Heavy use of alcohol by adolescents is a serious problem in the United States. According to the most recent Monitoring the Future Study (Johnston et al., 1998
), 31% of 12th-graders drank at least five drinks in one occasion during the 2 weeks before the survey. This drinking pattern, often labeled “binge drinking” in studies of adolescents, probably is most critical for the occurrence of negative outcomes (Kandel, 1980
; White, 1987
; Zucker, 1987
). In fact, several studies have found that high quantity drinking among adolescents results in numerous, immediate negative consequences, such as fatal automobile accidents (National Highway Traffic Safety Administration, 1997
), injuries, and risky sexual behavior (Fergusson and Lynskey, 1996
; Quigley and Marlatt, 1996
; Smart, 1996
; Thakker, 1998
, Zucker, 1994
). Less is known about more long-term consequences of such use.
The present study was undertaken to examine the young adult outcomes of adolescent binge drinking from an integrated person- and variable-centered perspective. The study had two purposes: (1) to identify and characterize different developmental trajectories of binge drinking in adolescence; and (2) to examine the age 21 outcomes of these trajectories. Although many researchers concentrate on negative consequences, such as criminality and health problems, it also is possible that heavy drinking in junior and senior high school could interfere with the development of social competencies and impair prosocial development. Therefore, we examined both negative outcomes, including depression, criminality, and alcohol and drug dependence, and positive outcomes, including graduation from high school, productive involvement, prosocial involvement, and parental bonding. Our general hypothesis was that the consequences of adolescent binge drinking would differ for youths with developmentally different binge drinking patterns.
Most of the research done to date on the more long-term consequences of adolescent alcohol use has employed a variable-centered approach. Such an approach focuses on level of binge drinking as the predictor, and examines trends in the sample. For example, Kandel and her colleagues (Kandel et al., 1986
) examined the consequences of age 15–16 drug use (past 12 months of use adjusted by frequency of use) on age 24–25 outcomes. They found that the strongest consequence of drug use is on subsequent use itself. In addition, they found several drug-specific predictions. Among these, adolescent alcohol use predicted increased shifts in employment, frequency of periods of unemployment, and health problems. However, these effects were nonsignificant when other drug use was controlled. Adolescent alcohol use did not predict educational attainment, crime, family roles, or depression. Kandel et al. found that most of the unique consequences of adolescent substance use came from the use of illicit drugs. Newcomb and Bentler (1988)
averaged frequency of alcohol use across several ages in adolescence to examine young adult consequences. These analyses produced mixed effects of adolescent alcohol use: adolescent alcohol use appeared to decrease criminal activities and loneliness while at the same time decreasing traditional pursuits such as college involvement.
One of the limitations of a variable-centered approach is that it does not capture possible heterogeneities in the developmental paths of a behavior. The variable-centered approach produces models that best describe the average behavior of the sample, but that are least applicable to those individuals showing the greatest deviations from the sample mean (Labouvie et al., 1991
). However, often it is precisely these more extreme groups (e.g., those who are chronically heavy drinkers over several years in adolescence, or those who escalate only in later years) who may be most interesting to study (Bates, 1993
). Zucker (1979)
suggests that one goal of alcohol research is to map out the developmental courses adequately for different types of drinkers.
Further, not only heterogeneities in binge drinking trends but also heterogeneities in broader social development are of interest. Youths at different ages face different developmental challenges. The early adolescent is engaged primarily in developing academic competence and peer relationships. By late adolescence, additional challenges, such as determining one’s place in the world and developing skills for intimacy, become more important (Clausen, 1991
; Erikson, 1982
). These developmental challenges are epigenetic, that is, the success of each depends in part on the successes of earlier periods. This principle is represented in our own Social Development Model (SDM; Catalano and Hawkins, 1996
), which presents four developmental periods: early childhood, late childhood, adolescence, and young adulthood. Within the SDM, problem behavior, such as binge drinking, is influenced by the prosocial and antisocial opportunities, involvement, rewards, bonding, and norms experienced during that period. In turn, binge drinking itself reduces the likelihood of successful prosocial socialization in the next period. For example, young adults who were hampered in developing academic and interpersonal competence in early adolescence, perhaps by early binge drinking, will have a more difficult time adapting to the adult world of work and interpersonal intimacy. Thus, not only are patterns of binge drinking important to understanding adult outcomes, but their developmental timing must be taken into account.
Researchers have noted the importance of conducting both variable-centered and person-centered analyses for some outcomes (Bates, 2000
; Cairns, 1986
; Magnussen and Bergman, 1988
; Muthén, 2000
). However, most of the work done thus far on the consequences of adolescent drinking has been exclusively variable-centered. To some extent, this is because the methods of person-centered analyses of intraindividual change have lagged behind. There is growing recognition, however, that greater attention needs to be directed toward individual growth curves and the description and explanation of differences in intraindividual change (White et al., 1998
). In the past decade, major methodological advances have been made that permit person-centered techniques based on the concept of individual growth trajectories. These techniques include latent growth modeling (Duncan and Duncan, 1996
; Muthén and Muthén, 2000
), hierarchical linear modeling (Bryk and Raudenbush, 1992
; Willett and Sayer, 1994
), and semi-parametric group-based modeling (Nagin 1999
; Nagin and Land, 1993
). However, the extent to which they may contribute to a better understanding of the behavioral outcomes of intraindividual change has not been well established.
For example, latent growth modeling (LGM) does employ an individual growth-curve methodology, because the intercepts and slopes of individual trajectories of alcohol use constitute the data examined. However, this method does not attend to the possibility of the existence of different types of trajectories with different shapes, and at different developmental periods. For example, Duncan et al. (1997)
used latent growth modeling to examine the effects of level of alcohol use and development of alcohol use during adolescence. They found that, on average, both the level and the slope of alcohol use in adolescence were important for future adjustment outcomes in young adulthood, including higher young adult alcohol use, alcohol-related problems, aggressive behavior, theft, and suicide ideation. One could perform LGM analyses on different subgroups to examine possible differences in trajectories, but these subgroups would need to be identified using some other method, or by a priori theory.
One study recently reported by White et al. (1998)
used adolescent-to-adult growth curves in illicit drug use (marijuana, cocaine, psychedelics, heroin, and nonprescription use of sedatives, tranquilizers, stimulants and analgesics) as independent variables to predict differences in adult outcomes. Normative and individual patterns of intraindividual change in drug use were described as composites of average levels, linear trends, and quadratic trends. This process-oriented approach to individual change was compared to an incremental approach (which used level of drug use at a single point in time) to examine the outcomes of physical and mental health, role functioning, and substance use and dependence in adulthood. Regardless of the analysis strategy, adolescent drug use was related to a lower likelihood of being married and to higher levels of alcohol and drug dependence. However, psychological distress, occupational status, and criminal behavior outcomes were significantly predicted by drug use only when contemporary young adult use was included in the trajectory. Thus, in this study a trajectory method revealed neither particularly different nor stronger predictions than would occur simply using a contemporaneous drug use measure. However, White et al. suggest that a more useful analysis might “be based on clustering or grouping individuals on the basis of their trajectories into relatively homogenous subgroups, each representing a distinct trajectory” (White et al., 1998
, p. 172).
A study reported by Schulenberg et al. (1996)
did employ a person-centered, cluster analytic technique to examine the consequences of binge drinking. Like the present study, Schulenberg et al. sought to identify different trajectories of binge drinking and to examine their consequences. The authors proposed a priori, and then confirmed with cluster analysis, six binge drinking trajectory groups: never; rare; chronic; decreased; increased; and “fling.” Schulenberg et al. examined binge drinking trajectories from age 18 to age 24. They found that the chronic and increased frequent binge drinking trajectories were associated with difficulties in the transition to adulthood by age 24.
Recently, Muthén and Shedden (1999)
employed finite mixture modeling to examine latent trajectories of binge drinking from age 18 to age 24. From this analysis they determined a three-trajectory model best fit their data: “high” for those who were high at baseline (age 18); “up” for those who were lower, but accelerated their use; and “norm,” a relatively low, normative curve of frequency of binge drinking. They then estimated the effect of trajectory class on alcohol abuse and dependence. Using this methodology, Muthén and Shedden found the “up” (or Increasers) had the highest estimated probability of alcohol dependence, followed by the highs and the norms, respectively.
The Schulenberg et al. (1996)
and Muthén and Shedden (1999)
studies provide important person-centered views of the consequences of binge drinking in young adulthood. However, it is during adolescence that most people initiate alcohol use, and it is during adolescence that many begin to show serious difficulties with alcohol use (Anthony et al., 1997
; Nelson et al., 1998
; Vaillant, 1983
; Zucker, 1987
Zucker, 1995). The present study seeks to understand the patterns of binge drinking in adolescence, and the consequences of these patterns on social functioning, criminal behavior, and mental health in the transition to adulthood. Specifically, it seeks to determine (1) if qualitatively different binge drinking trajectory groups can be discerned in adolescence (ages 13–18); and (2) if these different trajectory groups experience different positive and negative outcomes in the transition to adulthood at age 21.