Relationship Between Alcohol Use and Violent Behavior Among Urban African American Youths From Adolescence to Emerging Adulthood: A Longitudinal Study

Abstract

Objectives. We examined developmental trajectories of alcohol use and violent behavior among urban African American youths and the longitudinal relationship between these behaviors from adolescence to emerging adulthood.

Methods. Our sample included 649 African American youths (49% male) followed for 8 years. We assessed violent behavior and alcohol use by asking participants how often they had engaged in each behavior in the preceding 12 months. Growth curve analyses were conducted to identify the developmental trajectories of the 2 behaviors and to explore the longitudinal relationship between them.

Results. Violent behavior peaked in middle to late adolescence and declined thereafter, whereas the frequency of alcohol use increased steadily over time. These developmental trajectories varied according to gender. Among both male and female participants, early violent behavior predicted later alcohol use, and early alcohol use predicted later violent behavior. Moreover, changes in one behavior were associated with changes in the other.

Conclusions. Our results support a bidirectional relationship between alcohol use and violent behavior. Efforts to reduce one problem can be expected to reduce the other. Programs and policies aimed at reducing violence or alcohol use among adolescents should take into account this relationship.

Both alcohol use and violent behaviors are prevalent among urban adolescents and are important public health problems in the United States,^1,2 with violent injury being the leading cause of death among African American adolescents.³ According to a nationwide survey of high school students, 36% of students reported having committed violent acts in the preceding 12 months, and 19% reported having carried a weapon in the preceding 30 days.⁴ Another survey showed that two thirds of 12th-grade students had consumed alcohol in the past 12 months, and nearly half (45%) were current drinkers (i.e., they reported having consumed alcohol in the 30 days prior to the survey).² These problems have continued to receive increased attention in recent years.

In general, 4 competing theoretical explanations have been proposed for the relationship between alcohol use and violent behavior.^5,6 According to the first model, alcohol use causes violent behavior owing to psychopharmacological effects⁷ or a criminal subculture.⁸ The second model postulates that alcohol use is caused by violent behavior and is a consequence of a violent lifestyle because aggressive individuals are more likely to select or be pushed into social situations that encourage heavy drinking.⁹

The third model combines the first 2 models and argues that alcohol use and violence reinforce each other; in other words, alcohol use causes violence, and vice versa.¹⁰ The final model postulates that the relationship between alcohol use and violence is spurious.^10,11 Both behaviors are predicted by the same common set of risk factors and cluster together as a result of a single general problem behavior syndrome.^12,13

The relationship between alcohol use and violence has been documented in many studies.^5,6,14 In a large number of these studies, however, alcohol use and violence have been assessed at a single time point, and lead-lag effects (one variable correlating with another variable at a subsequent point in time) could not be studied. In the few studies that have examined longitudinal associations between alcohol use and violent behavior, findings have been mixed.

For example, some researchers have found that early alcohol use predicts later violent behavior,^15–18 and others have found that early violent behavior predicts later alcohol use.^11,19,20 White et al.,¹⁰ in their study of high-risk adolescent boys, demonstrated a bidirectional rather than unidirectional association between alcohol use and violent behavior. Other researchers^21–23 have also found that the relationship between alcohol use and violent behavior involves reciprocal influences.

In a meta-analysis of existing longitudinal studies examining the correlations between alcohol use and violence, Lipsey et al.²⁴ found that when common risk factors were taken into account, the strengths of the relationships were attenuated. In contrast, White et al.¹⁰ found that the strengths of cross-lagged associations between alcohol and aggression were not reduced significantly when they controlled for common risk factors.

One problem with existing research is that many of these studies have involved somewhat limited samples.^5,25 Some researchers draw their samples from the juvenile justice system, which may not be representative of youths more generally. Others draw their samples from schools but exclude dropouts and absentees, which may lead to undersampling of the most heavy alcohol users and delinquents. Most studies have included predominantly White youths in their samples. Little is known about the longitudinal relationship between alcohol use and violent behavior among African American youths. Perhaps most important, few studies have addressed the critical developmental period of emerging adulthood (age 18–29 years), which would allow an examination of associations between these 2 behaviors across different stages in the life cycle.

The objectives of our study were to address these gaps in the literature and test the 4 competing theories on the relationship between alcohol use and violent behavior with more than 600 African American youths followed for 8 years. Because our data were collected at multiple time points, we were able to examine the developmental trajectories of these behaviors, the lead-lag relationship between alcohol use and violent behavior, and how changes in one behavior were associated with changes in the other.

We investigated whether early alcohol use or violent behavior predicted later violent behavior or alcohol use from adolescence to emerging adulthood. We also examined the relationship between alcohol use and violent behavior from a developmental life course perspective in an attempt to determine whether the association differed in adolescence and emerging adulthood.

Researchers have found that rates of violent behavior²⁶ and use of alcohol and other drugs²⁷ are higher among adolescent boys than among adolescent girls. Violent behavior peaks in middle to late adolescence and declines thereafter,²⁶ whereas the frequency of alcohol and other drug use continues to increase throughout adolescence and declines in emerging adulthood.²⁸ In this study, we also assessed whether the developmental trajectories of alcohol use and violent behavior and the longitudinal associations between the 2 behaviors differ according to gender.

In our analyses, we controlled for several common risk factors for both alcohol use and violence described in the literature.²⁹ We included participants' academic achievement and depressive symptoms, parental drug use and violent behavior, family conflict, and peer drug use and violent behavior, as well as whether participants had sold illegal drugs. All of these measures were assessed at wave 1 of the study unless otherwise noted. We attempted to determine whether the relationship between alcohol and violent behavior could be explained by these common risk factors.

METHODS

We used data from an 8-year longitudinal study of substance use among youths at risk for dropping out of school.³⁰ Ninth graders in the second largest school district in Michigan who were enrolled in one of 4 public high schools in the fall of 1994 and whose grade point average (GPA) had been 3.0 or below in 8th grade were eligible for enrollment in the study.

The sample included 681 African American youths who were followed annually for 4 years during high school and another 4 years after high school. Thirty-two of these youths were excluded from our analyses as a result of missing data. The age range of the final sample (n = 649; 49% male) at year 1 was 13.87 to 16.88 years (mean = 14.86; SD = 0.65). Half the participants had an 8th-grade GPA of 2.0 or lower (mean = 2.20; SD = 0.70). A response rate of 91% was maintained from years 1 through 4; the overall response rate from years 1 through 8 was 68%.

Data Collection

At each wave of data collection beginning with 9th grade in 1994, trained interviewers conducted a structured 50- to 60-minute, face-to-face interview with each participant during regular school hours to collect data on demographics, family background, behaviors, relationships with friends and parents, and so forth. Youths who were not in school were interviewed in a community setting. After the face-to-face interview, participants completed a pencil-and-paper questionnaire focusing on sensitive issues such as substance use.

Measures

Violent behavior. Scores from 7 Likert-scale items were used to assess violent behavior. At each wave, youths indicated how often they had engaged in each behavior (ranging from 1 = 0 times to 5 = 4 or more times) during the preceding 12 months. Items asked how often participants engaged in the following behaviors: “gotten into a fight in school,” “taken part in a fight where a group of your friends were against another group,” “hurt someone badly enough to need bandages or a doctor,” “hit a teacher or supervisor (work supervisor),” “used a knife or gun or some other thing (like a club) to get something from a person,” “carried a knife or razor,” and “carried a gun.” Higher scores indicated more violent behavior. Cronbach α coefficients for this measure from waves 1 through 8 ranged from 0.75 to 0.81.

Alcohol use. Participants self-reported their frequency of alcohol use over the preceding 12 months during each wave. Response categories were 0 = 0 times, 1 = 1 to 2 times, 2 = 3 to 5 times, 3 = 6 to 9 times, 4 = 10 to 19 times, 5 = 20 to 39 times, and 6 = 40 or more times.

Depression. Six items from the Brief Symptom Inventory³¹ were used to measure depressive symptoms. Participants reported the extent to which loneliness, hopelessness, or worthlessness had caused them to be uncomfortable during the preceding week (ranging from 1 = not at all uncomfortable to 5 = extremely uncomfortable; Cronbach α = 0.79).

Academic achievement. Participants' 8th-grade GPAs were obtained from school records.

Selling drugs. Drug selling was measured with a single item: “During the last 12 months, how often have you sold an illegal drug?” (ranging from 1 = 0 times to 5 = 4 or more times).

Peers' violent behavior and drug problems. We used a 10-item scale (Cronbach α = 0.74) to assess the number of participants' friends who had substance use problems (e.g., drug use at school).³² Peers' violent behavior was assessed with 3 questions asking about the number of friends who got into fights, had carried a gun, or had carried a knife or razor (Cronbach α = 0.72). Response options for all items ranged from 1 = none to 5 = all.

Parental violent behavior and drug problems. The measure of parental drug problems consisted of 13 items assessing drug and alcohol use by the adults the youths said had raised them³³ (Cronbach α = 0.75). Parental violent behavior was measured with 2 items: frequency of carrying a gun and frequency of carrying a knife or razor. Response options for these items ranged from 1 = never to 5 = very often.

Family conflict. Participants were asked 5 questions about the frequency of fighting and acting out in their family³⁴ (ranging from 1 = hardly ever to 4 = often; Cronbach α = 0.76).

Demographic variables. The demographic variables included in our analyses were participant gender and socioeconomic status (SES) as reported by respondents at wave 1. SES was assessed as the highest occupational prestige score for either of the participants' parents.^35,36 (Mean occupational prestige scores within each major category defined by the National Opinion Research Center are as follows: operators, fabricators, and laborers, 33.38; service occupations, 34.95; farming, forest, and fishing occupations, 35.57; precision production, craft, and repair occupations, 38.51; technical, sales, and administrative support occupations, 40.43; and managerial and professional specialty occupations, 62.24.) The parental occupational prestige scores in our sample ranged from 29.3 to 64.4 (mean = 39.84; SD = 9.83). Most of the participants were from working-class backgrounds.

Growth Curve Analyses

We conducted 2-level growth curve analyses³⁷ with hierarchical linear modeling software.³⁸ Separate parallel analyses were performed with alcohol use and violent behavior as the dependent variables. The level 1 model examined within-individual patterns of change in alcohol use or violent behavior. The level 2 model included individual-level factors associated with trajectories of alcohol use or violent behavior (more details about the models are available as a supplement to the online version of this article at http://www.ajph.org).

RESULTS

Descriptive statistics for the 2 dependent variables by wave and gender (a table outlining the results is available as a supplement to the online version of this article at http://www.ajph.org) showed that the overall prevalence of violent behavior in the preceding 12 months was quite low across all waves. The frequency of alcohol use in the preceding 12 months was also relatively low (on average, a couple of times in adolescence and approximately once every other month in emerging adulthood).

We found an overall trend of increasing alcohol use and decreasing violent behavior from high school through emerging adulthood. Male participants reported more alcohol use than did female participants during emerging adulthood, whereas virtually no gender differences in alcohol use were observed during high school. Male respondents reported more frequent violent behavior than did female respondents from high school through emerging adulthood.

Trajectory of Alcohol Use and Gender Differences

Data on the alcohol use trajectory from high school to emerging adulthood and the predictive relationship between violent behavior and alcohol use are shown in Table 1. The unconditional model revealed a linear increase (evident through the positive age slope) in alcohol use from high school to emerging adulthood.

TABLE 1.

Regression Coefficients from Growth Curve Analyses with Violent Behavior Predicting Change in Alcohol Use Among Urban African American Youths: Michigan, 1994–2002

	Unconditional Model, b (SE)	Model 1, b (SE)	Model 2, b (SE)	Model 3, b (SE)	Model 4, b (SE)	Model 5, b (SE)
Alcohol use at age 18 y	1.500*** (0.052)	1.405*** (0.072)	1.443*** (0.069)	1.435*** (0.069)	1.384*** (0.072)	1.409*** (0.069)
Gender (male)		0.202* (0.104)	0.134 (0.099)	0.162 (0.103)	0.217* (0.103)	0.195 (0.103)
SES		0.025 (0.050)	0.024 (0.047)	0.069 (0.048)	0.010 (0.049)	0.062 (0.048)
Violent behavior in Year 1			0.431*** (0.051)	0.142* (0.065)
8th grade GPA				−0.092 (0.051)		−0.105* (0.050)
Depression				0.073 (0.054)		0.090 (0.053)
Selling drugs				0.143* (0.064)		0.174** (0.058)
Friends' violence				0.009 (0.065)		0.050 (0.062)
Friends' drug use				0.201** (0.067)		0.206** (0.066)
Parent drug use				0.132* (0.054)		0.139* (0.054)
Parent violence				0.027 (0.053)		0.044 (0.052)
Family conflict				0.130* (0.051)		0.133** (0.050)
Age slope
Intercept	0.168*** (0.012)	0.124*** (0.017)	0.119*** (0.017)	0.116*** (0.017)	−0.019 (0.030)	0.160*** (0.029)
Gender (male)		0.094*** (0.025)	0.101*** (0.025)	0.104*** (0.027)	0.084*** (0.060)	0.092*** (0.026)
Violent behavior in Year 1			−0.044*** (0.013)	−0.026 (0.017)
8th grade GPA				0.031* (0.013)		0.028* (0.012)
Depression				−0.001 (0.014)		0.003 (0.013)
Friends' violence				0.001 (0.016)		0.005 (0.016)
Friends' drug use				−0.016 (0.017)		−0.019 (0.017)
Parent drug use				−0.014 (0.014)		−0.016 (0.014)
Parent violence				0.005 (0.014)		0.007 (0.013)
Family conflict				−0.007 (0.013)		−0.013 (0.013)
Violent behavior					0.694*** (0.077)	0.672*** (0.077)
Violent behavior × age					−0.053** (0.019)	−0.020 (0.020)

Note. GPA = grade point average; SES = socioeconomic status. Age slope is the linear component of the growth curve.

*P < .05; **P < .01; ***P < .001.

Further examination of the data revealed gender differences in alcohol use trajectories (Table 1, model 1; a figure is available as a supplement to the online version of this article at http://www.ajph.org.) Alcohol use among male participants increased at a faster rate than it did among female participants. Although we found no gender differences in high school alcohol use, male participants reported drinking more frequently than did female participants in emerging adulthood.

Violent Behavior and Alcohol Use Trajectory

Predictive relationship. Year 1 violent behavior was positively associated with alcohol use at the age of 18 years but negatively associated with the rate of increase in alcohol use after controlling for gender and SES (Table 1, model 2), suggesting that the association was stronger in high school than in emerging adulthood (Figure 1). After we controlled for other common risk factors, the association between violent behavior and alcohol use remained significant but was reduced in magnitude and no longer changed with age (Table 1, model 3).

FIGURE 1.

Predictive relationship between violent behavior in year 1 and alcohol use trajectory among urban African American youths: Michigan, 1994–2002.

Note. Values on the y-axis are ratings from the 7-point scale used to assess alcohol use in the preceding 12 months (ranging from 0[0 times] to 6[≥ 40 times]). Low level of violent behavior is 1 standard deviation below the mean; high level of violent behavior is 1 standard deviation above the mean.

Concurrent relationship. When violent behavior at each wave was included as a time-varying covariate in alternative analyses, similar patterns were observed. Increases in violent behavior were associated with increases in alcohol use from adolescence to emerging adulthood. Moreover, we found a significant interaction effect between violent behavior and age (Table 1, model 4). A visual display of the results demonstrated trajectories similar to those shown in Figure 1, suggesting that the association was stronger during adolescence than during emerging adulthood.

After we controlled for common risk factors, the interaction between violent behavior and age dropped to nonsignificance, whereas the main effect of violent behavior remained virtually the same (Table 1, model 5). Gender was not significantly associated with the violent behavior slope, and thus it was dropped from the final models.

Trajectory of Violent Behavior and Gender Differences

The results of parallel growth curve analyses focusing on the violent behavior trajectory are shown in Table 2. The unconditional model revealed that violent behavior changed in a nonlinear fashion from adolescence to emerging adulthood, with the trajectory including linear, quadratic, and cubic components.

TABLE 2.

Regression Coefficients from Growth Curve Analyses with Alcohol Use Predicting Change in Violent Behavior Among Urban African American Youths: Michigan, 1994–2002

	Unconditional Model, b (SE)	Model 1, b (SE)	Model 2, b (SE)	Model 3, b (SE)	Model 4, b (SE)	Model 5, b (SE)
Violent behavior at age 18 y	1.332*** (0.017)	1.240*** (0.072)	1.231*** (0.023)	1.243*** (0.020)	1.245*** (0.023)	1.250*** (0.020)
Gender (male)		0.193*** (0.104)	0.209*** (0.032)	0.183*** (0.030)	0.185*** (0.032)	0.169*** (0.029)
SES		−0.013 (0.014)	−0.010 (0.014)	−0.006 (0.012)	−0.018 (0.014)	−0.008 (0.012)
Alcohol use in Year 1			0.129*** (0.014)	0.032* (0.014)
8th grade GPA				−0.043** (0.013)		−.0.047*** (0.013)
Depression				0.020 (0.015)		0.017 (0.015)
Selling drugs				0.081*** (0.017)		0.077*** (0.016)
Friends' violence				0.091*** (0.016)		0.087*** (0.016)
Friends' drug use				0.002 (0.017)		0.015 (0.017)
Parent drug use				0.012 (0.014)		0.016 (0.014)
Parent violence				0.047*** (0.013)		0.047*** (0.013)
Family conflict				0.014 (0.013)		0.015 (0.013)
Age slope
Intercept	−0.026*** (0.005)	−0.024*** (0.007)	−0.022*** (0.006)	−0.023*** (0.006)	−0.023** (0.007)	−0.027*** (0.007)
Gender (male)		−0.003 (0.010)	−0.006 (0.010)	−0.003 (0.010)	−0.012 (0.009)	−0.011 (0.009)
Alcohol use in Year 1			−0.019*** (0.003)	−0.007* (0.003)
8th grade GPA				0.007* (0.003)		0.007* (0.003)
Depression				−0.011** (0.003)		−0.012*** (0.003)
Selling drugs				−0.023** (0.004)		−0.024*** (0.004)
Friends' violence				−0.012*** (0.003)		−0.011*** (0.003)
Friends' drug use				0.002 (0.004)		0.001 (0.004)
Parent drug use				0.004 (0.003)		0.005 (0.003)
Parent violence				−0.008* (0.003)		−0.006* (0.003)
Family conflict				0.000 (0.003)		0.000 (0.003)
Alcohol use					0.052*** (0.006)	0.049*** (0.006)
Alcohol use × age					−0.004** (0.001)	−0.002 (0.001)
Age2	−0.001 (0.001)	0.001 (0.002)	0.001 (0.002)	0.002 (0.002)	0.002 (0.002)	0.002 (0.002)
Gender (male)		−0.005* (0.002)	−0.005* (0.002)	−0.005* (0.002)	−0.004 (0.002)	−0.004 (0.002)
Depression				0.002* (0.001)		0.002* (0.001)
Selling drugs				0.003** (0.001)		0.003*** (0.001)
Age3	0.001* (0.0003)	0.000 (0.0004)	0.000 (0.0004)	0.000 (0.0004)	0.000 (0.0004)	0.000 (0.0004)
Gender (male)		0.001* (0.0006)	0.001* (0.0006)	0.001* (0.0006)	0.001* (0.0006)	0.001* (0.0006)

Note. GPA = grade point average; SES = socioeconomic status. Age slope is the linear component of the growth curve; age² is the quadratic component of the growth curve; age³ is the cubic component of the growth curve.

*P < .05; **P < .01; ***P < .001.

Further examination indicated gender differences in violent behavior trajectories (Table 2, model 1). We found a linear decrease in violent behavior among female participants from high school to emerging adulthood, whereas the trajectory among male participants was cubic: a nonlinear increase in high school and then a decline in emerging adulthood (a figure is available as a supplement to the online version of this article at http://www.ajph.org).

Alcohol Use and Violent Behavior Trajectory

Predictive relationship. In the model that included year 1 alcohol use as a predictor (Table 2, model 2), alcohol use was positively associated with violent behavior at age 18 years but negatively associated with the age slope after we controlled for gender and SES. A visual display of the results demonstrated that the association between alcohol use and violent behavior was stronger in adolescence than in emerging adulthood (Figure 2). After we adjusted for other common risk factors, the association between violent behavior and alcohol use and the age effect remained significant, but the magnitudes of the effects were reduced (Table 2, model 3).

FIGURE 2.

Predictive relationship between alcohol use in year 1 and violent behavior trajectory among urban African American youths: Michigan, 1994–2002.

Note. Values on the y-axis are ratings from the 5-point scale used to assess violent behavior in the preceding 12 months (ranging from 1[0 times] to 5[≥ 4 times]). Low level of alcohol use is 1 standard deviation below the mean; high level of alcohol use is 1 standard deviation above the mean.

Concurrent relationship. Alternative analyses that included alcohol use at each wave as a time-varying covariate showed a similar pattern. Change in alcohol use was positively associated with change in violent behavior from high school to emerging adulthood. We also found an interaction effect between alcohol use and age; the association between alcohol use and violent behavior was stronger in high school than it was in emerging adulthood (Table 2, model 4). A visual display of these results demonstrated similar trajectories as those shown in Figure 2, suggesting stronger association during adolescence than during emerging adulthood.

After we adjusted for common risk factors, the interaction between alcohol use and age dropped to nonsignificance, whereas the main effect of alcohol use remained the same (Table 2, model 5). Gender was not associated with the alcohol use slope and thus was not included in the final models shown in Table 2.

DISCUSSION

This study is among the few that have examined the longitudinal relationship between alcohol use and violence trajectories. Our findings of developmental trajectories in alcohol use and violent behavior in a sample of African American youths are consistent with those of other longitudinal studies,^26,28 suggesting that the development of these behaviors is not population specific. Moreover, our findings of gender differences in alcohol use and violent behavior trajectories add to our understanding of the development of these behaviors.

We focused on the general pattern of alcohol use and violent behavior rather than acute effects, examining the relationship between these 2 behaviors with longitudinal data across adolescence and emerging adulthood in a sample of at-risk African American adolescents. As a result, our findings provide more insight into the developmental associations between the 2 behaviors over time among African American youths than one finds in the current literature.

Our results support a bidirectional relationship between alcohol use and violent behavior^10,21; that is, we found that early violent behavior predicted later alcohol use, and early alcohol use predicted later violent behavior. Although the developmental trajectories of alcohol use and violent behavior differed between male and female participants, the relationships between the 2 behaviors did not vary according to gender.

We also assessed how changes in one behavior were associated with changes in the other and found a bidirectional relationship between the behaviors over time from adolescence to emerging adulthood. The reciprocal associations remained even after we controlled for several possible spurious variables. Our results suggest that the presence of one of these behaviors may increase the risk for the other.^14,39

Furthermore, we found that the strength of the associations between alcohol use and violent behavior varied depending on our participants' stage in the life cycle. In the case of both behaviors, the associations were stronger during adolescence than in emerging adulthood. After we controlled for the common risk factors, however, age effects on the associations between these 2 behaviors were no longer significant. This implies that the stronger associations between the 2 behaviors during adolescence might have been due to the risk factors assessed.

Our findings suggest that it may be more informative to examine the relationship between alcohol use and violent behavior from a developmental life course perspective because we can gain a better understanding of the trajectories of these behaviors. During adolescence, alcohol consumption is illegal and may reflect a degree of unconventionality among users. Adolescents who use alcohol may be less bonded to conventional norms and more deviant. Thus, the reason for the stronger reciprocal associations between alcohol use and violent behavior may be that they fit a single general problem behavior syndrome.¹³

In emerging adulthood, however, alcohol use is legal and socially acceptable, and it does not necessarily reflect rejection of social norms. Notably, we observed a decline in the strength of the relationship between alcohol use and violent behavior over time. Therefore, after control for common risk factors, the associations between the 2 behaviors no longer varied by age.

An alternative explanation for the attenuated association between these 2 behaviors is Moffitt's distinction between adolescence-limited and life-course–persistent antisocial behavior.⁴⁰ In our study, a reduction in the association between alcohol use and violent behavior was observed during emerging adulthood, when participants whose violent behavior was confined to their adolescent years continued to increase their alcohol use.

Future research examining the mechanisms that account for the longitudinal associations between alcohol use and violent behavior and the factors that protect African American youths from developing these deviant behaviors would be useful. We investigated alcohol use only in terms of frequency of use. Future research that assesses more serious drinking problems such as binge drinking or drunkenness may reveal different patterns of relationships in adolescence and emerging adulthood than those we found in this study.

Limitations and Strengths

Some limitations of this study should be noted. First, our sample included urban African American youths who were at risk for negative outcomes because of low school achievement. Therefore, our findings may not be generalizable to African American youths as a whole. However, they may be relevant for a subgroup of at-risk African American youths who have been understudied and may be especially informative for policies and programs designed to address the problems of alcohol use and violent behavior among these youths. It is also notable that by the 12th grade the GPA distribution was more evenly spread out over all grade levels.⁴¹

A second study limitation is our reliance on self-report data, which may be affected by social desirability. However, the paper-and-pencil questionnaire format reduced this concern somewhat. Finally, although we controlled for a set of common risk factors that are representative across different adolescent development domains, these risk factors were not exhaustive. Other risk factors not included in this study may have explained the associations between alcohol use and violent behavior.

Despite these limitations, our study involved several strengths. First, we focused on a large sample of urban African American youths at risk for alcohol and other drug use and violence, a population that has been understudied in the past. Second, we collected data from participants on multiple occasions across different life cycle stages, enabling us to investigate longitudinal associations between alcohol use and violent behavior from a developmental life course perspective. Finally, we not only tested how early levels of one behavior predicted later levels of the other but also examined how changes in one behavior were associated with changes in the other.

Conclusions

Our findings indicating that there are longitudinal associations between alcohol use and violent behavior have important implications for preventive interventions. Long-term associations between alcohol use and violent behavior suggest that prevention efforts may be most effective when they are initiated early in adolescence. Screening for alcohol use and violent behavior early in the life cycle may be useful in preventing progression of alcohol misuse and serious violence.⁴²

The bidirectional relationship between alcohol use and violent behavior observed here suggests that reducing one problem will reduce the other. Therefore, programs and policies aimed at reducing youth violence or alcohol use should take into account the link between these behaviors. Efforts to prevent violence and alcohol use should be combined and should target adolescents exhibiting multiple risk behaviors.