In recent years, a good deal of attention has been turned toward examining how young people successfully transition to their adult roles. Although few coherent theoretical accounts have emerged to explain individual differences in such transitions, there is a dawning consensus in the field that the years between 18 and 25 may represent a distinct developmental period in postindustrialized societies, which Arnett (2000) has dubbed “emerging adulthood.” According to Arnett, these transition years are marked by relative instability in terms of youths' experiences in various salient life domains, a theoretical supposition that appears consistent with emerging evidence (Cohen, Kasen, Chen, Hartmark, & Gordon, 2003). Although some have gone so far as to suggest that no normative pattern therefore adequately characterizes the lives of emerging adults, a somewhat broader view suggests that what emerging adulthood may represent is a period within which prior adaptations are consolidated toward scaffolding the emergence of the twin developmental tasks of adulthood: romantic and work competence.

In a recent article, Roisman, Masten, Coatsworth, and Tellegan (2004) formalize such a perspective by suggesting that the transition years are indeed a period where young people develop the skills necessary to fill adult roles as workers and romantic partners. However, adaptation within these domains is not yet fully mature, hence the presence of both major inroads toward accepting adult roles (e.g., engaging in longer term relationships, preparing for a career) as well as significant retrograde experiences (e.g., breaking up with a long-term partner) during these years. Importantly, Roisman et al. (2004) found little evidence that successfully engaging the emerging domains of work and romantic relationships within the transition years predicted adult adaptation within these roles at age 30, demonstrating instead that social skills underlying the development of successful friendships and academic performance in the transition years set the stage for adult success in work and love.

Roisman et al. (2004), nonetheless, emphasized the conditional significance of work and romantic experiences in the transition years. Substantial evidence suggests that, under certain circumstances, work and romantic experiences may play supportive or undermining roles in the lives of emerging adults, particularly when such experiences are promotive of or interfere with educational performance and opportunities. For example, intense work experiences are clearly associated with disengagement from school, even after controlling for selection effects (Roisman, 2002; Steinberg, Fegley, & Dornbusch, 1993). Nonetheless, when coupled with saving for college, sociological data suggest that high quality involvement in work in the transition years can also represent a positive marker (Marsh, 1991). Incipient romantic experiences may likewise play positive or negative roles in the lives of emerging adults in that they can provide opportunities to move beyond a troubled family background or to be further victimized (Collins, 2003).

Perhaps the most important role for romantic and work experiences in the transition years, however, is described in the literature on desistence in antisocial offending. According to turning points theory (Laub, Nagen, & Sampson, 1998), what is coming on-line in the transition to adulthood is an unprecedented opportunity structure for youth, perhaps particularly so for those with antisocial histories. Although the work and romantic domains are likely not full-fledged developmental tasks in the early 20s, they provide an opportunity for youth to get involved in negotiating tasks that will ultimately become salient markers of being a successful adult in arenas that may be quite outside of the contexts of child and adolescent development that have historically sustained their developmental failures (e.g., their family, school system, etc.). This article examines the unique opportunities available to youth as they transition to adulthood to better understand mechanisms of continuity and change in antisocial behavior.

Substantial evidence has emerged supporting the existence of multiple developmental pathways to antisocial behavior problems in childhood, adolescence, and beyond (see, e.g., Aguilar, Sroufe, Egeland, & Carlson, 2000; Farrington & Loeber, 2000; Moffitt, Caspi, Harrington, & Milne, 2002; Patterson & Yoerger, 1997). Because early-emerging antisocial behavior problems have shown strong continuity across time (Loeber & Farrington, 2000; Olweus, 1979; Robins, 1966), much of this research has focused on the Early-Onset/Persistent (EOP) pathway. Although there is some disagreement regarding the progression of manifest antisocial behaviors within this pathway (e.g., Loeber's overt versus covert pathway; Loeber & Stouthamer–Loeber, 1998), most authors agree that the typical progression is: tantruming in early childhood; oppositional, defiant, and aggressive behaviors during middle childhood; shoplifting, lying, fighting, and involvement in more serious crimes during adolescence; and various forms of violent and serious criminal and antisocial behavior in adulthood (see Coie & Dodge, 1997, for a review).

In addition to high-level antisocial behavior, the EOP pathway has been reliably linked with problems in multiple domains of functioning across all major developmental periods (Aguilar et al., 2000; Moffitt, Caspi, Harrington, & Milne, 2002; Robins, 1966). In reviewing the consequences of early-starting and persisting antisocial behaviors, Farrington and Loeber (2000) concluded the following: (a) the repeated practice of disruptive behaviors and delinquent acts during formative periods contributes to the subsequent stability of such acts over time; (b) desistence is less likely, especially for those who advance to more serious forms of antisocial behavior; (c) years of engaging in antisocial behavior tend to rob youth of many opportunities to learn and practice prosocial behavior; (d) engaging in persistent antisocial behavior is often associated with poor social skills, leading to major and continued disturbances in relationships with relatives, peers, partners, employers, and coworkers; (e) low interest and motivation in acquiring an education leads to major disruption in classrooms, exposure to remedial/alternative/special education services, and eventually to chronic truancy and dropout; (f) chronically antisocial youth are more likely to become parents at a young age but are unable or unwilling to fulfill parental duties and often provide undesirable role models for their children; and (g) these youth have a high risk for depression and suicide, are at high risk for being killed or maimed for life, and often start using illicit substances at a young age, which increases their risk for becoming a persistent substance abuser.

Although adolescent antisocial behavior has its origins in childhood for some youth, there is a dramatic increase in the prevalence of antisocial behavior during the adolescence period. Citing these data, Moffitt (1993) suggested that there exists a distinguishable group of youth who show antisocial behavior that is limited to the adolescent period; they show no notable history of childhood behavior problems, and are expected to desist from offending postadolescence. Substantial research can be now cited in support of Adolescence-Onset (AO) antisocial behavior as a distinguishable pathway. Aguilar and colleagues (2000), for example, recently reported evidence for the distinctiveness of a group of youth who showed an onset of clinically elevated levels of antisocial behavior during adolescence, but did not show significant antisocial problems during earlier years.

Investigators have begun to describe and delimit pathways associated with antisocial behavior, but comparatively scant attention has been paid in the literature to understanding the (presumably differing) processes associated with the initiation of clinically significant antisocial behavior problems in adolescence versus childhood. Moffitt (1993) has theorized that, unlike the EOP group, AO antisocial behavior is not a psychopathological syndrome because it is presumed to show marked discontinuity and is statistically normative. She has argued that secular changes in health and work in Westernized countries have lengthened the duration of adolescence, leaving modern teenagers in a 5- to 10-year gap between biological and social maturity (a view not inconsistent with Arnett's [2000] theory of “emerging adulthood”). According to this view, EOP youth appear to be unaffected by this maturity gap, displaying their autonomy by engaging in such “mature” behaviors as drinking, smoking, and sexual activity. The AO pattern is thought to arise from an attempt by otherwise well-adjusted adolescents to display their independence by engaging in a kind of social mimicry of deviant peers because they want to acquire the mature status that they believe deviant behavior accords (Caspi & Moffitt, 1995).

Although the maturity gap hypothesis is intriguing and provocative, the proposition that AO antisocial behavior is essentially normative and nonpsychopathological has not yet been fully addressed with empirical research, particularly with regard to its developmental consequences in adulthood. Moreover, the emerging evidence that does exist suggests that high-level AO antisocial behavior is associated with concurrent problems such as high levels of depression/anxiety (Aguilar et al., 2000; Loeber, Russo, Stouthamer–Loeber, & Lahey, 1994) and disrupted parenting (Patterson & Yoerger, 1997), both of which are likely to have consequences beyond adolescence (Brook, Whiteman, Finch, & Cohen, 1996). In addition, although Moffitt et al. (2002) recently found partial support for their theory that the problems experienced by AO men around age 26 were of lower severity compared with those of the EOP group, the Dunedin birth cohort data nevertheless also showed that AOs have more problem behaviors in young adulthood compared with their Never Antisocial (NA) counterparts.

Despite these emerging findings regarding AO antisocial behavior, some important issues need further research emphasis. In particular, important contributions will be made by prospective, longitudinal studies that thoroughly and comprehensively examine the quality of antisocial youths' adaptation at all major developmental periods from birth to adulthood. Our own work (Aguilar et al., 2000) has shown that AOs are distinguishable from EOP youth in that their lives are marked by significantly fewer psychosocial risk factors before age 4, and they are less likely to come from a single-parent home or to have been maltreated. However, their profile beyond adolescence has not been thoroughly examined. Furthermore, extant research has not completely addressed as to whether youth who begin to show significant antisocial behavior in adolescence, in fact, desist from offending in young adulthood. Put simply, is AO antisocial behavior really limited to adolescence? Preliminary data suggest that it may not be for some individuals. As described earlier, some empirical studies have reported that a significant percentage of AO antisocial youth continue to show evidence of criminal behavior at age 26 (Moffitt et al., 2002), and as late as age 32 (Nagen, Farrington, & Moffitt, 1995).

Using data from a high-risk poverty sample, the first goal of this study was to assess whether adaptational problems during young adulthood can significantly distinguish EOP and AO youth from one another and from a NA control group. In addition to examining problem behaviors such as symptoms of externalizing and internalizing problems as well as life stress, the present study focused on success in the emerging developmental tasks of young adulthood (Roisman et al., 2004), including the engagement of the work and romantic domains. A related aim of this study was to examine whether AO antisocial youth are significantly less likely to show clinically elevated levels of antisocial behavior during young adulthood compared with EOPs.

The final goal of this study was to examine the correlates of lower levels of antisocial offending among AOs and EOPs in the transition to adulthood. Turning points theory (Sampson & Laub, 1993) posits that the successful engagement of the romantic and work domains in young adulthood represents an important opportunity for disrupting antisocial pathways. Nonetheless, Moffitt et al. (2002) have recently speculated that such turning points may be most applicable to the majority of antisocial youth who begin offending in adolescence (AO), rather than those who begin early in childhood and persist (EOP). This assertion is surprising for at least two reasons: (a) their study did not attempt to examine factors that distinguish desisters from persisters within the AO or EOP groups; and (b) given that the EOP antisocial type may result from failures in the developmental tasks of childhood and adolescence (Aguilar et al., 2000), it may be that positive experiences in the romantic and work arenas in young adulthood are particularly effective mechanisms for derailing the progression of antisocial behaviors among EOPs into adulthood. In other words, although EOP youth may be much less likely to be successful at work and love in the transition to adulthood, variation within this group may be largely attributable to success within these emerging developmental tasks. In addition to work and romantic engagement, in this study we examined the role of illicit substance use and deviant peer association in maintaining antisocial pathways. Moffitt (1993) has theorized that drug use and involvement with criminal associates are snares that maintain antisocial behavior into young adulthood, particularly among AO youth. We test this prediction here using prospective, longitudinal data.

Results

Analytic plan

To examine group differences between AOs, EOPs, and NAs, planned comparisons were used because they are ideal in protecting against Type II error (e.g., they eliminate the need for omnibus F tests). Because a maximum of k − 1 comparisons can be completed per outcome variable using planned comparisons (where k is defined as the number of levels in the independent variable), each set of analyses described below contrasts the NAs with the EOPs and AOs as well as the AOs with the EOPs. It should be noted that, although we created a CL antisocial behavior group for later regression analyses, we opted to focus our discussion of group differences among AOs, EOPs, and NAs for two important reasons. First, the number of CL youth was especially small (n = 13), which leads us to be hesitant about making substantive interpretations regarding their developmental patterns. Second, this study's aims and hypotheses are primarily focused around differences between antisocial adolescents (EOPs vs. AOs) and how they compare with the NA control group at age 23.

In addition to examining mean-level problem behaviors and adaptation across the three antisocial behavior groups, we used hierarchical regressions to examine the correlates of higher and lower levels of antisocial behavior at age 23. In each regression, dummy variables were entered reflecting antisocial history (AO, CL, EOP).¹

In each case, dummy coded group variables were coded 1 = member of group or 0 = not member of group. To prevent multicollinearity, the NA dummy variable was not entered in regressions.

Next, emerging developmental tasks variables, such as work involvement, were entered in a second block. Finally, interactions between antisocial developmental history and emerging developmental task variables were examined (e.g., AO × Work Ethic; CL × Work Ethic, EOP × Work Ethic) to determine if turning point experiences were specifically associated lower levels of antisocial behavior at age 23 given a particular developmental history of antisocial behavior in childhood/adolescence. As discussed in the Introduction, we expected that work/romantic competence and externalizing problems would be negatively correlated among EOP (but not AO) youth.

Of note, EOPs were found to be more impaired than AOs by their teachers (but not by their mothers or via self-report) at age 16. To minimize the possibility that group effects (NA, CL, AO, EOP) might be spurious due to adolescent differences in severity of impairment, we covaried an average of the three informants' continuous scores on externalizing at age 16 in all regression analyses. Although we chose a multiinformant composite to maximize the validity of the assessment of antisocial impairment at age 16, the results reported replicated controlling for data from any single informant.

Group differences

Planned comparisons described in Table 1 revealed that AOs and EOPs consistently reported significantly more total psychological problems, externalizing problems, anxiety/depression, deviant peer association, drug use, and life stress at age 23 than their NA counterparts. In addition, EOPs showed significantly higher levels of externalizing and illegal drug use than AOs (the difference between EOPs and AOs approached significance, p = .06, on total problem behaviors). EOPS and AOs, however, were not distinguishable on anxiety/depression, deviant peer association, and life stress. The final analysis within this set focused on the percentage of participants identified as antisocial in adolescence that either persisted or desisted in antisocial behavior using the criteria described in the Methods section. Table 2 shows that AOs were significantly more likely to have desisted by age 23 than EOPs. Raw data relevant to the other groups are also included in Table 2.

Symptoms of psychopathology, problem behaviors, and life stress at age 23 by antisocial group

Adolescent antisocial group by antisocial behavior at 23 years

In terms of adaptive functioning, Table 3 shows that only EOPs appeared to show significant problems compared to NAs at age 23 in the domains of work/school status, work ethic, and educational attainment. AOs were comparable to NAs on these measures and had significantly higher work ethic and educational attainment scores than EOPs (the AO vs. EOP difference on work involvement was marginally significant). Interestingly, no group differences emerged on the degree to which individuals across the antisocial groups were involved in long-term romantic relationships between ages 21 and 23.

Adaptation between ages 21 and 23 by antisocial group

The independent and interactive role of emerging developmental tasks

A set of three hierarchical regressions (see Tables 4, 5, and 6 for the final step) were estimated to examine the independent and interactive role of antisocial history and successful work/romantic engagement in predicting externalizing problems at age 23. In line with the group comparison analyses already presented, antisocial history (whether early, late, or both) was significantly associated with externalizing behavior at age 23. It is noteworthy that in all regressions antisocial history variables, as well as level of impairment (the multiinformant, continuous externalizing score composite) at age 16, strongly predicted levels of antisocial behavior at age 23 (R² = .34 for step 1, including antisocial history and severity in all equations).

Results of regression demonstrating whether variation in externalizing behavior at age 23 is associated with developmental history of antisocial behavior, school/work status between ages 21 and 23, and their interactions

Results of regression demonstrating whether variation in externalizing behavior at age 23 is associated with developmental history of antisocial behavior, work ethic between ages 21 and 23, and their interactions

Results of regression demonstrating whether variation in externalizing behavior at age 23 is associated with developmental history of antisocial behavior, patterning of romantic relationship involvement between ages 21 and 23, and their interactions

Although school/work status, work ethic, and long-term romantic involvement in early adulthood were not directly associated with reports of externalizing behavior at age 23 across the sample (main effects were nonsignificant), in each regression these variables were associated with lower levels of externalizing in the EOP group (EOP interactions were significant). No other Antisocial Group × Turning Point Variable interactions were significant. Note that all significant interactions reflected the fact that EOP youth with positive adaptation in the work and romantic domains showed levels of externalizing problems consistent with others (at the sample mean). Successful adaptation to the work and romantic domains was essentially unrelated to externalizing problems for non-EOP youth.

In a set of two regressions focused on deviant peer association and illegal drug use (see Tables 7 and 8), a different pattern of results emerged. For both regressions, antisocial history and concurrent deviant peer association/drug use were associated with higher levels of externalizing at age 23 (results were marginally significant for illegal drug use). Although interactions are not reported due to high levels of multicollinearity in these regression equations, none of the antisocial group interactions were significant (in addition, within group analyses revealed no obvious differences in effects by antisocial group).

Results of regression demonstrating whether variation in externalizing behavior at age 23 is associated with developmental history of antisocial behavior and deviant peer association at age 23

Results of regression demonstrating whether variation in externalizing behavior at age 23 is associated with developmental history of antisocial behavior and illegal drug use at age 23