Adolescents’ Changing Future Expectations Predict the Timing of Adult Role Transitions

Abstract

Individual differences in the transition to adulthood are well established. This study examines the extent to which heterogeneity in pathways to adulthood that have been observed in the broader U.S. population are mirrored in adolescents’ expectations regarding when they will experience key adult role transitions (e.g., marriage). Patterns of change in adolescents’ expectations and the relations between their expectations and subsequent role transitions are also explored. Data from 626 youth in Grade 11 (M age = 16), Grade 12, and early adulthood (M age = 23) are analyzed using mover-stayer latent transition analysis. Results indicate three profiles of expected timing, corresponding to youth who anticipate early role entry (i.e., early starters), youth who anticipate earlier entry into employment but no other roles (i.e., employment-focused), and youth who anticipate delays in role transitions favoring increased education (i.e., education-focused). Two-thirds of youths changed their expectations from Grade 11 to 12. Grade 11 and 12 profile membership predicted role transitions in early adulthood. These findings highlight the importance of adolescents’ expectations and changes in expectations across time in shaping entry into adulthood.

The transition to adulthood is made up of multiple role transitions, typically represented as changes in status within the domains of education, work, independent living, romantic relationships, and family formation (e.g., Hogan & Astone, 1986). It is widely assumed that adolescence sets the stage for these transitions, as adolescents make decisions about education, work, and family formation that have implications for their future life course (Shanahan, 2000). Consistent with this notion, prior research has shown that adolescents formulate goals for their adult lives and have a sense of when key role transitions will occur (Greene, 1990; Nurmi, 1993; 2004). Furthermore, adolescents’ expectations regarding their future transitions are presumed to influence the decisions they make and their actual pattern of role transitions (Crockett & Beal, 2012). Indeed, findings show that adolescents’ expectations regarding their adult roles and the timing of role transitions are associated with their outcomes in early adulthood (Beal & Crockett, 2010; Johnson & Mortimer, 2000; Mello, 2008). Taken together, theory and empirical evidence support the notion that expectations for adult role transitions provide a cognitive framework (or road map) that guides adolescents’ behavior and influences their transition to adulthood.

Although these findings provide a promising foundation, several key questions about the process linking adolescents’ expectations and their actual patterns of role transitions remain unanswered. These include: a) whether and how expectations in one domain (e.g., work) are synchronized with expectations in other domains (e.g., education, partnering), suggesting cross-domain coherence; b) whether adolescents systematically change their expectations over time; and c) whether patterns of expectations across domains, and changes in these patterns, predict subsequent role transitions and related adult outcomes. This study seeks to address these gaps by identifying distinct patterns in adolescents’ expectations and linking them to young adult outcomes. More precisely, we identified patterns of expectations regarding the timing of multiple adult transitions (e.g., age of school completion, marriage), examined changes in these patterns during adolescence, and explored their implications for adult outcomes.

Adolescents’ Expectations about Early Adulthood

Future orientation is an over-arching term that refers to the extent to which adolescents consider the future, the content of their future goals, and when particular goals will be achieved (Seginer, 2009). Theory suggests that adolescents are developing a sense of personal identity that includes who they will be and what they will do in the future (Erikson, 1968; Markus & Nurius, 1986; Nurmi, 1993). Erikson (1968) argued that the role selections young people make in specific domains as part of their identity formation (e.g., occupation, religion, sex roles) are related to a coherent sense of self that also reflects their anticipated niche in society. Today there appears to be theoretical consensus that adolescents are integrating the various aspects of their identity into a cohesive sense of self in the present and the future, primarily through social processes unfolding in family, peer, education, work, and neighborhood settings (Crosnoe & Johnson, 2011; Greene, 1990; Nurmi, 2004).

According to Erikson, identity achievement is associated with a sense of wholeness and coherence (Erikson, 1968; 1980). This implies that adolescents’ identity choices in particular domains are coordinated with each other and form coherent patterns. By extension, expectations about adult roles in various domains should also be compatible, where youth expecting to enter a particular profession simultaneously expect to complete the required training for that career. In this way, expectations channel youth toward specific life-course trajectories and ultimately influence their adult outcomes (Nurmi, 2004). There is also a dynamic element to the development of expectations, such that expectations continue to be modified based on experience and may change during adolescence. Thus, the expected timing of adult role transitions in different domains may be coordinated and form coherent patterns, and change in expectations (which could also be coordinated) may also occur.

While many studies have focused on future-oriented cognitions in a single domain (e.g., work), fewer studies have examined cross-domain coherence in the expected timing of role transitions. The studies that have examined cross-domain relationships (e.g., associations between educational plans and expectations regarding family formation) have provided some evidence of coherence, particularly for girls. For example, girls may anticipate both work and family roles and perceive them to be of equal importance (Johnson & Mortimer, 2000); girls also appear to consider potential competition between family and career goals when considering career choices (Stevens, Puchtell, Ryu, & Mortimer, 1992). Furthermore, the average expected ages of role transitions appear to follow a sequential pattern (e.g., school completion and job entry prior to marriage and parenting) that is fairly consistent throughout adolescence (Greene, 1990). Furthermore, family plans that are made during adolescence predict the subsequent sequencing of transitions to adulthood (Johnson & Mortimer, 2000). There is also some evidence that adolescents adjust educational and career goals to align with each other (Beal & Crockett, 2013). Additional research that considers an expanded range of domains (i.e., education, work, independent living, and family formation) longitudinally is needed to determine whether there is broader coherence among adolescents’ future plans that persists across time.

Understanding the social context in which adolescents form their expectations (e.g., the influence of familial expectations, cultural norms and opportunities, community resources) is also critical for understanding adolescents’ expectations for adulthood. The normative sequencing of events during the transition to adulthood in broader society may serve as a template for the sequencing of adolescents’ expectations regarding those transitions (Greene, 1990). Gender and socio-economic status are also important to understand expectations and observed outcomes in the transition to adulthood, as constraints based on gender and SES are prevalent (Johnson & Mortimer, 2000). Expectations held by youth in particular communities are further constrained by institutional supports, including the school system, the availability of higher education and job training programs to support work, and the employment sectors represented in the community (Nurmi, 2004). Additional factors that contribute to the future-oriented goals selected by youth include the expectations, resources, and norms of parents and other family members (Seginer, 2009) and their own experiences and performance (e.g., grade point average; Andrew & Hauser, 2011). These mechanisms not only shape the initial range of choices young people consider for their futures, but likely also influence how future expectations change across time (Nurmi, 2004).

Changes in Expectations during Adolescence

The beliefs individuals hold about their future are thought to change during adolescence (e.g., Gottfredson, 1981; Nurmi, 2004). Gottfredson (1981) hypothesized that adolescents use their understanding of what opportunities for future occupations are available and what resources, skills, and preferences they have in order to narrow their range of possible occupations until a single occupation is selected and pursued. This process of modifying future goals based on experiences and environmental feedback is also reflected in Nurmi’s (2004) theory of goal selection and attainment in adolescence. Nurmi argues that adolescents identify possible goals, are channeled by existing social structure, personal preferences, and opportunities toward or away from particular goals, refine and re-visit their choices based on feedback and experience, and ultimately shape their own transitions into future adult roles by selecting which goals are pursued. In support of this notion, empirical evidence indicates that adolescents modify work-related expectations as they approach adulthood. Armstrong and Crombie (2000) found that 8^th to 10^th graders adjusted their occupational aspirations and expectations across time until they co-aligned. Similarly, Beal and Crockett (2013) reported that adolescents adjusted their occupational goals to align with educational goals across high school. Finally, Willoughby (2010) reported that adolescents altered their marital attitudes across adolescence, with adjustments toward prioritizing marriage, but that there was less priority for marriage with higher educational aspirations. In all cases, these findings provide evidence of increasing coherence in expectations across domains as youth get older. Further, there is some evidence of SES modifying changes in expectations; Bozick and colleagues (2010) reported that educational expectations were relatively stable for high SES youth across childhood and adolescence, while expectations for low and middle SES youth were less stable.

While changes in expectations over time have been detected in the literature, studies have generally not been successful in identifying developmental trends in expectations. For example, Crockett and Beal (2012) used multilevel modeling to examine trajectories of the expected timing of young adult transitions, and detected relative stability across high school. Similarly, Schmid, Phelps, and Lerner (2011) found relative stability between Grades 7 and 8 in mean scores for adolescents’ expectations related to future college attendance, health, work, and family. In one of the rare instances where a general pattern of change was detected, Mello (2008) reported distinct trajectories in adolescents’ expectations for education and work using hierarchical linear modeling: educational expectations initially decreased and then increased in adolescence while occupational expectations increased and then decreased as youth approached age 20. However, only age 14 expectations predicted young adult outcomes in that study, suggesting that change in expectations (i.e., slope) is not associated with adult outcomes. Thus, there are mixed findings regarding a) whether adolescents’ expectations change in a systematic way as they near adulthood, and b) whether change in expectations is important for understanding adult outcomes.

Adolescents’ Expectations Predict their Adult Outcomes

Prior studies indicate that adolescents’ goals and aspirations regarding future educational and occupational achievement predict their educational and occupational outcomes in early adulthood (Beal & Crockett, 2010; Johnson & Mortimer, 2000; Mello, 2008). To date, the link between adolescents’ expectations for the future and young adult outcomes has been studied primarily with regard to educational and occupational attainment (Johnson & Mortimer, 2000; Lee, Hill, & Hawkins, 2012; Masten et al., 2004; Mello, 2008; Reynolds & Burge, 2008; Schoon & Parsons, 2002; Schoon & Polek, 2011) and to a lesser extent family roles (Johnson & Mortimer, 2000; Willoughby, 2014). For example, Beal and Crockett (2010) found that adolescents who had higher educational expectations subsequently completed higher levels of education. Reynolds and Burge (2008) reported a similar relationship between expectations to finish college and subsequent college graduation, which contributed to greater educational attainment for women in particular. With regard to family formation, Finlay and colleagues (2014) found that adolescents who reported believing that family would be important to them as adults were more likely to be married and have children, but less likely to be working in early adulthood. Likewise, Johnson & Mortimer (2000) reported that, in addition to predicting educational and occupational outcomes, adolescent expectations related to education, work, and family roles predicted delayed entry into cohabitation for both genders in young adulthood, and delayed entry into marriage for women. Willoughby (2014) similarly found that adolescents’ expected timing and attitudes about marriage predicted the probability of transitioning into marriage (but not cohabitation) in young adulthood. It is worth noting that most existing studies rely on adolescents’ expectations at a single point in time and with a limited number of domains to predict adult outcomes. This approach likely over-simplifies the complexity of the links between expectations and young adult outcomes. However, to our knowledge, no previous study has examined patterns of expectations (i.e., how expectations across domains are related) and changes in those patterns of expectations among adolescents or used those patterns to predict adult outcomes.

Observed Patterns in the Transition to Adulthood

There is a significant body of work demonstrating that indicators of the transition to adulthood (i.e., education, work, independent living, marriage, parenthood) are sequenced and dependent on each other, such that experiencing one role transition typically relates to the timing of another role transition. For example, early entry into intensive work is associated with a subsequent earlier education completion and career acquisition, as well as and earlier entry into independent living, marriage, and parenthood (Bachman, Staff, O’Malley, Schulenberg, & Freedman-Doan, 2011; Mortimer, Vuolo, Staff, Wakefield, & Xie, 2008; Staff & Mortimer, 2007; Staff, Vaneseltine, Woolnough, Silver, & Burrington, 2012), while lower levels of consistent part-time work can contribute to higher educational attainment (Jeremy Staff & Mortimer, 2007). Likewise, Johnson and Reynolds (2013) found that while adolescents’ expectations to attend college were associated with college attendance in young adulthood, experiencing a role transition in work or family domains led to a decline in college attendance. Similarly, Reynolds and colleagues (2007) reported that delaying role transitions in marriage and parenting was associated with an increase in achieving occupational goals in adulthood. This sequencing of role transitions is also reflected in the normative expected ages for role transitions identified in the United States, where independent living is expected to normatively proceed marriage and parenting transitions (Settersten, 1998; Settersten & Hägestad, 1996).

Insights into how adolescents’ expectations about the transition to adulthood are integrated across domains can also be drawn from the literature examining patterns in observed pathways to adulthood. Research on the actual timing and sequencing of roles as young people transition in education, work, independent living, romantic relationships, and family formation has demonstrated diverse pathways to adulthood, where subsets of young people experience qualitatively different routes to becoming adults (Schulenberg & Zarrett, 2006; Settersten & Ray, 2010). For example, Osgood and colleagues (2005) reported six pathways to adulthood among youth from the Midwest: fast starters who transitioned in education, work, partnering, and parenting relatively early compared to their peers, slow starters who struggled to make any transitions, working singles who transitioned early in education and work domains but delayed marriage and parenting, educated singles who delayed entry into work and family roles in favor of higher education, educated couples who partnered early but delayed other transitions in favor of education, and parents without careers. These patterns are similar to those described in a British cohort (Schoon, Chen, Kneale, & Jager, 2012). When fewer transitions are assessed, the patterns differ somewhat, but heterogeneity is still the norm (Braboy Jackson & Berkowitz, 2005; Maggs et al., 2012; Oesterle et al., 2012). For example, Vuolo, Mortimer, and Staff (2014) report four pathways from school to work: Four-year college degree to career, Two-year college degree to career, some college to no career, and no training to no career; education, work experience, and socioeconomic status were important predictors of which youth were on each of those pathways. When independent living and family formation were considered with the same sample of young people, three patterns emerged: Early transitions across domains, Delayed transitions across domains, and Multifaceted transitions, with delays in some domains but not others (Mortimer, Staff, & Lee, 2005); again, gender and SES played an important role in determining who belonged to each of the identified pathways to adulthood. Consistent with these findings, three patterns have been found across a variety of studies and samples: a group of young people who transition out of education and into work and/or family formation (i.e., partnering and parenting) early, a group that delays some of the transitions to adulthood in favor of higher education, and a group that enters the workforce relatively early without family formation. Given that adolescent future-oriented cognitions have been shown to predict young adult outcomes (Beal & Crockett, 2010; Finlay, Wray-Lake, Warren, & Maggs, 2014; Guo, Marsh, Morin, Parker, & Kaur, 2015; Khattab, 2015; Mello, 2008; 2009; Mike Vuolo, Jeylan T. Mortimer, & Jeremy Staff, 2014) pathways to adulthood may offer an important framework for understanding how adolescent expectations fit together across domains.

The Current Study

Based on prior theory and research, the goal of this study was to provide a nuanced picture of adolescents’ expectations about adult role transitions, including the extent to which these expectations cohere across domains, how they change during adolescence, and their contributions to adult outcomes. We examined developmental changes in adolescents’ expectations about the timing of five young adult role transitions (school completion, work entry, independent living, marriage, and parenthood) and identified distinct profiles of expectations, patterns of movement between identified profiles, and outcomes associated with them. The analysis took into account demographic characteristics that have been found to predict actual pathways to adulthood, including parental education, gender, academic performance, and educational expectations (Amato et al., 2008; Maggs, Jager, Patrick, & Schulenberg, 2012; Räikkönen, Kokko, Chen, & Pulkkinen, 2012; Salmela-Aro et al., 2012; Mike Vuolo et al., 2014). Based on previous research, we hypothesized that (1) there would be multiple distinct patterns of expectations regarding the timing of adult role transitions, consistent with diverse pathways to adulthood; (2) some adolescents would change their expectations during high school; and (3) expectations would predict actual timing of adult role transitions. Specifically, we expected that anticipating delayed onset of adult roles would be associated with a delay in observed role transitions in adulthood. We also anticipated that there would be variations in patterns and outcomes based on gender and SES, drawing on the work described previously. The theoretical model integrating these hypotheses is depicted in Figure 1. The figure indicates that there are distinct profiles of expectations at each Grade (Grades 11 and 12) and also that an adolescent’s profile can change from Grade 11 to Grade 12 (i.e., movement). The figure also indicates that the initial profile in Grades 11 and 12 and movement between profiles predict the actual timing of adult role transitions.

Figure 1.

The Mover-Stayer Latent Transition Analysis, k = 3 latent profiles were identified: Early Starters, Employment-Focused, and Education-Focused. Movement between Grade 11 and Grade 12 profiles were rrecoded to be −2 (moved from Grade 11 education-focused to Grade 12 early starter), −1 (moved from Grade 11 education-focused to Grade 12 employment-focused or Grade 11 employment-focused to Grade 12 early starter), 0 (no movement between Grade 11 and Grade 12), 1 (moved from Grade 11 early starter to Grade 12 employment-focused or from Grade 11 employment-focused to Grade 12 education-focused), or 2 (moved from Grade 11 early starter to Grade 12 education-focused).

Method

Participants

Data came from the Rural Adolescent Development (Vicary, 1991) and Rural Young Adult Transitions (masked citation) studies, which were conducted from 1985 to 1995. The target population for the initial 5-year study included all junior high students (in Grades 7, 8, and 9) in a rural school district in the eastern U.S. Participants in these three cohorts were surveyed annually through 12^th Grade (from 1985 to 1990). Greater than 90% of students invited to participate completed the first survey; 60% of the original sample completed the young adult follow-up, and 55% of the total sample were present for all waves of data collection. In addition, all participants were surveyed in 1995 when they were young adults between the ages of 21 and 26 (M age = 23.25, SD = 1.25). For the present analysis, the adolescent data (N = 626) were restructured by grade, so that T1 corresponds to Grade 11 for all participants (M age = 16.13, SD = 0.50). Thus, the assessment timeline for the present analysis included Grades 11, 12, and the young adult follow-up (see Table 1 for descriptive statistics). Grades 11 and 12 were chosen for these analyses due to their proximity to the young adult follow-up.

Table 1.

Descriptive statistics for expected ages, demographics, and control variables in grades 11 and 12 and observed outcomes in young adulthood.

Adolescent Predictors	Grade 11 M (SE)	Grade 12 M (SE)
Expected Age of
Finishing school	20.74 (0.12)	20.74 (0.10)
Starting to work	20.67 (0.13)	20.94 (0.12)
Living on own	20.28 (0.11)	20.67 (0.12)
Getting married	23.27 (0.14)	23.47 (0.13)
Have first child	24.40 (0.14)	24.82 (0.13)
Demographics
Age	16.13 (0.02)	17.14 (0.02)
Maternal education	2.64 (0.07)	2.60 (0.06)
Control Variables
Grades in school	7.59 (0.08)	7.81 (0.08)
Educational expectations	4.09 (0.06)	4.11 (0.06)
Occupational aspirations	53.64 (0.67)	52.51 (0.68)
Occupational expectations	53.09 (0.76)	53.13 (0.74)
	Young Adult Outcomes M (SE) or %
Age	23.25 (0.05)
Age of educational completion	20.03 (0.14)
Educational attainment	4.30 (0.08)
Age entered workforce	18.18 (0.06)
Work status	24% not employed
	56% employed part time
	20% employed full time
Age of independent living	18.67 (0.09)
Independent living status	56% living independently
Age of marriage	20.79 (0.15)
Marital status	37% married
Age of parenting	20.17 (0.18)
Parenting status	27% with children

Roughly half (52%) of participants were boys, and the sample was almost entirely white. In Grade 12, 15% of the mothers of youth in this study had not graduated from high school, 51% had finished high school, and 34% completed at least some education beyond high school. In contrast, 18% of fathers had completed some high school, 40% finished high school, and 42% completed at least some education beyond high school. In order to focus on expected timing during the high school years, we excluded any adolescents who reported improbable expectations (e.g., first child at age 50) or who experienced an adult transition by Grade 12 (n = 75; e.g., became a parent during high school).

Measures and Procedure

Participants in Grades 11 and 12 completed written surveys in the fall of the academic year during regular school hours. Parents provided passive consent. The survey was administered by project staff, and school personnel were not present. The young adult follow-up survey was mailed to participants in 1995; 88% of the high school cohort used in this study (present in Grades 11 or 12) also participated in the young adult follow-up. Young adults provided active consent and returned surveys via mail.

Expected timing of adult role transitions (Grades 11 and 12)

The core measures for this study were adolescents’ annual reports of when (at what age) they expected to complete their schooling, get their first regular job, move out of their parents’ home, get married, and have their first child. All questions were asked using the stem, “At about what age do you expect to, or did you, if it’s already occurred…”

Young adult role transitions and related outcomes

As part of the follow up survey, participants reported whether or not they had ever experienced several role transitions: lived independently, been married, or had a child. Those who answered “yes” were asked the age at which they experienced that event (timing of role transitions). In addition, age of education completion was assessed using the question, “How old were you when you completed your education?” and educational attainment was assessed using the item “How far did you go in school?” with responses from 1 (some high school) to 7 (graduated from professional school). Age of entry into work was assessed using the question, “How old were you when you got your first job after you left high school?” Job status was assessed using the item “What is your current work situation?” with the following responses: employed part-time, employed full-time including military service, laid off, unemployed but looking for work, unemployed but not looking for work, on sick leave, maternity leave, or parental leave, or full-time homemaker. These response options were recoded as 0 (not working), 1 (working part time), or 2 (working full time) for the purpose of this study.

Demographic, psychological, and educational covariates

The model included the following covariates assessed in Grades 11 and 12: adolescents’ reports of gender (0 = male), age in years, each parent’s level of education (a proxy for SES) coded as 0 (Grade school only), 1 (some high school), 2 (finished high school), 3 (technical school), 4 (some college), 5 (finished college), or 6 (beyond college), and typical school grades from 0 (Mostly F’s) to 9 (Mostly A’s). Educational expectations (“How far do you plan to go in school?” 1 = some high school, 6 = professional degree), prestige scores for occupational aspirations (i.e., “What kind of work would you like to do?”), and prestige scores for occupational expectations (i.e., “What kind of work do you think you will probably do?”) were also included as covariates because increased occupational prestige and educational plans both contribute to delays in the age that education is completed and youth enter the workforce. Occupational aspirations and expectations were coded for prestige using National Opinion Research Council (NORC) codes (Davis & Smith, 2002), consistent with previous research (Beal & Crockett, 2010; Chang, Chen, Greenberger, Dooley, & Heckhausen, 2006; Dubow, Huesmann, Boxer, Pulkkinen, & Kokko, 2006).

Analytic Approach

To identify groups of individuals with similar profiles of adolescent expectations for the timing of adult role transitions in grades 11 and 12 and simultaneously estimate which teens were changing profiles between grades, a Mover-Stayer Latent Transition Analysis (MS-LTA; e.g., Nylund, 2007; Collins & Lanza, 2010) was used. MS-LTA is a data analytic technique that can model response variable heterogeneity in Grades 11 and 12 via identification of latent (i.e., unobserved) sub-groups, capture sub-group membership change over time, and use both subgroup membership at Grades 11 and 12 and sub-group membership change to predict subsequent adult outcomes. As shown elsewhere (e.g., see Peugh & Fan, 2013), MS-LTA extends latent profile analysis (LPA) to quantify heterogeneity in longitudinal data. LPA assumes that differences in response patterns across participants are due to the sample being comprised of a mixture of two or more (i.e., k ≥ 2) unobserved sub-groups with differing profiles of scores. In addition to estimating sub-group membership at Grade 11 and Grade 12, MS-LTA further extends LPA for longitudinal data by adding a second-order latent categorical variable that identifies which participants did and did not change their latent subgroup membership (profile) between two time points (e.g., Grade 11 and Grade 12).

Consistent with the work of Nylund (2007), the MS-LTA was conducted in a sequential, step-wise progression that involved: 1) a latent profile analysis to identify the presence of latent sub-groups separately for Grades 11 and 12 based on distinct profiles for expected ages, each profile differing in the average expected ages of role transitions in multiple domains; 2) a second-order latent transition analysis to detect which participants did and did not change (i.e., “movers” and “stayers”) their profile sub-group membership between Grades 11 and 12; and 3) a regression analysis regressing outcome variables onto (a) latent profile membership at Grade 11 and Grade 12 and (b) a variable that specified movement between profiles from Grade 11 to Grade 12. Covariates included in the analysis were age, gender, occupational and educational aspirations and expectations, parent educational attainment, and grades in school. In MS-LTA analyses, these covariates were used to estimate conditional profile membership and movement between profiles.

The presence of between 1 and 6 possible latent profiles was examined during the enumeration process to determine the optimal number of profiles for each grade. Up to six profiles were estimated because that is the maximum number of profiles found in the literature on pathways to adulthood (Osgood et al., 2005). Like other mixture model approaches, MS-LTA provides fit indices to evaluate each of the possible solutions. This includes the Akaike Information Criterion (AIC), Consistent AIC (C-AIC), Bayesian Information Criterion (BIC), Draper’s modified BIC (D-BIC), Hannan-Quinn Information Criterion (HQ), and Hurvich-Tsai AIC (HT-AIC) for model comparison, and Entropy, Normalized Entropy Criterion (NEC), Classification Likelihood Criterion (CLC), and the Integrated Completed Likelihood BIC (ICL-BIC) to examine classification quality. For all indices, lower values indicate better fit.

After identifying the K latent profile solution that best fit the data at both Grade 11 and Grade 12, a second-order latent categorical “movement” variable was specified and added to the joint Grade 11 and Grade 12 latent transition analysis model to identify those participants who changed (“movers”) and did not change (“stayers”) their latent profile group membership between Grade 11 and Grade 12. The addition of the second-order latent categorical “mover/stayer” variable defines an MS-LTA analysis (Collins & Lanza, 2010). In the final step of the MS-LTA, latent profile memberships at Grades 11 and 12, as well as whether a participant was a “mover” or “stayer,” served as predictor variables for each young adult outcome variable. Age, gender, occupational and educational expectations, parent educational attainment, and grades in school were included as covariates predicting latent profiles and movement between profiles. Age in young adulthood, gender, and parent educational attainment were included as covariates predicting all young adult outcomes. Consistent with best practices for mixture modeling analyses (see Enders & Gottshall, 2011), the default maximum likelihood parameter estimation algorithm in Mplus (MLR, see Muthén & Muthén, 1998–2015) was used to properly handle all missing data on predictor and response variables.

Results

Latent Profiles of Adolescents’ Expected Ages of Role Transitions

To address the first research hypothesis, that there are multiple distinct patterns of expectations regarding the timing of adult role transitions, we sought to determine the number and nature of distinct profiles observed in adolescents’ expectations about when they would make the five adult role transitions. Consistent with the recommendations of Li and Hser (2011), the data on expected ages of role transitions in Grades 11 and 12 were analyzed using a conditional LPA model (i.e., with covariates). The covariates included gender, age, mother’s educational attainment, and father’s educational attainment in Grade 11, and well as Grade point average, educational expectations, occupational expectations, and occupational aspirations in Grades 11 and 12. Results for the conditional LPA analysis in Grade 11 supported the presence of k = 3 distinct profiles; however, in Grade 12 there was support for either k = 2 or k = 3 distinct profiles. When Grade11 and 12 data were estimated together there was again support for the presence of either k = 2 or k = 3 distinct profiles, based on all available enumeration index criteria for a joint mixture model (Table 2). To aid with determining the mixture solution that best reflected the observed data, we then examined the classification accuracy for movement between Grades 11 and 12 (i.e., when estimated profile membership in Grades 11 and 12 suggested movement, was the individual assigned as a mover). There was improved accuracy for a k = 3 class solution (75.5% accurately assigned versus 58.3% accurately assigned in a k = 2 class solution). Accordingly, a k = 3 latent profile solution was selected for both the Grade 11 and Grade 12 data (see Figure 2).

Table 2.

Latent profile analysis results for k = 2 and k = 3 class solutions in a joint transition model for Grades 11 and 12.

		Model Comparison Statistics					Classification Quality Indices
		Information Criteria (IC)1					Entropy and Entropy Penalty2
Latent Trajectories	AIC	C-AIC (NC-AIC)	BIC (NBIC)	D-BIC (ND-BIC)	HQ (NHQ)	HT-AIC (N HT-AIC)	Entropy	NEC	CLC	ICL-BIC	NICL-BIC
k = 2	57890	60002 (58782)	59618 (58398)	58912 (57693)	58560 (58045)	57952 (68215)	0.82	0.4545	57124	59619	58400
k = 3	57602	60697 (58910)	60134 (58347)	59099 (57312)	58583 (57828)	57996 (80003)	0.81	0.4566	56477	60136	58348

¹IC indices with N-Adjusted IC indices shown in parentheses;

²Entropy Penalty indices; Bolded entries indicate the best-fitting k transition class model. AIC= Akaike Information Criterion, C-AIC = Consistent AIC, BIC = Bayesian Information Criterion, D-BIC = Draper’s modified BIC, HQ = Hannan-Quinn Information Criterion, HT-AIC = Hurvich-Tsai AIC.

²NEC = Normalized Entropy Criterion, CLC = Classification Likelihood Criterion, ICL = Integrated Completed Likelihood.

Figure 2.

Estimated group patterns for expected ages of transition in education, work, independent living, marriage, and parenting for the identified k = 3 solution in the Latent Profile Analysis for Grades 11 (left) and 12 (right).

In Grade 11, each of the k = 3 latent profiles differed in their average ages of expected role transitions. As shown in Figure 2, the first profile had the lowest average age of expected role transitions in Grade 11, which was replicated in Grade 12 (dark grey; “Early Starters”). The second profile (light grey, “Employment-focused”) had higher average ages of expected role transitions compared to “Early Starters” for all response variables in Grades 11 (Figure 2, left panel). In Grade 12, higher average ages were found for education, work, and independent living, and these youth had the highest average age of expected marriage and parenthood (Figure 2, right panel). The third profile had the highest average age of expected role transitions for all response variables in Grade 11 (black; “Education-focused”) and the highest average age of expected education and work role transitions in Grade 12. In Grade 11, 28% of youth were classified as “Early Starters,” 49% of youth were classified as “Employment-focused,” and 23% of youth were classified as “Education-focused.” In Grade 12, 23% of youth were classified as “Early Starters,” 13% of youth were classified as “Employment-focused,” and 64% of youth were classified as “Education-focused.”

Differences in latent profile membership by gender and SES

There were significant differences in profile membership based on gender and SES. In Grade 11, girls were significantly more likely to be classified as early starters, while boys were more likely to be classified in either the employment-focused or education-focused profiles (χ² = 16.76, p < .01). Similarly, in Grade 12 girls were significantly more likely to be classified as early starters or education-focused, while boys were more likely to be classified in the employment-focused profiles (χ² = 20.75, p < .01). With respect to SES, maternal education in Grade 11 was significantly lower for adolescents classified as early starters (F (2, 624) = 13.96, p < .01); this was also the case for paternal education in Grade 11 (F (2, 624) = 16.71, p < .01). In Grade 12, maternal education (F (2, 624) = 5.06, p < .01) and paternal education (F (2, 624) = 20.66, p < .01) continued to be significantly lower for adolescents classified as early starters.

Change in adolescents’ patterns of expectations across high school

To examine whether some adolescents change their expectations during high school we examined profile changes between Grades 11 and 12. Profile changes between two time points are typically modeled in MS-LTA as a binary latent variable indicating “movers” and “stayers.” In such models all movement, regardless of direction or magnitude of change, is treated as equivalent. However, in our case, all changes from one profile to another were not equivalent, because their meaning depended on both the direction and magnitude of change. For example, moving from “Education-focused” at Grade 11 to “Early Starters” at Grade 12 may not have the same impact on adult outcomes as moving from “Early Starters” to “Education-focused.” Furthermore, moving from “Early Starters” at Grade 11 to “Employment-focused” at Grade 12 represents a smaller change in expected ages than moving from “Early Starters” at Grade 11 to “Education-focused” at Grade 12, so the magnitude of change also needs to be captured. To create a variable that was sensitive to both direction and magnitude, each participant’s latent profile membership at each Grade was assigned a theoretical value based on the average ages of expected role transitions (i.e., “Education-focused” = 3, “Employment-focused” = 2, and “Early Starters” = 1), then each participant’s profile membership value at Grade 11 was subtracted from his or her latent profile membership value at Grade 12. This recoded ordinal movement variable ranged from −2 to +2, where positive values indicated movement from a latent profile characterized by lower average age of expected role transition at Grade 11 to a latent profile characterized by higher average age of expected role transition at Grade 12. Values of either −2 or +2 indicated movement between the extreme latent profile groups (i.e., “Early Starters” to “Education-focused” and vice versa), values of −1 or +1 indicated movement between adjacent latent profiles, and values of zero indicated no change in latent profile (i.e., stayers).

Movement was observed as follows: over one-third of adolescents (37.3%) showed a stable profile (did not switch profiles between Grades 11 and 12); about half showed movement toward delaying role transitions (12.8% moved from “Early Starter” to “Education-focused” profiles, and 37.8% moved from “Employment-focused” to “Education-focused” profiles); and about 12% showed movement toward earlier role transitions (2.2% moved from “Education-focused” to “Employment-focused” profiles, 2.1% moved from “Education-focused” to “Early Starter” profiles, and 7.7% moved from “Employment-focused” to “Early Starter” profiles). No youth moved from “Early Starter” to “Employment-focused” profiles.

Differences in movement by gender and SES

There were significant differences in movement by gender and SES. Specifically, girls were more likely to be classified as movers, both in delaying role transitions and in moving toward earlier role transitions, whereas boys were more likely to remain in the same profile in Grades 11 and 12 (χ² = 27.05, p < .01). With respect to SES, there were no significant differences in Grade 11 maternal education (p = .38) or Grade 12 maternal education (p = .88) for those who moved and those who did not move profiles. However, those who moved from profiles characterized by earlier role transitions to profiles characterized by delayed role transitions had significantly higher paternal education in Grade 11 (F (2, 624) = 3.65, p = .03) and in Grade 12 (F (2,624) = 4.15, p = .02).

Profiles of Expectations, Changing Profiles, and Adult Outcomes

The results described up to this point pertain to identifying latent profiles of expected timing for adult role transitions and estimating movement (switching profiles) between Grades 11 and 12. The final question was whether adolescents’ expectation profiles, and any movement between profiles, predicted the actual timing of role transitions and other adult outcomes. This was accomplished by estimating a model where gender, age at the time young adult data were collected (to adjust for censoring), parent education, profile membership in Grades 11 and 12, and movement between profiles were predictors of young adults’ reported role transitions and the age at which that role transition took place. Those model-based estimates are provided in Table 3. Effects were estimated simultaneously for all observed outcomes in a single analysis. Significant effects of at least one predictor or covariate were seen for all outcomes except current independent living status (results not shown).

Table 3.

Unstandardized regression estimates (standard errors) of group membership on young adult outcomes for early-starters and employment-focused youth compared to young people classified as education-focused in Grades 11 and 12.

				Young Adult (YA) Outcome Variables
Adolescent Predictor Variables	School Leaving Age	Education Attainment	Work Entry Age	Work Status	Independent Living Age	Marriage Age	Marriage Status	Parenting Age	Parenting Status
YA Age	0.34** (0.12)	−0.01 (0.06)	0.21** (0.06)	0.06 (0.11)	0.11 (0.10)	0.42** (0.14)	0.21 (0.14)	0.51** (0.17)	0.34* (0.16)
Gender	0.24 (0.23)	−0.09 (0.12)	−0.12 (0.12)	−0.62** (0.21)	−0.24 (0.18)	−0.20 (0.28)	−0.81** (0.27)	0.25 (0.36)	−0.46 (0.32)
Maternal Education	0.05 (0.09)	0.10* (0.05)	0.02 (0.05)	0.08 (0.08)	−0.08 (0.07)	0.11 (0.10)	0.03 (0.11)	0.06 (0.12)	−0.26 (0.14)
Paternal Education	0.20** (0.07)	0.13** (0.04)	0.01 (0.04)	−0.10 (0.07)	−0.14** (0.05)	−0.16 (0.09)	−0.14 (0.08)	0.16 (0.11)	0.01 (0.10)
Grade 11 Profile	0.42 (0.26)	0.02 (0.13)	−0.11 (0.13)	−0.22 (0.24)	0.57** (0.21)	0.59* (0.28)	−0.75** (0.30)	0.17 (0.32)	−0.64 (0.36)
Grade 12 Profile	0.94** (0.21)	0.93** (0.12)	0.17 (0.12)	−0.10 (0.21)	−0.61** (0.18)	0.50* (0.23)	−0.27 (0.25)	−0.32 (0.27)	−0.64* (0.29)
Movement	0.10 (0.24)	−0.17 (0.13)	−0.05 (0.13)	−0.01 (0.23)	0.13 (0.20)	0.02 (0.27)	−0.52 (0.28)	0.47 (0.30)	0.19 (0.32)

^**p < .01,

^*p < .05

NOTE: Comparisons in Grade 11 and Grade 12 are to the Education-Focused profile; Movement coded to capture direction (positive/negative) and magnitude, where early starters who did not change classes = 0, early starters who became education-focused = 2, employment-focused youth who became early starters = −1, employment-focused youth who did not move classes = 0, employment-focused youth who became education-focused = 1, education-focused youth who became early starters = −2, education-focused youth who became employment-focused =−1 and education-focused youth who did not change group membership = 0.

First, with respect to Grade 11 profile membership, being in a profile associated with delayed role transitions was associated with a significantly older age of independent living, a decreased likelihood of being married, and an increased age of marriage for those who had married by the young adult outcome. Being in a profile associated with delayed role transitions in Grade 12 was associated with having a significantly older age of school completion, completing more education, older age of marriage, and a decreased likelihood to be a parent at the young adult outcome. Only one effect, age of independent living, ran counter to what was hypothesized; being in a profile where role transitions were delayed in Grade 12 was associated with a younger age of entry into independent living at the young adult follow-up. Of note, there was no significant effect of movement when Grade 11 and Grade 12 profiles were included in the model.

In addition to the effects of profile membership, there were also significant effects of age, gender, and SES on young adult outcomes. Specifically, being older at the young adult follow-up was associated with older ages for exiting school, entering work, marriage, and parenting. This likely reflects that older youth had more time to make these transitions. With respect to gender, women were significantly less likely to be working full-time or to be married at the young adult follow-up. Finally, the effects of SES spanned both maternal and paternal education, where higher maternal education was associated with significantly higher educational attainment, while higher paternal education was associated with significantly higher age of exit from school and educational attainment, and an earlier age of entry into independent living.

Discussion

In this study, we sought to extend the literature on the transition to adulthood by characterizing adolescents’ expectations regarding the timing of five adult role transitions, examining changes in expectations over a one-year period, and testing the associations between patterns of expectations (and changes in them) and adult outcomes. When expected ages for all five role transitions were considered together, three coherent patterns of expected timing were identified: early starters, employment-focused youth, and education-focused youth. There was also evidence that many youth changed their expectations (i.e., profile membership) from Grades 11 to 12. Finally, profile membership in Grades 11 and 12 predicted actual role transitions and their timing in early adulthood.

Heterogeneity, Coherence, and Change in Adolescents’ Expectations

The three profiles of expectations we identified mapped onto the three most common pathways to adulthood in the literature (e.g., Braboy et al., 2005; Maggs et al., 2012; Oesterle et al., 2012; Osgood et al., 2006; School et al, 2012): early starters, employment-focused, and education-focused. These patterns generally reflected expectations of an early transition to adulthood across domains, a delayed transition to adulthood accompanied by extended schooling, and moderately early transition to work with some delays in family formation. Furthermore, the profiles we identified showed coherence across life domains in that adolescents who anticipated a relatively early (or delayed) transition in one domain (e.g., education) appeared to anticipate relatively early (or delayed) transitions in other domains as well. This coherence is consistent with the notion that adolescents consider role transitions in multiple domains when they imagine their future, and often focus on either an early track or a later track to adulthood across domains (Osgood et al., 2006). The exception was employment-focused youth, who initially anticipated intermediate ages in all domains but later expected a delayed transition to marriage and parenthood. As discussed below, this group also showed the greatest changes in membership between Grades 11 and 12. Consistent with the research done by others (e.g., Johnson & Mortimer, 2000; Markus & Nurius, 1986; Unemori, Omoregie, & Markus, 2004), our findings appear to suggest some identity integration across domains, such that adolescents tended to display cross-domain coherence in the age of education completion, work entry, and family formation, for example, regardless of which profile they belonged to. Further, across groups the sequence of young adult transitions was expected to be fairly similar (i.e., education and work before marriage and parenting transitions) consistent with societal prototypes for the transition to adulthood (Greene, 1990) and empirical findings from other studies (Johnson & Mortimer, 2000).

One profile we did not detect in our data that has been detected with observed young adult outcomes data is that of educated singles. It may be the case that our “education-focused” youth represent a mix of youth who become educated singles and educated couples, and that the timing of partnering is unexpectedly delayed for some education-focused young adults. Research on unexpected young adult outcomes is being conducted (e.g., Messersmith & Schulenberg, 2008; Jeremy Staff, Harris, Sabates, & Briddell, 2010) but that work has emphasized discordance between expectations and outcomes within specific domains and not across profiles. The current study was limited in that uncertainty was not explicitly measured, and all non-aged responses that might have indicated uncertainty were coded as missing. Future work examining expected delays in partnering among youth pursuing higher education could help us better understand whether youth anticipate pursuing education and delaying partnering. It is also noteworthy that none of the profiles we detected reflect non-normative sequences (e.g., parenthood before school completion). When more than three profiles of observed (rather than our expected) role transitions have been observed, they have included in-sequence transitions (i.e., fast starters, working singles, educated singles, and educated couples) and out-of-sequence transitions (e.g., parents without careers). Perhaps we failed to find out-of-sequence patterns because such non-normative patterns are not anticipated by young people.

We also observed changes in expectations between Grades 11 and 12, where nearly two-thirds of the sample changed their expectations sufficiently to be re-classified into a different profile in Grade 12. This movement is consistent with theories proposing that adolescents modify their expectations in light of social feedback and new information, particularly as they approach adulthood (e.g., Gottfredson, 1981; Nurmi, 2004). The high rate of change suggests that for many young people, plans are in flux during the junior year of high school, which may be indicative of general uncertainty about the future.

Regarding the direction of change, most movement was into profiles associated with expectations of later transitions than the adolescent initially anticipated; only 12% of youth moved into profiles associated with earlier transitions. Furthermore, the most common change (46% of the sample) was into the education-focused profile. Thus, the main tendency was for adolescents to adjust their expectations in the direction of extending their education and postponing other adult role transitions. This finding is similar to those reported by Mello (2009), who also observed a positive trend in educational expectations during the end of high school. It is also consistent with Nurmi’s (2009) theory of selection and refinement of future goals, which suggests that young people choose a goal and then adjust or refine their goal based on feedback from their environment, particularly when they near an anticipated social transition (i.e., leaving high school; Tynkkynen, Nurmi, & Salmela-Aro, 2010). Movement into the education-focused group meant that by Grade 12, over 60% of youth were classified as education-focused. The fact that so many youth shifted into the education–focused group between their junior and senior of high school suggests either that they were not seriously considering their futures until then or that they acquired critical information at that time, such as the educational requirements for desired occupations. Either way, the results point to a need for interventions earlier in the high school years to provide information on educational requirements and to help adolescents make informed decisions.

We also found significant effects of gender and SES on profile membership, where girls were more likely to be classified as early starters in Grade 11, as early starters or education-focused in Grade 12, and were significantly more likely to change profiles between Grades 11 and 12 compared to boys. With respect to SES, lower maternal and paternal education in Grades 11 and 12 were significantly associated with membership in the early starter profile, and higher paternal education in Grades 11 and 12 was also associated with movement toward profiles where role transitions were delayed (e.g., education-focused). These findings are consistent with previous studies examining the impact of gender and SES on adolescent expectations and the transition to adulthood (Amato et al., 2008; Goyette, 2008; Johnson & Mortimer, 2000; Johnson & Reynolds, 2013; Mello, 2009; Mortimer et al., 2008; Reynolds & Burge, 2008; Reynolds & Johnson, 2011). There are many reasons for gender and SES to influence both adolescents’ expectations for their futures and their young adult attainment, not the least of which is that previous research has demonstrated that social class connects adolescents to resources to ensure goal attainment with respect to adolescents’ plans for college and career – especially for girls (Hardie, 2015). Future work examining how to leverage change in these patterns (i.e., the impact of shifting career plans for socio-economically disadvantaged youth) will be critical for applying these findings in the future.

Expectations and Subsequent Role Transitions

The three profiles of expectations we identified were differentially associated with observed role transitions and status. Specifically, compared to education-focused youth, individuals who were classified as early starters in either Grade 11 or Grade 12 went on to complete their education earlier and had significantly lower educational attainment as adults; these youth also married at younger ages and were more likely to have been married by the time of follow-up. Thus, adolescents’ expectations about their future lives tended to predict their subsequent experience. Interestingly, the transitions experienced in young adulthood for this sample was generally not unlike that of the rural US population, based on 1990 US Census data (United States Bureau of the Census Vital Statistics Division, 1991). Specifically, the mean educational attainment for the full sample aligned with technical school, and 1990 Census data indicates that for white persons aged 20–24 years, 86% of the US population and 68% of the rural population completed high school or more than high school education, while only 12% of the US population and 4% of the rural population had completed a bachelor’s degree or higher – consistent with the rates observed in the current sample. Further, 67% of all persons and 68% of rural persons ages 20–24 were in the labor force, compared to 76% of this sample. With respect to marriage, 21% of the US population 26% of the rural population under age 24 and had ever married, compared to 37% in this sample. With respect to parenting, 19% of the US population and 25% of the rural population under age 24 were parents, compared to 27% of this sample.

One counter-intuitive finding was that youth classified into profiles where role transitions were delayed reported a significantly earlier transition into independent living in young adulthood. The rural community where participants lived did not have local options for post-secondary education. It may be the case that young people who pursued college had to leave their communities, and reflected that independence in their survey responses.

Of note, once profile membership in grades 11 and 12 and covariates were included, movement between profiles did not offer any additional predictive utility with respect to young adult outcomes. It may be that once start and end points are included, movement does not provide any additional unique information. Future studies examining differences between youth who do and do not change profile membership in longitudinal studies are warranted to investigate this further.

Limitations and Future Directions

The data used for this study are unique in several ways which strengthen this study: the majority of adolescents in the community participated in the surveys during high school, there was consistency in the domains of expected transitions and observed transitions in the young adult follow-up, and a variety of indicators of the transition to adulthood were assessed. However, the findings from this study should be considered in light of several limitations. The methods used here, LPA and MS-LTA, are hypothesis-generating by nature, and as a result some of the findings here may be unique to this sample; thus, replication is needed. The sample was primarily white and from a single rural community that was undergoing economic stress at the time of data collection. Future studies with other birth cohorts and more diverse samples will help to inform the generalizability of these findings. Nonetheless, the fact that the expectation profiles observed are similar to those found in studies of actual role transitions supports their validity.

Additionally, the young adult follow-up occurred early in the life course, when participants were age 23 on average. While transitions out of education and into the workforce were experienced by most participants, only half of participants reported living independently, one-third reported being married, and one-fourth reported having children. Thus, for marriage and parenting in particular, these findings are only relevant for distinguishing young people who experienced transitions relatively early in their life course. We included age at the young adult follow-up to address this limitation statistically; however, future research that follows adolescents farther into adulthood would help determine whether the observed differences persist. This is particularly important for considering the impact of education and employment-focused youth who expect delaying partnering and parenting. A broader index of partnering that includes cohabitation as well as marriage might also be better in order to capture current trends in partnering. It may also be worthwhile to consider patterns of expectations and changes in patterns prior to the end of high school. Previous research, for example, has found that educational expectations are relatively stable from childhood into adolescence (Bozick et al., 2010). Knowing when change is and is not occurring with respect to patterns across domains of expectations is important for understanding the course of expectations.

Finally, the data were collected in the 1990s, and represent Generation X rather than Millennials. While expectations to attend and graduate college have increased, with variation by gender and SES (Reynolds & Johnson, 2011), these generations do not differ substantially with respect to college attendance, which has remained relatively stable at around 20% since the early 1990s; however, college graduation rates have increased by close to 20% for Millennials compared to Generation X, with stronger effects for women (The Council of Economic Advisors, 2014). Further, Millennials are much less likely to join the labor market while pursuing education, and have experienced increased unemployment and underemployment as a result of the recession. Millennials are also delaying marriage, with marriage rates near 50% for individuals ages 25–34, and little variation by educational attainment. Fewer Millennials anticipate ever marrying compared to Generation X, while expecting to transition into parenting has not changed. Finally, with respect to age of entry into parenting, median age has remained relatively stable, near age 22, since 1990 for young people who do not pursue any post-secondary education, while delayed entry into parenting is observed for those pursuing higher education. Based on these generational differences, we would anticipate changes in the content and timing of expectations between the cohort used in this study and those born in more recent decades (Goyette, 2008), and replication with more recent cohorts is needed to ensure the relevance of our findings for contemporary youth.

However, there is little reason to think that the developmental process of forming and refining one’s expectations for the future would have changed. Therefore, we believe that this paper highlights relations among expectations and links between expectations and outcomes that are likely to persist in more recent samples. Future research addressing these limitations will necessarily be longitudinal, spanning adolescence and adulthood, and should capture not only expectations and outcomes but also the cognitive and metacognitive processes involved in formulating and refining expectations for adulthood, perhaps by including interview measures. Understanding such processes will be useful for refining theories that address future orientation and for designing interventions to support successful transitions into adulthood for all young people.

Despite these limitations, the present findings extend our current understanding of adolescents’ future-oriented cognitions in at least three ways. First, they provide evidence of coherence among adolescents’ expectations regarding role transitions in diverse domains, as reflected by three distinct profiles. This is consistent with the notion that young people consider multiple adult roles simultaneously and may have a holistic view of their future life course, but that not all youth anticipate the same life course. Second, our finding that some adolescents changed their expectations from Grade 11 to Grade 12 indicates that youths continue to formulate, review, and update their expectations, as some theorists have argued (e.g., Gottfredson, 1981; Nurmi, 2004). Third, the profiles of expectations and changes in those patterns predicted observed outcomes in early adulthood, consistent with the notion that adolescents’ expectations foreshadow their future behavior. Determining whether this foreshadowing represents a causal effect of expectations on behavior (e.g., a self-fulfilling prophecy) or simply an accurate estimate of future outcomes without causal influence is an important task for future research.