to feel older and adult all the time. Additional control variables predicted feeling adult but not feeling older, including being female, being white compared to Latino, and being US-born.

Table 2.

Binary Logistic Regression of Subjective Age on Hardship and Potential Mediators

	Feels Older All The Time			Feels Adult All The Time
	Model 1:	Model 2:	Model 3:	Model 1:	Model 2:	Model 3:
	Baseline model	Baseline + hardships	All mediators included	Baseline model	Baseline + hardships	All mediators included
Age at Wave III (years)	.05* (.02)	.04 (.02)	−.05 (.03)	.22*** (.03)	.20*** (.03)	.09** (.03)
Femalea	.04 (.07)	.09 (.07)	−.01 (.07)	.20*** (.06)	.24*** (.06)	.11 (.07)
Race/ethnicityb
Non-Latino Blacka	.41*** (.10)	.20 (.11)	.30** (.10)	.58*** (.09)	.47*** (.10)	.59*** (.10)
Latinoa	.03 (.11)	−.07 (.11)	.02 (.11)	−.34** (.13)	−.37** (.12)	−.32** (.11)
Asiana	−.22 (.15)	−.21 (.15)	−.08 (.17)	−.26 (.23)	−.24 (.23)	−.07 (.27)
Native Americana	.15 (.32)	−.03 (.36)	−.07 (.33)	−.31 (.47)	−.43 (.51)	−.59 (.49)
Other race/ethnicitya	.00 (.36)	.00 (.35)	.15 (.33)	−.24 (.29)	−.26 (.29)	−.11 (.28)
Parents’ Educational Attainment (Years)	−.06*** (.01)	−.03* (.02)	−.01 (.02)	−.13*** (.02)	−.11*** (.02)	−.07*** (.02)
Nativity (U.S. born)a	.11 (.16)	.03 (.16)	−.10 (.16)	.52** (.17)	.47** (.17)	.35 (.19)
Years in Residence at Wave I	−.03*** (.01)	−.02** (.01)	−.01* (.01)	−.02* (.01)	−.01 (.01)	.00 (.01)
Family Structure at Wave Id
Other two-parent familya		.22** (.08)	.11 (.08)		.30*** (.08)	.16 (.08)
Single-mother familya		.32*** (.08)	.29** (.08)		.13 (.10)	.05 (.11)
Single-father familya		.19 (.20)	.11 (.19)		.26 (.21)	.21 (.21)
Other family structurea		.44** (.15)	.31 (.16)		.47* (.21)	.25 (.21)
Household Poverty Statusc
Below poverty linea		.25 (.16)	.16 (.16)		.18 (.13)	.04 (.14)
100–200% of linea		.22 (.14)	.12 (.14)		.24* (.11)	.07 (.11)
200–300% of linea		.04 (.12)	−.01 (.13)		.09 (.11)	.01 (.12)
300–400% of linea		.12 (.13)	.13 (.13)		−.14 (.13)	−.15 (.14)
Family Income Missinga		−.03 (.13)	−.09 (.13)		.07 (.12)	−.06 (.12)
Feels Unsafea		.46*** (.09)	.47*** (.09)		.08 (.10)	.05 (.11)
Victim or Witness of Violencea		.48*** (.09)	.36*** (.10)		.34** (.11)	.21 (.11)
Housework Every Day at WIa			.27*** (.06)			.29*** (.07)
Works 20+ Hours/Week at WIa			.26** (.10)			.16 (.13)
Expects to Be Alive at 35a			.00 (.10)			.03 (.11)
Expects to Be Killed by 21a			.09 (.10)			.08 (.11)
Has a Child at WIIIa			.22* (.09)			.67*** (.08)
Lives on Her/His Own at WIIIa			.22*** (.06)			.08 (.07)
Not a Student at WIIIa			.20* (.09)			.20* (.09)
Married/Cohabiting at WIIIa			.43*** (.11)			.49*** (.09)
Full-time Worker at WIIIa			.12 (.08)			.27*** (.08)
Constant	−1.93** (.56)	−2.46*** (.59)	−1.27 (.65)	−3.78*** (.77)	−4.02*** (.76)	−2.90*** (.74)
Design-based F Df	7.71*** (10, 119)	8.95*** (21, 108)	10.64*** (30, 99)	17.32*** (10, 119)	9.87*** (21, 108)	14.43*** (30, 99)
Incremental F Df		9.65*** (11)	12.03*** (9)		4.55*** (11)	31.90*** (9)
% Correctly Predicted Values
Overallg	83.10%	83.10%	83.10%	64.11%	64.95%	67.53%
DV = 0	100	100	99.8	89.0%	88.1	85.4
DV = 1	.0	.001	.01	22.8	26.5	38.0

Source: National Longitudinal Study of Adolescent Health (1995). N = 13,230 for feeling older; 7,863 for feeling adult.

Notes:

^a1 = yes.

Reference categories:

^bNon-Latino White

^c> 400%

^d2 biological parents.

^gEstimation correcting for survey design effects (stratification and clustering) render other indicators of model fit, including pseudo R-square, inappropriate.

^*p < .05

^**p < .01

^***p < .001, comparing younger/similar to older subjective age. Standard errors in parentheses.

We introduced the hardship measures in Model 2.⁵ With respect to the family, living in other two-parent families, living with a single mother, or living in “other” family structures during Wave I increased the likelihood of feeling older in Wave III by 25 percent, 38 percent, and 55 percent, respectively. In terms of feeling adult, these increased likelihoods were 35 percent for other two-parent families and 60 percent for “other” family structures, but living with a single mother was not significant. The coefficient for single father was quite close to that for “other two-parent families” (the category dominated by stepfamilies), but did not reach statistical significance, owing in part to its smaller size. When economic hardship is considered without the other hardship domains, the two poorest income categories significantly predicted feeling older compared to the highest category (supplemental model not shown), but in Model 2 the only significant income difference was that respondents from 100–200 percent of the poverty line were 27 percent more likely to feel adult than those from the highest income category. We found through estimating several additional models (not shown), which considered the indicators of hardship singly and in combination, that adolescent family structure accounted for the relationship between economic hardship and feeling older. Feeling unsafe as an adolescent also predicted feeling older in young adulthood, as did experience of violence, increasing the likelihood of feeling older by 58 percent and 62 percent respectively. For feeling adult, only exposure to violence was significant, raising the likelihood by 40 percent. For both subjective age measures, “other family structures” compared to two-biological-parent families and experience of violence (as well as feeling unsafe for feeling older only) had the largest effects of all the hardship measures.

Several of the control variables’ effects were attenuated with the introduction of these hardship measures in Model 2. Most notably, the coefficients for non-Latino black were reduced significantly. Reductions were also evident for chronological age, parental education, nativity, and residential mobility.

Model 3 examines whether the effects of hardship were mediated by adolescents’ involvement in frequent housework or intensive paid work, anticipating a curtailed life course, and making adult role transitions. Frequent housework in adolescence predicted feeling adult and feeling older in young adulthood, and intensive paid work predicted feeling older in young adulthood. But because supplemental analyses (see supplemental tables provided at SPQ website, www.asanet.org/spq) show that adding only these variables to Model 2 did not significantly reduce the size of any of the hardship coefficients, they did not mediate any of the estimated hardship effects. We examined whether the effects of housework and paid work were stronger for younger sample members, but interaction terms were not statistically significant (not shown). Neither measure of expecting a very curtailed life course predicted subjective age in the multivariate models and therefore did not mediate any of the hardship effects.

Experiencing some types of adult role transitions by Wave III, however, was not only strongly associated with having an older subjective age, but also mediated some of the hardship effects (see Baron and Kenny 1986 for a discussion of mediation criteria).⁶ Having a child, not being a student, and being married or cohabiting predicted both feeling older and feeling adult. Living on one’s own only predicted feeling older, while being a full-time worker only predicted feeling adult. Among the role transitions, marriage/cohabitation had the largest effect on feeling older, while parenthood had the largest effect on feeling adult. Among the hardship and role transition measures in Model 3, feeling unsafe had a larger effect than any of the role transitions on feeling older, but the effect of parenthood was largest on feeling adult; indeed, none of the hardship measures significantly predicted feeling adult once role transitions and the other potential mediators were included.

In conjunction with supplemental analyses (see note 6 and Model 3 of supplemental table A2 provided at SPQ website), the analyses show that the significant effects of hardships from each of the three domains were fully mediated by role transitions: “other” family structures for both dependent variables, “other two-parent family” structures for feeling older only, and 100–200 percent of the poverty line and experience of violence for feeling adult only.⁷ The associations of having a single mother and experience of violence with feeling older were partially mediated by role transitions, and “other two-parent family” structures were partially mediated by feeling adult.⁸ Of the hardships that were significant in Model 2, only the effect of feeling unsafe on feeling older was not at least partially mediated by role transitions. It is also notable that the effects of chronological age, gender, and nativity on feeling adult were substantially mediated in Model 3, as were the effects of parental education on both measures of subjective age.

Of course, hardship is not randomly distributed across the population; rather, it tends to be compounded. In bivariate analyses, respondents who experienced one category of hardship were significantly more likely to have experienced each other type of hardship (all p<.001; not shown). The cumulative effects of hardship across domains are evident in Figure 1, in which we present the predicted probabilities of feeling older and feeling adult all the time for different hypothetical respondents. While experience with each type of hardship was positively associated with experiencing each other type, there was no small number of prevalent combinations of hardships which we could use to portray predicted probabilities. Instead, we illustrate the compounding effect by introducing each dimension of hardship and building on previous ones. The first set of predictions is based on Model 2, which includes all hardship measures. In these predictions, we set the value of continuous measures to their mean and categorical measures to their mode. We manipulated the values of several hardship measures systematically.

Figure 1. Predicted Probabilities of Older Subjective Age for Hypothetical Respondents by Levels of Hardship and Adult Role Transitions.

Notes: The first 5 predictions are based on Table 2, Model 2.

The “advantaged” hypothetical respondent has no hardships (for household income we selected > 400% of poverty line) and average or modal values for other variables. The last 2 predictions are based on Table 2, Model 3. The hypothetical respondent is the same as immediately previous (other 2-parent family, poor, unsafe and violence exposure, and otherwise “average”). “Other 2-parent family” means living with two parents, but both are not biological parents.

At the far left, we present the predicted probability of feeling older for an advantaged hypothetical respondent whose household income was in the highest category, who had two biological parents in the household, and did not feel unsafe or experience violence (with average values for all other variables). This scenario represents a very common pattern for respondents, 48 percent of whom did not experience any type of hardship. This hypothetical respondent had a low predicted probability of feeling older all the time at Wave III, at 0.12 (or a 12 percent likelihood of feeling older). His predicted probability of feeling adult all the time was 0.35, or a 35 percent likelihood of feeling adult. In comparison, 17 percent of all respondents felt older all the time and 37 percent felt adult all the time (weighted means).

The effect of changing the hypothetical respondent’s family from two biological parents to another type of two-parent family (these were mostly stepfamilies) is shown next, increasing the probability of feeling older to 0.14 and of feeling adult to 0.42. Keeping that change in family structure and manipulating the hypothetical respondent’s Wave I household income to below the poverty line raised these predicted probabilities to 0.18 and 0.46 respectively. Retaining these values and adding both feeling unsafe in one’s neighborhood or school and being exposed to violence doubled the predicted probability of feeling older, to 0.36, while the predicted probability of feeling adult rose to 0.57. These predictions demonstrate that the effects of hardship on subjective age are cumulative across domains. As hardships compounded across the figure, the predicted probability of feeling older tripled and the predicted probability of feeling adult increased by 63 percent.

For the second comparison in Figure 1, we used Model 3 from Table 2 and manipulated the number of adult role transitions made by a hypothetical respondent who has experienced hardship across all domains. When he experienced no adult role transitions, his predicted probabilities of feeling older and feeling adult were 0.17 and 0.30, respectively, compared to 0.40 and 0.70 when he made all five transitions.

Sensitivity Analysis

Although our interest was in those with the oldest subjective ages (i.e., those who felt older or adult “all of the time”), we assessed whether our findings would change by altering the cut-off and including “most of the time” in our older subjective age category; we summarize the differences here (tables are available from the authors). For the analysis of feeling older in Table 2, hardship results differed in that the other two-parent family coefficient was not fully mediated in Model 3. The household income effects were also stronger. The 100–200 percent and 200–300 percent of poverty threshold coefficients were significant in Model 2, and they were reduced but remained significant in Model 3. So while below-poverty incomes are associated with feeling older “most” and “all the time” only through their association with family structure, incomes in the two categories just above the poverty line are associated with older relative age independent of family structure when we consider those who feel older most or all of the time. Feeling unsafe was not significantly related to feeling older in either model, indicating its relevance only for those with the oldest relative ages. There were also two changes in the effects of the adult role transitions. Those working full-time were more likely to feel older most or all of the time though they were not more likely to feel older all of the time, and student status was no longer predictive of subjective age when we considered feeling older most or all of the time.

For the analysis of feeling adult in Table 2, the combination of “most” and “all” of the time represented 71 percent of the sample. Hardship findings differed in that “other two-parent family” structures and experience of violence remained significant predictors of feeling adult in Model 3 rather than being fully mediated by adult role transitions. The single-mother family coefficient was significant in Model 2, but as with the other coding scheme, it was not significant in Model 3. Household incomes at 100–200 percent of the poverty line and “other” family structures were not significantly associated with feeling adult in Model 2. In Model 3, housework and student status were not significant.

DISCUSSION

The process of growing up, and aging more broadly, varies in pace across individuals. We combine insights from the life-course perspective, which draws our attention to age norms, the age-grading of roles, and how life stages can serve as master statuses, with existing conceptualizations of identity to understand individuals’ subjective understanding of age. What we argue and find in this study has implications for understanding age identities as well as identities more broadly, and contributes to several important life-course debates.

The current study generally supports our argument that hardship in childhood and adolescence fosters an older subjective age in early adulthood. There were certainly variations in the effects of hardship in the family, economic, and safety/violence domains across the two indicators of older subjective age and by their operationalization, but indicators within each of these domains predict older subjective age within at least one and often more than one of these specifications. These hardships share a challenge to contemporary cultural expectations for childhood or adolescence that lead to the sense that one is older than one’s age peers. We do not measure those age-related expectations directly, but instead identify a set of experiences we hypothesize are age non-normative, each stemming from one or more hardships. We grouped these age non-normative experiences into three (non-exhaustive) types, including transitioning into adult roles earlier than one’s age peers, having had a lot of early responsibilities, and perceiving one’s life expectancy as highly curtailed.

Respondents who had made adult role transitions held older subjective ages than those who had not yet done so. Partnering, becoming a parent, and completing one’s education predicted feeling older and adult all of the time. Establishing an independent household was also important to feeling older, and working full-time was important for feeling adult all of the time. In terms of taking on responsibilities, frequent housework in adolescence was associated with feeling older and feeling adult and intensive paid work in adolescence was associated with feeling older. Contrary to our hypotheses, however, anticipating a sharply curtailed life span during adolescence was not associated with an older subjective age in multivariate models. Previous consideration of this process has focused on older populations. It may be that nearness to the end of life only shapes subjective age in middle and late adulthood, when physical decline from age and specific health conditions serve as a constant reminder of one’s mortality.

Of these three types of experiences, only with respect to adult role transitions did we find evidence of mediation. All but one of the significant hardships were either fully or partially mediated by role transitions. The effects of hardships on feeling adult were mediated somewhat more strongly than on feeling older. The introduction of role transitions and high-intensity paid work also mediated several of the control variables, including gender, parents’ education, nativity, and residential stability. It is interesting to note that the effect of chronological age on feeling older is fully mediated by hardship (though the coefficient changes little until role transitions, housework, and high-intensity paid work are added), and the effect of chronological age on feeling adult is partially mediated (55 percent of the effect) by role transitions and housework. In other words, social processes account for most or all of the effect of chronological age on subjective age identity, as many social scientists working in this area would expect.

Our analyses of the hypothesized mediation processes were limited in two important ways that need to be addressed in future research. First, we lacked measures of very high levels of involvement in housework, and of financial responsibilities, contributing income to the family, and significant involvement in caring for siblings or older family members that may mediate the effects of hardship we observe. Second, subjective age was only measured at one time-point in our study, and as a result, we cannot know whether the responsibilities and roles we examine precede older age identities. For example, high-intensity employment in adolescence or early entry into adult roles could be more common among those who already felt older. Longitudinal measures would enable researchers to address this possibility and examine the extent to which behaving in an “older” fashion and having an older subjective age may be reciprocally linked over time.

Our consideration of age-related cultural expectations and examination of the case of age identities also advances our understanding of identities more broadly. Identity theory (Stryker and Burke 2000), which encapsulates the primary scholarship on identities in contemporary sociological social psychology, conceptualizes identities as based in roles. Scholarship in this tradition tends to move forward from this conceptualization to better understand identity salience, behavioral choice, self-verification, and other aspects of identity processes, with little attention to other bases of identity. Our findings affirm the importance of roles, but also suggest the value of continued consideration of the bases of identities. Adult social roles had some of the largest effects on feeling older and self-perceived adulthood. Yet non-role factors also shape subjective age identities. These factors include age-graded responsibilities, which we tapped into through the frequency of performing housework and intensity of paid employment, as well as lack of safety and exposure to violence, which constituted an experience of hardship significantly related to feeling older even with our identified mediators controlled. With respect to self-perceived adulthood, race/ethnicity and parental education levels also shaped age identity in ways not fully accounted for by differential acquisition of adult social roles, nor by exposure to hardship or our theorized mediators. These findings suggest that the social context of growing up also directs one’s understanding of whether one has reached adult status. Further attention to these sources of age identities will likely prove useful in building a broader theoretical understanding of how identities develop.

Our findings also speak to several ongoing discussions in the life-course tradition. First, findings from this study support a key distinction from this tradition: statistical age normativity and social age norms are not one and the same (Settersten 2003). The distribution of responses to our measure of relative age indicates this most clearly: 59 percent of respondents feel older than others their age most or all of the time. It is thus statistically normative to feel non-normative. Our findings suggest that comparisons are made to cultural understandings of the behaviors one engages in at different ages, rather than to the actual behaviors of others one’s age. Like the children of Lake Wobegon who are all above average, young people in the United States disproportionately feel older than others their age. Previous studies indicate this changes as people get older. Adults in their thirties and beyond tend to report younger subjective ages compared to their chronological age (Montepare and Lachman 1989; Galambos et al. 2005). Our findings on the predictors of subjective age also support that it is not what adolescents’ age peers are doing (i.e., what is statistically normative), but ideas about what one should be doing at a given age (social norms) that shapes subjective age.

Second, rapid change in the demography of the transition to adulthood has led to an important debate on the relevance of adult social roles in understanding the contemporary nature of adulthood in Western societies (e.g., Arnett 2000; Furstenberg et al. 2003; Shanahan et al. 2005). We find in a nationally representative sample that entry into adult roles is associated with age identity in a way that affirms their importance in defining and signifying adulthood for today’s young people. These role transitions also partially mediate the effect of chronological age on feeling adult, showing that social processes, more than just advancing years, influence young adults’ subjective sense of their own age.

Finally, the effect of early hardship on the self-concept has long been of interest to life-course social psychologists and the findings of this study contribute to this line of inquiry as well. The primary foci in this area have been the implications of hardship for self-efficacy and self-esteem (e.g., Whitbeck et al. 1991, 1997). We join another recent study in expanding the scope of consideration, showing that hardship is associated with another aspect of the self-concept: age identity. Foster et al. (2008) also recently found that childhood neglect and adolescent intimate-partner violence accelerate “subjective weathering,” a concept overlapping with relative age. The ways that young people interpret their place in the life course and temporal dimensions of their lives, including the pace of aging, are linked to early hardship.

Much work is still needed on the implications of older subjective ages during adolescence and young adulthood. Our contributions in the current study do not hinge on whether older age identities in this stage of the life course are detrimental or beneficial, but that question deserves additional attention. Growing up too fast can mean “pseudomaturity” (Galambos et al. 1999; Greenberger and Steinberg 1986), in which feeling older and mimicking adult behaviors is not matched by genuine psychosocial maturity. Having an older subjective age is associated with adolescent substance use, problem behavior, and more involvement with opposite-sex peers (Arbeau, Galambos, and Jansson 2007; Galambos et al. 1999), some of which can be thought of as adult behaviors assumed too early. Subjective age could be a mechanism reproducing or maintaining disadvantages from childhood or adolescence into adulthood.

And yet, feeling older may also facilitate taking more mature approaches to solving problems and accepting responsibility for oneself and others. Moreover, having an older subjective age may be an adaptive response that enables adolescents to make the best of difficult situations. For example, a teenage parent who comes to see him/herself as older, more mature, adult even, may be more likely to behave in adult ways. Shanahan (2000:677) noted, “The manner in which people respond to challenges and stressors may account for differences in the transition to adulthood.” To the extent that feeling adult is linked to more mature reactions to stress, feeling adult sooner could be beneficial. Initial positive benefits could, alternatively, be combined with later difficulties, however, and vice versa.

Perhaps the larger question might be posed in terms of when having an older subjective age is detrimental and when it is beneficial. We see several potentially important distinctions to consider. First, the implications of developing older subjective ages in adolescence may depend on the level of agency afforded the adolescent. Some young people may choose to take on greater responsibilities and make earlier transitions. Young people who do not anticipate attending college in the future and who invest heavily in paid work during adolescence (Mortimer 2003) may fall into this group. In contrast, hardship by its very definition is something that is thrust upon children and adolescents, and thus we might expect more negative consequences. A second distinction is likely important, though, and that is whether in developing an older subjective age, young people also develop genuine maturity. Even those thrust into demanding situations may rise to the challenge, developing successful coping skills and becoming more autonomous and socially responsible. The extant literature suggests it is pseudomaturity—having an older subjective age without genuine maturity—that is linked to problematic outcomes during adolescence. Such behaviors could not be viewed as beneficial outcomes even if agentically selected. Whether genuine or pseudomaturity develops may rest on family and community contexts, a third distinction in need of scrutiny. One can imagine the adolescent who takes on extra responsibilities in the face of hardship but who works alongside parents or other individuals who model maturity and provide support. One can also imagine an adolescent burdened with responsibilities without such support (e.g., assuming a parental role for siblings and managing the household for an overworked or otherwise unavailable parent). As this brief discussion indicates, we must examine the context of “growing up faster” during the early years if we are to understand its implications for the life course.

Biographies

Monica Kirkpatrick Johnson is an associate professor in the department of sociology at Washington State University. Her research focuses on education and work-related processes during adolescence and the transition to adulthood, and particularly the social psychological experience of this life-course transition. Her current work examines processes of subjective age identity and formation and change in ambitions during this period of the life course.

Stefanie Mollborn is an assistant professor of sociology and faculty in the Health and Society Program of the Institute of Behavioral Science at the University of Colorado at Boulder. Her research focuses on social psychological approaches to understanding health behaviors and outcomes over the life course. Current projects include an analysis of the importance of resources for the early development of teenage parents’ children and an examination of the antecedents and consequences of social norms about adolescent pregnancy.