Early Adult Transitions and Their Relation to Well-Being and Substance Use
Abstract and Keywords
Evidence shows that well-being increases during the period between late adolescence and early adulthood, but questions remain about how widespread this increase may be and why it occurs and, more generally, how the course of well-being relates to the various diverse pathways out of high school. Substance use also tends to increase during this period, reaching its lifetime peak during the early twenties, depending on the given cohort and substance. Well-being and substance use, while not necessarily sharing a common etiology or developmental course across the life span, may increase among young adults during transition in part because of the new roles and contexts that provide more freedom and selection of opportunities. Using data from four waves of nationally representative U.S. panel data spanning ages eighteen to twenty-four, this chapter investigates early adult transitions and their relation to well-being and substance use. It analyzes the timing, sequencing, and covariation of social role transitions related to school and work, romantic involvement (specifically marriage), parenthood, and independence in the form of leaving the parental home.
In recent decades, the period between adolescence and adulthood has become extended for many segments of the population, making this period more than simply a staging ground for adulthood (Arnett 2000). At the same time, traditional sequences of events that mark adult status (e.g., complet-ing school, obtaining full-time employment and gaining financial independence, getting married, and starting a family) appear to have become less central to the definition of adulthood (if not less common). Nevertheless, embedded within this period of life are multiple and specific developmental tasks and transitions in the domains of achievement, affiliation, and identity (Oerter 1986; Schulenberg and Maggs 2002). Although there is not a single normative or prescribed pathway through these various tasks and transitions (Cohen et al. 2003; Settersten 2003; Shanahan 2000), successfully negotiating at least some of them (and particularly those viewed as central by the young person) is likely to be associated with more salutary trajectories of health and well-being and to provide a foundation for optimal development during adulthood (Masten and Curtis 2000; Ryff, Singer, and Seltzer 2002; Schulenberg, Bryant, and O̓Malley, in press; Weise, Freund, and Baltes 2000).
Well-being has been found to increase during the period between late (p.418) adolescence and early adulthood (Gore et al. 1997; Schulenberg et al. 2000), but questions remain about how widespread this increase may be and why it occurs and, more generally, how the course of well-being relates to the various diverse pathways out of high school. Substance use also tends to increase during this period, reaching its lifetime peak during the early twenties, depending on the given cohort and substance (Chen and Kandel 1995; Jackson et al. 2002; Johnston, O̓Malley, and Bachman 2003). While it has been found that changes in substance use relate to various social role transitions during emerging adulthood (e.g., Bachman et al. 1997; Brook et al. 1999; Jessor, Donovan, and Costa 1991), questions remain about how various transitions work together in contributing to increases and decreases in substance use during this time. Well-being and substance use, while not necessarily sharing a common etiology or developmental course across the life span, may increase during the transition to adulthood in part because of the new roles and contexts that provide more freedom and selection of opportunities (Bachman et al. 1997; Schulenberg and Maggs 2002). Furthermore, while substance use has clear negative and often dangerous correlates and consequences (e.g., Hawkins, Catalano, and Miller 1992), experimental substance use during late adolescence may also serve constructive purposes in regard to developmental tasks related to, for example, peer bonding, independence striving, and identity experimentation (Chassin, Presson, and Sherman 1989; Maggs 1997; Schulenberg, Maggs, and O̓Malley 2003a).
In this chapter, we analyze data from four waves of nationally representative U.S. panel data spanning ages eighteen to twenty-four, and we offer a “big picture” about the timing, sequencing, and covariation of social role transitions related to school and work, romantic involvement (specifically marriage), parenthood, and independence in the form of leaving the parental home. At wave 1 in our study, young people were nearing the end of their senior year of high school (modal age of eighteen), allowing us to follow their “launching” into post-high school transitions. During this important launching period, initial plans first combine with new experiences to place individuals on paths that will lead them into adulthood (Clausen 1991; Gore et al. 1997). In aggregating across these specific transitions at wave 2 (modal ages of nineteen to twenty) to construct mutually exclusive transition groups, we focus on both the number of transitions and the distinct patterning of various transitions, defining and offering prevalence estimates of the multiple pathways through emerging adulthood. Building on some of our previous research (Schulenberg et al. 2000), we consider associations between the wave 2 transition groups (i.e., aggregated (p.419) by number and by unique patterns) and trajectories of well-being and substance use across the four waves (spanning ages eighteen to twenty-four). And in the last phase of the analyses, we examine diversity within transition groups, focusing specifically on how the differential transitional experiences that occur between waves 2 and 4 relate to trajectories of well-being and substance use.
In conceptualizing how the different transition groups might relate to trajectories of well-being and substance use, we draw from Coleman's focal theory (1989) regarding transition effects during early adolescence. According to focal theory, the number of transitions a young person goes through relates to the amount of difficulty the young person experiences; that is, numerous and simultaneous transitions can overwhelm one's coping capacity, and well-being can suffer (Schulenberg and Maggs 2002). Thus, it might be expected that the number of transitions in the year or two immediately following high school is negatively related to well-being and positively related to substance use. But it is also possible that those most willing to take on more transitions at once might have more psychological resources to begin with, suggesting an opposite direction of relations. More broadly, we draw from Elder's (1998) conceptualizations concerning the social life course and Rutter's (1996) conceptualizations regarding transitions as potential turning points with regard to ongoing functioning and adjustment (for additional details on our conceptual approach, see Schulenberg et al. [2003a]).
Our approach is largely descriptive, which is appropriate given that our purpose is to map the milestones of the broader critical developmental transition from adolescence to adulthood. But we also try to provide some preliminary explanations of our findings. The different pathways and their relations to trajectories of well-being and substance use may vary by gender, cohort membership, and race/ethnicity (Schulenberg et al. 2000), and we investigate these possibilities in our analyses. We take a pattern-centered (rather than single variable-centered) approach to considering the different transitions. Such an interaction-based approach to change (see Cairns and Cairns 1994; Magnusson 1995; Singer et al. 1998) seeks to extend previous maineffects findings, such as the effects of marriage and living away from parents, that we and others have demonstrated in previous analyses (Bachman et al. 1997; Graber and Dubas 1996; Leonard and Rothbard 1999; Schulenberg et al. 2000). This pattern-centered approach is more complex than typical variable-centered (main effects) approaches, but the additional complexity is warranted given that certain transitions tend to co-occur during emerging adulthood. (p.420)
We examine national panel data spanning ages eighteen to twenty-four from the Monitoring the Future (MTF) project (Johnston et al. 2003), which is an ongoing, cohort-sequential longitudinal project funded by the National Institute on Drug Abuse. It is designed to understand the epidemiology and etiology of substance use and, more broadly, behavior and psychosocial development during adolescence and young adulthood. The project has surveyed nationally representative samples of approximately 17,000 high school seniors in the United States each year since 1975, using questionnaires administered in classrooms. Approximately 2,400 individuals are randomly selected from each senior year cohort for follow-up. Follow-up surveys are conducted by mail every two years.
The panel sample used in the present study consisted of nineteen consecutive cohorts of respondents who were surveyed as high school seniors (wave 1, age eighteen) from 1977 through 1995, and who participated in follow-up surveys one or two years after high school (wave 2, ages nineteen to twenty), three or four years after high school (wave 3, ages twenty-one to twenty-two), and five or six years after high school (wave 4, ages twenty-three to twenty-four). Differences in year of follow-up occur because the biennial follow-up surveys begin one year after high school for one random half of the panel drawn from each cohort, and two years post high school for the other. For these analyses, the two random halves were combined.
To increase the breadth of areas covered by the surveys, MTF uses six different questionnaire forms (questionnaires are distributed randomly within schools at senior year, and a given individual's questionnaire form is consistent across waves). Because the items that comprise the well-being measure are located on only one of the forms, only one-sixth of the sample was available for the present study. This included 3,912 weighted cases (1,666 males and 2,243 females). Drug users are oversampled for follow-up, and corrective weighting is used to reflect population estimates.
We focus primarily on transitions that occur during the first year or two immediately following high school (by wave 2, ages nineteen to twenty), an appropriate time frame given our emphasis on the launching into emerging (p.421) adulthood. We examine longitudinal trajectories of well-being and substance use in order to try to capture the course of these constructs prior to, during, and after the wave 2 transitions. This makes it possible to consider selection effects, as well as to examine whether the transitions serve to alter the ongoing trajectories of substance use and well-being. We selected the age twenty-three to twenty-four survey as the final (fourth) wave because this age is beyond the normative ending time for full-time college attendance (age twenty-two for most of the cohorts included here), allowing us to consider postcollege experiences.
Transitions at Wave 2
We consider a variety of transitions that occur between wave 1 (age eighteen, senior year in high school) and wave 2 (ages nineteen to twenty), including entering college, entering the workforce, leaving the parental home, getting married, and entering parenthood. Seven transitions were examined based on items concerning full-or part-time college attendance during the past year, full-or part-time employment during the past year, current living arrangements (specifically, living in parental home), current marital status, and parenthood. This is not a list of mutually exclusive transitions, of course, nor is it a comprehensive list of all of the important milestones during this period. But it is a reasonable group of normative social-role transitions that reflects the diversity of life paths during this launching period. We aggregated across the various transitions, in terms of their number and patterning, to form mutually exclusive transition groups (details provided below).
Gender, Cohort, and Race/Ethnicity
We considered gender, cohort, and race/ethnicity effects, particularly how they impinged on relationships between transition groups and trajectories of well-being and substance use. Senior-year classes were grouped into three cohorts (1977–82, 1983–89, 1990–95). Given our emphasis on multiple transition groups, our available sample, and the national political and substance-use cycles over the two-decade period (Johnston et al. 2003;Schulenberg et al. 2000), these cohort groups reflect logical and meaningful categories. Race/ethnicity was considered in terms of white (83% of the sample), African American (8%), and other race/ethnicity groups (9%, the majority of whom were Hispanic American). This three-way grouping is less than satisfying in some ways, but given the sample size and focus on multiple transition groups, it was our best option.
Based on previous analyses (Schulenberg et al. 2000)and the work of Ryff and colleagues on well-being during adulthood (p.422) (e.g., Ryff and Keyes 1995), overall well-being was considered in terms of a composite of three interrelated constructs: self-esteem (based on Rosenberg 1965), self-efficacy (similar to Nowicki and Strickland's  internal locus of control subscale), and social support (similar to Newcomb and Harlow 1986). Each item was a statement about one's self (e.g., I feel I am a person of worth); for all items, possible responses were 1 (disagree), 2 (mostly disagree), 3 (neither agree nor disagree), 4 (mostly agree), and 5 (agree), with responses reversed if necessary so that high scores reflect high well-being. The same measures were used at all four waves, and cross-sectional exploratory factor analyses of the three scales suggested one underlying dimension at each wave. Alpha coefficients for this overall score exceeded .75 at each of the four waves.
Substance use measures for these analyses included binge drinking (frequency of having five or more drinks in a row during the past two weeks) and marijuana use (occasions of use in the past twelve months). The Monitoring the Future substance-use items have been shown to demonstrate excellent psychometric properties, and their reliability and validity have been reported and discussed extensively (Johnston and O̓Malley 1985; Johnston et al. 2003; O̓Malley, Bachman, and Johnston 1983). Possible responses for occasions of binge drinking in the past two weeks were 1 (none), 2 (once), 3 (twice), 4 (3–5 times), 5 (6–9 times), and 6 (10 or more times); for occasions of marijuana use in the past twelve months possible responses were 1 (0 occasions), 2 (1–2 occasions), 3 (3–5 occasions), 4 (6–9 occasions), 5 (10–19 occasions), 6 (20–39 occasions), and 7 (40 or more occasions). The same measures were used at all four waves.
To address the aims of this chapter, there were five phases of analysis, which dealt with (1) average trajectories of well-being and substance use (binge drinking and marijuana use) across the four waves (ages eighteen to twenty-four); (2) description of wave 2 (ages nineteen to twenty) transitions; (3) how the number of wave 2 transitions relate to trajectories of well-being and substance use; (4) how the wave 2 transition groups relate to trajectories of well-being and substance use; and (5) within specific wave 2 transition groups, how wave 4 (ages twenty-three to twenty-four) transitions relate to trajectories of well-being and substance use.
We relied typically on within-time and repeated-measures ANOVAs, considering transition groups (along with gender, cohort, and race/ethnicity) as the predictors, and substance use and well-being (within-time and (p.423) across time) as the outcomes. Despite the implied causal ordering in the analyses, bidirectional influences very likely occur between the transition groups and dependent variables; ANOVAs provide a straightforward way of connecting a categorical variable (transition groups) with longitudinal trajectories of continuous variables (substance use and well-being). In the repeated-measures ANOVAs, time effects (i.e., change across the four waves) were partitioned into orthogonal polynomial contrasts to test for linear, quadratic, and cubic effects in well-being and substance use over time. The time-interaction terms provided tests of whether and how the transition groups (and gender, cohort, and race/ethnicity) were associated with different trajectories of well-being and substance use. For significant time-by-transition group interactions, comparisons were made among the change coefficients of the various subgroups to determine significant differences.
Clearly, given the wealth of findings yielded by the analyses, not all findings can be presented here. To simplify our presentation in this chapter, and consistent with the overall approach of this volume, we focus primarily on the patterns of significant findings relevant to how the transition groups relate to well-being and substance-use trajectories. (A full report of all findings and additional detail about sample and measures is provided in MTF Occasional Paper 56 [Schulenberg et al. 2003b].)
Findings are presented according to the five analysis phases. We limit our consideration of findings to those differences and changes over time that were significant at least at the p 〈 .01 level (a level justified by the size of our sample and the number of analyses conducted).
Average Trajectories of Well-Being and Substance Use
We start by examining average trajectories of well-being and substance use (binge drinking and marijuana use) across four waves from senior year in high school to ages twenty-three to twenty-four for the total sample and by gender, cohort, and race/ethnicity.
As shown in figure 13.1, well-being increased across the waves, with a faster rate of change between earlier waves than later waves. Men and women started with identical levels of well-being, but the increase over time was significantly greater for men than for women, and the leveling off with age was stronger for men than for women. There were no significant cohort or race/ethnicity interaction effects with time.1 (p.424)
Binge drinking tended to increase immediately following high school and was consistently higher for men than women; this gender difference increased with age, as shown in figure 13.2a. Figure 13.2b shows that binge drinking varied as a function of cohort group, with the three groups starting off quite differently in terms of initial level of binge drinking at age eighteen but then converging by ages twenty-one to twenty-two. Specifically, binge drinking for the most recent cohort group (1990–95) increased more rapidly over time than for the other cohort groups, and all groups followed a quadratic trend in which binge drinking peaked by wave 3 or 4 and then decreased. Binge drinking varied by race/ethnicity, with binge drinking being higher for whites than for African Americans and the other race/ethnicity groups; whereas the trajectory for whites increased then decreased, the trajectories for African American and other race/ethnicity groups remained flat.
The findings for marijuana use are very similar to those for binge drinking. As shown in figure 13.3a, marijuana use was, on average, higher for men than for women, and women decreased their use more rapidly over time than did men. Figure 13.3b indicates that the overall level and trajectory (p.425)
Overall, then, well-being was found to increase during the transition, especially over the first few years out of high school. This was true for both men and women, although the rate of increase was faster for men. These time trends held regardless of cohort or race/ethnicity. Binge drinking and marijuana use were, on average, higher for men than women, and higher for whites than African American and other race/ethnic groups. Cohort effects were striking for the trajectories of substance use (see Johnston et al. 2003), with evidence of convergence across cohorts during the mid-twenties when substance use declined for all cohorts (although the oldest cohort group maintained its higher level of marijuana use).
Transitions at Wave 2 (Ages Nineteen to Twenty)
In the second phase of the analyses, we examined the percentages of individuals in our national panels making the various post-high school transitions between waves 1 (age eighteen) and 2 (ages nineteen to twenty). These percentages are shown in figure 13.4 by gender. Note that these are not mutually exclusive transitions, with the exception of full-versus part-time work and full-versus part-time college. The most common post-high school transition was entering full-time college, with almost 60% of the sample doing this. Only 8% were attending college part time. About 33% of the men and 25% of the women made the transition into full-time work, and another 29% of the men and 35% of the women were working part time. (Part-time work does not necessarily represent a transition, given that most had worked part time during high school; however, for other purposes considered below, we wanted to include post-high school part-time work as an important activity.) Moving away from one's parents was very common, with about half the sample doing so and the other half living with one or both parents. Only about 10% of the women and 5% of the men were married by wave 2, and 7% of the women and 4% of the men had children. Significant cohort differences were evident for full-time work and full-time college, with the former decreasing and the latter increasing from earlier to more recent cohorts (see also Bachman et al. 1997). In terms of significant race/ethnicity differences, (p.428)
We next considered two ways of aggregating across the individual transitions and then examined how the aggregates related to well-being and substance use. First, the number of transitions was simply summed (presented in the next section). Second, we considered all possible combinations of transitions and focused only on those combinations that encompassed sufficient portions of the sample to permit meaningful consideration.
Number of Transitions and Well-Being and Substance Use
A straightforward way of thinking of the transitions in aggregate is to sum the number of transitions a given individual makes at wave 2. As we discussed earlier, this approach draws from Coleman's focal theory (1989) in which the number of transitions a young person goes through during early adolescence is negatively related to well-being and positively related to difficulties; conversely, especially during the transition from adolescence into early adulthood, those most willing to take on more transitions at once might have more psychological resources to begin with, suggesting an opposite direction of relations.
The number of transitions any one individual could make ranged from zero to five (although there are seven possible transitions, (p.429) two mutually exclusive pairs—part- and full-time work and part- and full-time school—make five the top of the range). The mode for men and women was two transitions (49% and 48%, respectively), followed by one transition (29% of men and 26% of women), and then three (15% of men and 20% of women). About 5% of the sample experienced no transitions; at wave 2 they were still living with their parents, were not married, had no children, were not enrolled in college full or part time, and were not working full or part time. Less than 2% experienced four or five transitions. There were no significant gender, cohort, or race/ethnicity differences in the average number of transitions.
Figure 13.5 shows the trajectories of well-being over time by the number of wave 2 transitions (the arrow at wave 2 signifies when transition groups are defined). Well-being increased for all groups between waves 1 and 2. It continued to increase for the one, two, and three transition groups across the waves and leveled off (quadratic effect) for the no and four/five transition groups. With only minor exceptions, the transition groups maintained their relative ordering across the four waves, with well-being scores overall being significantly higher than average for those making two and three transitions and significantly lower than average for those making no transitions and one transition. Transition group interactions involving gender, cohort, and race/ethnicity were not significant. The fact that the differences in well-being were in place at wave 1 prior to graduating from high school indicates a selection effect: those who are higher in well-being in high
Figures 13.6 and 13.7 show the trajectories of binge drinking and marijuana use, respectively, for the transition groups. In contrast to what was found for the well-being trajectories, the substance-use trajectories show a fair amount of differential change as a function of number of transitions. Those in the no, one, and four/five transition groups had significantly higher binge drinking and marijuana use at wave 1 than did those in the two and three transition groups. As shown in figure 13.6, the binge drinking trajectory remained relatively flat for those in the no and one transition groups, decreased sharply for those in the four/five transition group, and increased then decreased for those in the two and three transition groups. As shown in figure 13.7, very similar results were found with regard to marijuana-use trajectories; it is noteworthy that marijuana use did not decline over time for the no transition group. For both binge drinking and marijuana use, none of the time-by-transition group interactions involving gender, cohort, or race/ethnicity was significant.
Overall, there was little evidence to suggest that experiencing more transitions immediately following high school contributes to poorer functioning and adjustment. Indeed, well-being tended to be higher
Transition Groups: Well-Being and Substance Use
Construction and Prevalence of Transition Groups
In this fourth phase of the analyses, we wanted to assemble a limited set of naturally occurring, mutually exclusive configurations of various transitions and then consider the trajectories of well-being and substance use as a function of these constructed transition groups. This was a potentially cumbersome process, for up to 240 unique categories (i.e., 2(5!)) were possible. But as shown in figure 13.8, we were able to construct nine mutually exclusive transition groups, with a tenth “unclassified” group (not illustrated).
Making some logical decisions, we began this analysis by isolating the (p.432)
As shown in figure 13.8, the other groups that we found to encompass sufficient numbers of young people included three groups who were similar in terms of not being married, not having children, and living away from parents at wave 2: those who attended college full time and worked full or part time (11% of men and 13% of women), those who attended college full time and did not work (23% of men and 20% of women), and those who worked (p.433) full or part time and did not attend college (6% of men and 5% of women). These three groups were analogous to another set of three, with the difference being that this second set of three lived home with one or both parents at wave 2: those who attended college full time and worked full or part time (14% of men and women), those who attended college full time and did not work (8% of men and women), and those who worked full or part time and did not attend college (17% of men and 13% of women).
These nine transition groups were mutually exclusive, and together, they accounted for 90% of the sample (leaving 10% in the unclassified group). Across these nine transition groups, prevalence rates did not vary significantly by gender, cohort, or race/ethnicity.
The well-being trajectories of the nine wave 2 transition groups are illustrated in figure 13.9. The trajectory for the unclassified group is not shown, but this group was included in the analyses. As is clear, well-being increased for each group, and the transition groups generally maintained their relative ranking in well-being over time, once again indicating selection effects. Compared to the total sample, well-being was significantly higher in the “not married, no children, live away, college and work,” “not married, no children, live away, college only,” and “not married, no children, live with parent, and college and work” groups; and it was significantly lower in the two “work-only” groups and the no transition and single parent groups. Interactions involving gender, cohort, and race/ethnicity were not significant.
The trajectories for binge drinking and marijuana use for the nine wave 2 transition groups are illustrated in figures 13.10 and 13.11, respectively. In both analyses, the between-subjects effect for transition group was significant, the time-by-transition interaction was significant for both linear and quadratic trends, and none of the interactions involving gender, cohort, or race/ethnicity was significant.
Overall across the waves, compared to the total sample, binge drinking was significantly higher in the two “work-only” groups (who also had the highest level of binge drinking at wave 1) and the “live away, college only” group. It was significantly lower in the “married, live away” group. Compared to the total sample trajectory (see figure 13.2a), the binge-drinking trajectory decreased for the “not married, no children, live away, work-only group,” decreased sharply then leveled off for the “married, live away” group, and increased then decreased for the “not married, no children, live away, college only” and “not married, no children, live away, college and work” (p.434)
Marijuana use overall—across the waves and compared to the total sample—was significantly higher in the two work-only groups and significantly lower in the two remaining groups that were “not married, had no children, and lived with parents” (the college and work, and work-only groups). Compared to the total sample trajectory (see figure 13.3a), the trajectories for the “not married, no children, live away, and college only” and “not married, no children, live away, and college and work” groups showed significantly less linear decrease and greater negative quadratic effect; the trajectory for the “not married, no children, live with parent, and work-only” group showed a significantly greater linear decrease; and the trajectory for the “married, live away” group showed a positive quadratic effect, reflecting the sharp decrease with marriage at wave 2 with this group.
While differences were evident in well-being trajectories across the nine transition groups, these differences were in place at wave 1, indicating selection effects. In general, well-being was higher for those who, at wave 2, were not married, did not have children, and were in college. More differential change as a function of the nine transition groups was evident in the trajectories of binge drinking and marijuana use. In particular, for both binge drinking and marijuana use, there are sharper increases then decreases over time for those who at wave 2 are attending full-time college, not living with parents, not married, and have no children and sharper decreases for those who at wave 2 were married and not living with parents.
Examining Wave 4 Transitions within Wave 2 Transition Groups
In the final set of analyses, we wanted to examine what happened at wave 4 for some of the key wave 2 transition groups. In keeping with our pattern-centered approach, we looked within specific wave 2 groups, or appropriate combinations of “adjacent groups,” and examined how wave 4 transitions related to variations in well-being and substance-use trajectories. We considered two groups that involved sufficiently large segments of the sample: (a) the wave 2 “living away, not married, no children, full-time college” combined group, which included working and nonworking subgroups and contained about 33% of the sample (1,294 individuals); and (b) the wave 2 “not married, no children, no college, work-only” combined group, which included both those living away and those living with parents and contained about 20% of the sample (787 individuals). In these analyses, we considered (p.438) gender-by-transition-by-time interactions, none of which was significant, but were unable to consider interactions involving cohort and race/ethnicity due to sample size limitations.
Wave 2 “Live Away, not Married, no Children, Full-Time College” Group
All individuals in this group were (at wave 2) enrolled full time in college, lived away from home, were not married, and had no children (n = 1,294). Based on the transitions that had occurred at wave 4 (ages twenty-three to twenty-four), we formed six groups and a seventh unclassified group. Three of the groups were similar in that their members still lived away from parents, were not married, did not have children, and either worked only (27%), attended college/graduate school only (12%), or worked and attended college/graduate school (14%). The remaining groups were the following: lived away from parents, married, no children, working (13%); moved back with parent(s), not married, no children (most completed college and were working full time) (20%); lived away from parents, with children (5%); and unclassified (9%).
Figures 13.12, 13.13, 13.14 show the trajectories of well-being, binge drinking, and marijuana use, respectively, for the six different wave 4 trajectory groups (the arrow at wave 4 signifies when transition groups are defined). As shown in figure 13.12, well-being increased for all six groups, especially across the earlier waves. The six groups did not differ from each other in their levels or trajectories of well-being, with one exception: the trajectory for those who remained in college full time, including graduate school, did not level off between waves 3 and 4 (significant cubic effect).
Figures 13.13 and 13.14 demonstrate graphically that substance-use levels were fairly equivalent across the six subgroups at wave 1 and then began to diverge considerably at wave 2 when everyone was still a full-time student living away from home and was neither married nor had children. Of particular interest, binge drinking and marijuana use at wave 2 were significantly lower compared to the group total for those who subsequently got married by wave 4, suggesting that that substance use at wave 2 foreshadows a quicker subsequent entry into marriage and, for binge drinking only, a quicker subsequent entry into parenthood. This is consistent with our earlier finding that becoming engaged is associated with—and perhaps causal of—declines in substance use (Bachman et al. 1997). More generally, binge drinking increased more rapidly for those groups who at wave 4 still lived away from home, were not married, and had no children; and it decreased more rapidly for the wave 4 “live away, married, no children, working” and “live away, with children” groups. And whereas marijuana use generally (p.439)
Wave 2 “Not Married, No Children, No College, Work Only” Group
All individuals in this group were (at wave 2) employed full time, were not in college full or part time, lived away from home, were not married, and had no children (n = 787). Based on consideration of what transitions had occurred by wave 4 (ages twenty-three to twenty-four), we formed five groups (plus a sixth unclassified group): lived away, worked full time, not married, and had no children (22%); lived away, worked full time, married, and had no children (14%); lived away, worked full time, married, had children (13%); lived away and neither attended college nor worked (10%); lived with parent(s), worked full time, not married, and had no children (22%); and unclassified (18%).
Figures 13.15, 13.16, 13.17 show the trajectories of well-being, binge drinking, and marijuana use, respectively, for the five different wave 4 trajectory groups. For well-being, all five groups had similar well-being scores at wave 1, and over time, only the well-being trajectory of the “live away, no college, not working” group was different than the trajectories for the other groups. Specifically, well-being declined between waves 3 and 4 to a greater extent for this group than for the total, which very likely relates to this group neither working nor attending college at wave 4.
For both the binge drinking and marijuana-use trajectories, the transition groups most different from the others were the wave 4 “live away, work, not married, no children” and “live away, work, married, with children” groups: compared to the total, the former group had significantly higher-than-average substance use across waves, and the latter group had significantly lower-than-average substance use across the waves. Of particular interest concerning the “foreshadowing” (and likely engagement effect) mentioned earlier in the other subgroup analysis, the two groups that were the same except for marriage at wave 4 (i.e., lived away, had no children, worked full time) had similar levels of binge drinking at wave 1; then they quickly diverged in binge drinking by wave 2—when both groups were working full time, were not attending college, were not married, and had no children—and remained significantly different at waves 3 and 4.
On the whole, these two final analyses showed that within homogenous transition groups defined at wave 2, differences in subsequent transitional experiences between waves 2 and 4 were associated with divergences in trajectories of substance use and, to a lesser extent, of well-being. (p.443)
(p.446) And in the case of substance use, some of these divergences were evident at wave 2 when the groups were homogenous with respect to the various transitions; in particular, in both sets of comparisons, a greater decline in substance use—especially binge drinking—between waves 1 and 2 fore-shadowed a greater likelihood of marriage by wave 4. Divergences in well-being trajectories in both sets of comparisons were limited but telling: in the wave 2 “live away, not married, no children, full-time college” group, well-being leveled off between waves 3 and 4 for all subgroups except those who remained at wave 4 in college or graduate school full time and were not working; in the wave 2 “not married, no children, no college, work-only” group, well-being dropped between waves 3 and 4 only for those who at wave 4 were neither working nor attending college and were living away from parents.
Before discussing specific findings regarding relations between various transitions and the well-being and substance-use trajectories, it is important to note first that even with the lengthening of the period between adolescence and adulthood (Arnett 2000) and the increased diversity in pathways (Shanahan 2000), our findings suggest that any attempt to understand emerging adulthood would benefit from considering traditional indicators of developmental milestones such as marriage and full-time work. Indeed, given the foreshadowing we found, these milestones represent more than simply external markers. What is especially important about these developmental milestones is how they work together—the fact that out of 240 possible combinations of transitions we were able to place 90% of our sample in one of nine mutually exclusive transition groups suggests a deliberate patterning of transitions (see Cairns and Cairns 1994). Furthermore, while there were some significant gender, cohort, and race/ethnicity differences in transitions and in trajectories of well-being and substance use, interactions with transition groups were in large part nonsignificant, indicating that the links between transition experiences and the trajectories were fairly pervasive and did not vary as a function of gender, cohort, and race/ethnicity (at least in the late twentieth century in the United States).
Trajectories of Well-Being
Looking at the sample as a whole, well-being increased during the first few years out of high school for both men and women (at a faster rate for men) and then began to level off by the midtwenties. This was true regardless (p.447) of race/ethnicity or cohort. Linking well-being to the number of transitions one makes revealed some rather surprising findings. Although, as with the sample as a whole, well-being increased steadily for each group examined (i.e., those making no, one, two, three, or four or more transitions), those making more transitions had consistently higher well-being. This effect suggests that the mechanisms of stress suggested by focal theory (described previously)—in which the numerous and simultaneous transitions of early adolescence may overwhelm one's coping capacity—are not likely to be operating here. Perhaps the difference here is that, for life after high school, young people have more choice in their transitions and new social roles, thus increasing the match between what they wish to do and available opportunities. While interesting, an exclusive focus on the number of transitions falls short of offering insight into the underlying processes.
In considering the patterning of the transitions, focusing on nine mutually exclusive transition groups based on the social role changes they experienced by wave 2 (ages nineteen to twenty), we again found that well-being increased for all groups. The groups with the highest well-being were those who, at wave 2, had not yet married, had no children, lived away from home, and were attending college, with or without combining work. Those with significantly lower well-being were, at wave 2, single parents, those working without attending college, and those who had yet to make any transition. The nine transition groups tended to maintain their relative ranking in well-being over time (see fig. 13.9). This stability of intergroup differences strongly suggests a selection effect, in which well-being or a correlate of well-being, like academic achievement (Clausen 1991) during the senior year of high school or before, sets the stage for the type/patterning of transitions one plans to take on after high school. Nevertheless, when looking within groups defined at one point in time (our wave 2) to consider how within-group diversity in life paths unfolds over time (our wave 4), we learned that the course of well-being can be somewhat sensitive to the experience of transitions, particularly those related to achievement domains of school and work; such patterns suggest that the course of well-being is not entirely a function of selection effects.
The fact that nearly all transition groups, including the group that did not experience any transitions by wave 2, showed an increase in well-being over time suggests that there is some “niche picking” going on, with young people selecting the transitions or experiences that match best with their developmental needs and desires (Schulenberg and Maggs 2002). More broadly, the increase in well-being for all groups suggests the utility of Baltes's (1987) selection, optimization, and compensation life span development (p.448) model for understanding how young people successfully negotiate the many changes and demands of emerging adulthood (see also Schulenberg et al., in press; Wiese et al. 2000).
Trajectories of Substance Use
For the group as a whole, substance use (marijuana and alcohol) among men and women tended to peak by the early twenties, although use among men was consistently higher than among women. The level and trajectory of substance use varied by cohort (see Johnston et al. 2003). The earlier cohorts had the highest levels of use, but within cohorts, use declined as the youth aged. Both binge drinking and marijuana use were highest among whites. The trajectory of binge drinking for whites increased and then decreased over time, while for African Americans and the other race-ethnicity groups, the trajectories were flat.
Considering the number of transitions, substance-use trajectories were a bit more varied than well-being trajectories were. Those who, by age nineteen to twenty, had already experienced four or more transitions saw the steepest decline in substance use, suggesting the effect of a combination of transitions, although marriage likely had the strongest effect (Bachman et al. 1997). It is interesting that those who made no transitions by wave 2 (ages nineteen to twenty) had a relatively high and flat trajectory of marijuana use across the waves, suggesting some effects of avoiding the tasks of early adulthood. But again, while interesting, the focus on the number of transitions is unsatisfying in regard to possible processes that connect transitions to trajectories of substance use.
As we found, trajectories of well-being are influenced considerably by the specific patterning of transitions. Certain post-high school contexts or experiences, specifically living away from home and not being married, contribute to a relative increase and delayed decrease in substance use (see Bachman et al. 1997). This provides additional evidence that the emerging adulthood period is a time of experimentation (Arnett 2000) and that once typical adulthood roles are assumed, experimentation tends to be left behind. One could also explain this pattern by means of changed willingness to take risk and/or associated changes in constraining social influences (e.g., presence of parents, fiancé, and/or spouse). Furthermore, declines in substance use appear to foreshadow upcoming transitional experiences that move the individual more firmly into adulthood status. In general, the findings that the course of substance use was more influenced by the transitions than was the course of well-being suggest that while many transitions do indeed serve as turning points, such turning point influences are not necessarily (p.449) pervasive with respect to multiple indications of functioning and adjustment (Rutter 1996; Schulenberg and Maggs 2002).
Strengths and Limitations
Strengths of this study include the use of U.S. national multicohort panel data spanning the transition to young adulthood; limitations include measure limitations, the restricted set of transitions, and some degree of imprecision in defining transition groups (e.g., we may have missed some important events during the two-year lag between waves). The pattern-centered, interaction-based approach is both a strength and a limitation. Clearly, our “big picture” approach works best in combination with other more finegrained studies that can provide more of the interesting detail about life's milestones and processes of change during this period of life.
The global transition to adulthood can serve as an important proving ground where one's accumulated talents, support, and hopes interact with the new opportunities and challenges of post-high school life. For most young people, the trajectories of functioning and adjustment established through-out childhood and adolescence likely carry into emerging adulthood and work, together with (or against) the pervasive changes that may come with this transition, yielding continuity in overall functioning and adjustment into adulthood. But this transition period can also serve as a turning point for many young people, a time when established trajectories of functioning and adjustment change direction (for better or worse) due in part to the experiences of emerging adulthood. In this study, we found extensive mean level changes in well-being and substance use during emerging adulthood, with considerable differential change in substance use as a function of transition group. We also found considerable continuity in well-being in terms of a general lack of differential change as a function of transition group. The diverse pathways from adolescence to adulthood are rooted in earlier experiences and plans that set the stage for continuity in well-being, but the experiences of the different pathways contribute to discontinuities in substance use.
Arnett, Jeffery J. 2000. Emerging Adulthood: A Theory of Development from the Late Teens through the Twenties. American Psychologist 55:469–80.
Bachman, Jerald G., Katherine N. Wadsworth, Patrick M. O̓Malley, Lloyd D. Johnston, and John E. Schulenberg. 1997. Smoking, Drinking, and Drug Use in Young Adulthood: The Impact of New Freedoms and New Responsibilities. Mahwah, NJ: Lawrence Erlbaum Associates.
Baltes, Paul B. 1987. Theoretical Propositions of Life-Span Developmental Psychology: On the Dynamics between Growth and Decline. Developmental Psychology 23:611–26.
Brook, Judith S., Linda Richter, Martin Whiteman, and Patricia Cohen. 1999. Consequences of Adolescent Marijuana Use: Incompatibility with the Assumption of Adult Roles. Genetic, Social, and General Psychology Monographs 125: 193–207.
Cairns, Robert B., and Beverley D. Cairns. 1994. Lifelines and Risks: Pathways of Youth in Our Time. New York: Cambridge University Press.
Chassin, Laurie, Clark C. Presson, and Steven J. Sherman. 1989. “Constructive” vs. “Destructive” Deviance in Adolescent Health-Related Behaviors. Journal of Youth and Adolescence 18:245–62.
Chen, Kevin, and Denise B. Kandel. 1995. The Natural History of Drug Use from Adolescence to Mid-thirties in a General Population Sample. American Journal of Public Health 85:41–47.
Clausen, John A. 1991. Adolescent Competence and the Shaping of the Life Course. American Journal of Sociology 96:805–42.
Cohen, Patricia, Stephanie Kasen, Henian Chen, Claudia Hartmark, and Kathy Gordon. 2003. Variations in Patterns of Developmental Transitions in the Emerging Adulthood Period. Developmental Psychology 39:657–69.
(p.451) Coleman, John C. 1989. The Focal Theory of Adolescence: A Psychological Perspective. Pp. 43–56 in The Social World of Adolescents: International Perspectives, edited by Klaus Hurrelmann and Uwe Engel. Berlin: Walter de Gruyter.
Elder, Glen H., Jr. 1998. The Life Course and Human Development. Pp. 939–91 in Handbook of Child Psychology. Vol. 1, Theoretical Models of Human Development, edited by Richard Lerner. New York: Wiley.
Gore, Susan, Robert Aseltine Jr., Mary Ellen Colten, and Bin Lin. 1997. Life after High School: Development, Stress, and Well-being. Pp. 197–214 in Stress and Adversity over the Life Course: Trajectories and Turning Points, edited by Ian H. Gotlib. New York: Cambridge University Press.
Graber, Julia A., and Judith Semon Dubas, eds. 1996. Leaving Home: Understanding the Transition to Adulthood. San Francisco: Jossey-Bass.
Hawkins, J. David, Richard F. Catalano, and Janet Y. Miller. 1992. Risk and Protective Factors for Alcohol and Other Drug Problems in Adolescence and Early Adulthood: Implications for Substance Abuse Prevention. Psychological Bulletin 112: 64–105.
Jackson, Kristina M., Kenneth J. Sher, M. Lynn Cooper, and Phillip K. Wood. 2002. Adolescent Alcohol and Tobacco Use: Onset, Persistence and Trajectories of Use across Two Samples. Addiction 97:517–31.
Jessor, Richard, John E. Donovan, and Frances M. Costa. 1991. Beyond Adolescence: Problem Behavior and Young Adult Development. New York: Cambridge University Press.
Johnston, Lloyd D., and Patrick M. O̓Malley. 1985. Issues of Validity and Population Coverage in Student Surveys of Drug Use. Pp. 31–54 in Self-report Methods of Estimating Drug Use: Meeting Current Challenges to Validity, edited by Beatrice A. Rouse, Nicholas J. Kozel, and Louise G. Richards. NIDA Research Monograph No. 57. Washington DC: National Institute on Drug Abuse.
Johnston, Lloyd D., Patrick M. O̓Malley, and Jerald G. Bachman. 2003. National Survey Results on Drug Use from the Monitoring the Future Study, 1975–2002. Vol. 1, Secondary School Students. Vol. 2, College Students and Young Adults. NIH Publication No. 03–5375 and 03–5376. Bethesda, MD: National Institute on Drug Abuse.
Leonard, Kenneth E., and Julie C. Rothbard. 1999. Alcohol and the Marriage Effect. In “Alcohol and the Family: Opportunities for Prevention.” Special Issue, Journal of Studies on Alcohol 13:139–46.
Maggs, Jennifer L. 1997. Alcohol Use and Binge Drinking as Goal-Directed Action during the Transition to Post-Secondary Education. Pp. 345–71 in Health Risks and Developmental Transitions during Adolescence, edited by John Schulenberg, Jennifer L. Maggs, and Klaus Hurrelmann. New York: Cambridge University Press.
(p.452) Magnusson, David. 1995. Individual Development: A Holistic, Integrated Model. Pp. 19–60 in Examining Lives in Context: Perspectives on the Ecology of Human Development, edited by Phyllis Moen, Glen H. Elder Jr., and Kurt Lüscher. Washington DC: American Psychological Association.
Masten, Ann S., and W. John Curtis. 2000. Integrating Competence and Psychopathology: Pathways toward a Comprehensive Science of Adaptation in Development. Development and Psychopathology 12:529–50.
Newcomb, Michael D., and Lisa L. Harlow. 1986. Life Events and Substance Use among Adolescents: Mediating Effects of Perceived Loss of Control and Meaninglessness in Life. Journal of Personality and Social Psychology 51:564–77.
Nowicki, Stephen, and Bonnie R. Strickland. 1973. A Locus of Control Scale for Children. Journal of Consulting and Clinical Psychology 40:148–54.
Oerter, Rolf. 1986. Developmental Tasks through the Life Span: A New Approach to an Old Concept. Pp. 233–71 in Life Span Development and Behavior, edited by Paul B. Baltes, David L. Featherman, and Richard M. Lerner. Vol. 7. Hillsdale, NJ: Lawrence Erlbaum.
O̓Malley, Patrick M., Jerald G. Bachman, and Lloyd D. Johnston. 1983. Reliability and Consistency of Self-reports of Drug Use. International Journal of the Addictions 18: 805–24.
Rosenberg, Morris. 1965. Society and the Adolescent Self-image. Princeton, NJ: Princeton University Press.
Rutter, Michael. 1996. Transitions and Turning Points in Developmental Psychopathology: As Applied to the Age Span between Childhood and Mid-adulthood. International Journal of Behavioral Development 19:603–26.
Ryff, Carol D., and Corey Lee M. Keyes. 1995. The Structure of Psychological Well-Being Revisited. Journal of Personality and Social Psychology 69:719 –27.
Ryff, Carol D., Burton H. Singer, and Marsha Mailick Seltzer. 2002. Pathways through Challenge: Implications for Well-Being and Health. Pp. 302–28 in Paths to Successful Development: Personality in the Life Course, edited by Lea Pulkkinen and Avshalom Caspi. Cambridge: Cambridge University Press.
Schulenberg, John E., Alison L. Bryant, and Patrick M. O̓Malley. In press. Taking Hold of Some Kind of Life: How Developmental Tasks Relate to Well-Being during the Transition to Adulthood. Development and Psychopathology.
Schulenberg, John E., and Jennifer L. Maggs. 2002. A Developmental Perspective on Alcohol Use and Heavy Drinking during Adolescence and the Transition to Young Adulthood. Journal of Studies on Alcohol, Suppl. 14:54–70.
Schulenberg, John E., Jennifer L. Maggs, and Patrick M. O̓Malley. 2003a. How and Why the Understanding of Developmental Continuity and Discontinuity Is Important: The Sample Case of Long-Term Consequences of Adolescent (p.453) Substance Use. Pp. 413–36 in Handbook of the Life Course, edited by Jeylan T. Mortimer and Michael J. Shanahan. New York: Plenum Publishers.
Schulenberg, John, Patrick M. O̓Malley, Jerald G. Bachman, and Lloyd D. Johnston. 2000. “Spread Your Wings and Fly”: The Course of Health and Well-Being during the Transition to Young Adulthood. Pp. 224–55 in Negotiating Adolescence in Times of Social Change, edited by Lisa J. Crockett and Rainer K. Silbereisen. New York: Cambridge University Press.
Schulenberg, John E., Patrick M. O̓Malley, Jerald G. Bachman, Lloyd D. Johnston, and Virginia B. Laetz. 2003b. How Social Role Transitions from Adolescence to Adulthood Relate to Trajectories of Well-Being and Substance Use. Monitoring the Future Occasional Paper No. 56. Ann Arbor, MI: Institute for Social Research.
Settersten, Richard A., Jr. 2003. Age Structuring and the Rhythm of the Life Course. Pp. 81–98 in Handbook of the Life Course, edited by Jeylan T. Mortimer and Michael J. Shanahan. New York: Plenum Publishers.
Shanahan, Michael J. 2000. Pathways to Adulthood in Changing Societies: Variability and Mechanisms in Life Course Perspective. Annual Review of Sociology 26:667–92.
Singer, Burton, Carol D. Ryff, Deborah Carr, and William J. Magee. 1998. Linking Life Histories and Mental Health: A Person-Centered Strategy. Sociological Methodology 28:1–51.
Wiese, Bettina S., Alexandra M. Freund, and Paul B. Baltes. 2000. Selection, Optimization, and Compensation: An Action-Related Approach to Work and Partnership. Journal of Vocational Behavior 57:273–300.
(p.450) (1.) This summary of the findings is based on the following ANOVA results. Among the between-subjects (i.e., ignoring time) main and interaction effects involving gender, race/ethnicity, and cohort, only the gender main effect (p 〈 .05) and race/ethnicity main effect (p 〈 .01) were significant. In the within-subject effects, the overall time effect was significant (p 〈 .001; both linear and quadratic trends were significant, p 〈 .001); the time-by-gender interaction was significant (p 〈 .01; this interaction was significant for both linear and quadratic trends, p 〈.05); and none of the time-by-cohort or race/ethnicity interactions (two-, three-, and four-way) was significant. Such details are not provided in our other summaries of findings in this chapter, but they are provided in MTF Occasional Paper 56 (Schulenberg et al. 2003b).