Prospective Associations between Delinquency and Suicidal Behaviors in a Nationally Representative Sample

Article Outline

• Abstract
• Methods
  • Sampling procedures and sample
    • Sample
  • Measures
    • Predictor variable: delinquency
    • Dependent variables: suicidal behaviors
    • Covariates
    • Interaction terms
  • Data analytic strategy
• Results
• Discussion
• References
• Copyright

Abstract 

Purpose

We examined the prospective associations between delinquency and suicidal behaviors among adolescents using a nationally representative sample.

Methods

The National Longitudinal Study of Adolescent Health was used to assess the prospective associations between delinquency at ages 12–17 years and suicidal behaviors (ideation, attempts, treatment for attempts) 1 and 7 years later.

Results

Controlling for demographic variables, delinquency was prospectively associated with an increased risk for all three suicidal behaviors 1 year later, and to ideations and attempts 7 years later. After controlling for other risk factors (i.e., depression, self-esteem, problem drinking, impulsivity, religiosity), delinquency was still significantly related to an increased likelihood for suicidal ideation 1 and 7 years later, and to suicide attempts 1 year later. Further, delinquency was significantly related to suicidal ideation for girls more than for boys.

Conclusions

These findings suggest that delinquent youth, particularly delinquent girls, should be targeted in interventions to reduce suicidal behaviors.

Keywords: Suicidal behavior, Delinquency, Gender differences

Suicide claimed the lives of more than 5000 12–26-year-olds in 2003 [1] and was the third leading cause of death for this age range, preceded only by unintentional injury and homicide. Clearly, suicide represents a significant public health problem among youth and younger adults in the United States.

The most significant predictors of suicide completions are nonfatal suicidal behaviors, such as attempts and ideations [2], [3]. Nationally representative data indicate that 8.4% of high school students attempted suicide in 2005 [4]. Further, 17% reported having seriously considered suicide during the past 12 months, and 2.3% reported that they had a suicide attempt that resulted in an injury, poisoning, or overdose that had to be treated by a nurse or doctor [4]. To prevent suicide, it is first necessary to identify the risk factors for suicidal behaviors so that appropriate intervention targets can be elucidated.

Research indicates several risk factors for suicidal behavior among youth, including depression, substance use, and impulsivity [5], [6], [7], [8], [9], [10], [11]. In addition to these risk factors, research also indicates that delinquency is associated with suicidal behaviors [6], [7], [8], [10], [12]. This finding is not surprising in light of Durkheim’s position that suicide rates vary inversely according to the degree of social integration [13]. Thus, individuals who are socially disaffiliated from the larger population are at greater risk for suicidal behavior, and research indicates that delinquents are less likely to be socially integrated [14].

Investigations on the association between suicidal behavior and delinquency have typically relied only on delinquent samples. Because these studies lacked comparison groups of nondelinquents, they were not able to estimate the relative risk of suicidal behavior associated with delinquency [2], [15], [16], [17], [18], [19]. Studies that have included comparison groups of nondelinquents have used community-based samples [8], [20], [21]. Two exceptions to this were studies that found a significant association between delinquent behaviors and suicidal behaviors in nationally representative samples of high school students [12], [22]. Both of these studies, however, used cross-sectional data and thus it was not possible to determine if delinquency predicted the risk for suicidal behavior over time.

The purpose of the current study was to examine prospective associations between delinquency and three measures of suicidal behaviors (seriously considered committing suicide, attempted suicide, had an attempt result in an injury, poisoning, or overdose that had to be treated by a doctor or nurse) in a nationally representative sample of youth and young adults. The youth were 12–17 years of age when their delinquent behaviors were assessed (Wave 1), and 1 year older (i.e., 13–18 years old; Wave 2) and 7 years older (i.e., 19–26 years old; Wave 3) when their suicidal behaviors were assessed. A prospective design enabled us to determine if delinquency predicted the initiation of suicidal behavior between Waves 1 and 2 and between Waves 1 and 3. Multivariate analyses that controlled for other risk factors for suicidal behavior, such as depression, allowed us to determine the unique role delinquency played in predicting suicidal behavior over time.

Back to Article Outline

Methods 

Sampling procedures and sample 

Data from the National Longitudinal Study on Adolescent Health (Add Health) were used to test the study hypotheses. Add Health is a longitudinal survey of adolescents who were in grades 7–12 at baseline. The study was funded primarily by the National Institute of Child Health and Human Development, and conducted by the Carolina Population Center at the University of North Carolina at Chapel Hill. The purpose of Add Health is to investigate factors that influence adolescents’ health and health-related behavior in areas such as diet, physical activity, health service use, morbidity, injury, violence, sexual behavior, contraception, sexually transmitted infections, pregnancy, suicidal intentions/thoughts, substance use/abuse, and runaway behavior.

The survey used a multistage stratified cluster design to sample high schools, both public and private, in the United States. To be eligible, a high school had to include an 11th grade and have at least 30 students. Of the 26,666 eligible high schools, a systematic random sample of 80 high schools was selected. Schools were stratified by region, urbanicity, school type, and percentage white. For each high school selected, the largest feeder school, most typically a middle school, was also recruited for study participation. Seventy-nine percent of the recruited schools agreed to participate, resulting in a sample size of 132 schools. Each school provided a student roster, and all students who were listed on the school rosters were eligible to be selected for the in-home survey. The participation rate was 75.6%, resulting in a sample of 18,924 adolescents who completed the Wave 1 in-home survey. These interviews were completed between April and December of 1995 and took from one to two hours to complete. Data were recorded directly onto laptop computers. For less sensitive questions, interviewers read the questions and entered respondents’ answers. For more sensitive questions, the respondent listened to prerecorded questions through earphones and entered the answers directly (audio-CASI). This procedure reduced the potential for respondents’ answers to be influenced by either their parents or the interviewer.

All 7th–11th graders who completed an in-home survey at Wave 1 were eligible to be interviewed for a second time 1 year later (Wave 2), and a third time 7 years later (Wave 3). Approximately 88% of the eligible participants completed a Wave 2 in-home survey, and approximately 73% of eligible participants completed a Wave 3 in-home survey. For the purposes of the current study, only youth aged 12–17 years at the time of the Wave 1 interviews were included in the analyses. Respondents older than 17 years (n = 3198) were excluded, as they could not be classified as delinquents due to their adult status. We also excluded adolescents who reported having attempted suicide at Wave 1 (n = 692). Therefore, the analyses indicate if delinquency predicted the initiation of suicidal behavior between Waves 1 and 2 and between Waves 1 and 3. These restrictions resulted in a sample size of 15,034.

Measures 

Fifteen items were used to measure delinquent behaviors occurring over the past 12 months. Each item was answered on a four-point Likert scale ranging from 0 “not at all” to 3 “5 or more times.” Sample items included “How often did you deliberately damage property that didn’t belong to you?”; “Take something from a store without paying for it?”; “Hurt someone badly enough to need bandages or care from a doctor or nurse?”; “Drive a car without its owner’s permission?”; and “Steal something worth more than $50?” The scale has been used in prior research with Add Health data, and has good internal consistency reliability (α = .84). Because the sum of the continuously scored items was highly skewed, we created a dichotomous measure of delinquency. We first dichotomized each of the 15 items (0 = never engaged in, 1 = engaged in more than once) and then summed the items. Next, the summed distribution was dichotomized using the top 22% (most delinquent) vs. the rest, as this is how delinquency has been measured in extant research with Add Health data [23]. The top 22% included those respondents who had engaged in at least five of the 15 delinquent behaviors.

We included three measures of suicidal behaviors, including the following: (1) if the respondent reported having seriously considered committing suicide during the past 12 months (1 = yes, Wave – 2 8.3%, Wave 3 – 4.8%); (2) if the respondent attempted suicide during the past 12 months (1 = 1 or more times, Wave 2 – 2.2%, Wave 3 – 1.2%); and (3) if the respondent had any suicide attempt that resulted in an injury, poisoning, or overdose that had to be treated by a doctor or nurse (1 = yes, Wave 2 – .4%, Wave 3 – .4%). For all three measures, the reference group was coded as 0.

We controlled for three person-level demographic variables (age, gender, ethnicity) and one environmental-level (metropolitan status) demographic variable assessed at Wave 1. Age (0 = 12-, 13-, and 14-year-olds, 40%; 1 = 15-, 16-, and 17-year-olds, 60%), gender (0 = male, 51%; 1 = female, 49%), race (0 = white, 74%; 1 = nonwhite, 26%), and metropolitan status (0 = completely urban, 52%; 1 = not completely urban, 48%) were coded dichotomously.

We also controlled for five other risk factor variables that have been shown to be related to suicidal behavior. These five covariates included alcohol problems, depression, self-esteem, impulsivity, and religiosity. Respondents were classified as having an alcohol problem (14%) if they reported that they had been drunk at least 3–12 times or had experienced negative consequences of alcohol use at least twice in each of three or more areas in the past year [23]. The sum of 18 items answered using a four-point Likert scale was used to assess depressive symptoms experienced during the 2 weeks before the Wave 1 assessments (M = 10.70, SD = 6.97). Sample items included “you thought your life had been a failure,” and “you felt sad.” This scale has been used in prior research with Add Health data [23] and has good internal consistency reliability (α = .86). The mean of six items answered on a five-point Likert scale was used to assess self-esteem at Wave 1 (M = 4.13, SD = .58). Sample items included “you have a lot to be proud of,” and “you like yourself just the way you are.” This scale also has been used in prior research with Add Health data [23], and has good internal consistency reliability (α = .83). The mean of three items that were answered on a five-point Likert scale was used to assess impulsivity (M = 2.20, SD = .66). Sample items included “when you have a problem to solve, one of the first things you do is get as many facts about the problem as possible,” and “when making decisions, you generally use a systematic method for judging and comparing alternatives.” The scale had adequate internal consistency reliability (α = .70). Religiosity was assessed by asking respondents how important religion was to them (1 = very important or fairly important, 78%), 0 = fairly unimportant or not important at all, 22%).

We computed three interaction terms to represent the cross-products of gender and delinquency, race and delinquency, and age and delinquency. The components of the interaction terms were centered before computing the cross-products to reduce multicollinearity between the main effect variables and the interaction term [24].

Data analytic strategy 

Survey responses were weighted to provide data representative of youth in the United States. These sample weights adjusted for stratification and over-sampling of underrepresented groups. We used SPSS 14.0 (SPSS Inc., Chicago, IL) to analyze the data. SPSS Complex Samples was used to account for the complex sampling design of Add Health. When testing the longitudinal associations between delinquency and suicidal behavior, the sample was restricted to just those adolescents who had not attempted suicide at Wave 1. Therefore, the analyses indicated if delinquency predicted the initiation of suicidal behavior between Waves 1 and 2 and between Waves 1 and 3.

We first tested the bivariate associations between delinquency at Wave 1 and the three suicidal behavior variables at Waves 2 and 3. We then entered the demographic variables into the model to determine the effects of delinquency at Wave 1 on suicidal behaviors at Waves 2 and 3 not due to demographic variables. Next, we added the five risk factor variables into the multivariate logistic regression models to assess the unique effects of delinquency on suicidal behaviors, while holding constant the demographic variables and the risk factor variables. To determine if delinquency had differential effects on suicidal behaviors in terms of the respondents’ gender, race, or age, we conducted moderational analyses using standard criteria for testing moderation [25]. Specifically, moderation was shown if the interaction term between the predictor (delinquency) and the hypothesized moderator (e.g., gender) was significantly related to the dependent variables (suicidal behaviors) after controlling for the main effects of both the predictor and hypothesized moderator variable. The effects of each of the three interaction terms were tested simultaneously for each of the three outcome variables. Last, we conducted post hoc probing of significant moderational effects [26].

Back to Article Outline

Results 

Results from the bivariate analyses indicated that delinquency was prospectively associated with all three suicidal behaviors 1 year later (Wave 2). Specifically, delinquent youth were significantly more likely than their counterparts to have seriously considered suicide (crude odds ratio [COR] 1.71, 95% confidence intervals [CI] 1.46–1.99), attempted suicide (COR 2.00, 95% CI 1.45–2.75), and required medical treatment after an attempt (COR 2.02, 95% CI 1.02–4.00) 1 year later. Delinquency remained a significant predictor of suicidal ideation when the respondents were 7 years older (Wave 3). Specifically, delinquents were more likely than their nondelinquent counterparts to have seriously considered suicide (COR 1.56, 95% CI 1.22–2.01) when they were aged 19–26 years old.

We next examined the prospective associations between delinquency and suicidal behavior with the demographic covariates in the model. After controlling for age, race, gender, and metropolitan status, delinquency was prospectively related to all three measures of suicidal behaviors 1 year later, and to two of the measures (ideation, attempt) seven years later. One year later, delinquent youth were significantly more likely to have seriously considered suicide (adjusted odds ratio [AOR] 1.97, 95% CI 1.68–2.32), attempted suicide (AOR 2.41, 95% CI 1.74–3.33), and required medical treatment after a suicide attempt (AOR 2.21, 95% CI 1.06–4.58), compared with their less delinquent counterparts. Youth classified as delinquents at ages 12–17 years were significantly more likely than their less delinquent counterparts to have seriously considered suicide (AOR 1.64, 95% CI 1.28–2.09) and attempted suicide (AOR 1.60, 95% CI 1.01–2.54) 7 years later at ages 19–26 years.

Next, we added the other risk factor covariates into the model. Results from these multivariate logistic regression models are presented in Table 1. Findings regarding the covariates indicated that younger age at Wave 1 was predictive of suicide behaviors, such that youth ages 12–14 years at Wave 1 were more likely than youth ages 15–17 years at Wave 1 to have seriously thought about suicide and attempted suicide 1 and 7 years later. Females were significantly more likely than males to have seriously considered suicide 1 year later, and to have attempted suicide both 1 and 7 years later. Minority adolescents were less likely to have thought about suicide 1 year later, but there were no race differences for attempting suicide or requiring medical treatment. There also were no differences in the odds of engaging in any of the suicidal behaviors based on respondent’s metropolitan status, problem drinking status, or impulsivity. Youth lower in self-esteem were significantly more likely than their counterparts with higher levels of esteem to have seriously considered suicide one and seven years later. Adolescents reporting higher levels of depressive symptoms were significantly more likely than their less depressed counterparts to have thought seriously about suicide 1 and 7 years later, and to have attempted suicide and required medical treatment for a suicide attempt 1 year later. Youth who reported that religion was important to them were less likely than their less religious counterparts to have required medical treatment after a suicide attempt 7 years later.

Table 1. Adjusted odds ratios (AORs) and 95% confidence intervals (CI) for predicting suicidal behaviors 1 and 7 years later—United States, National Longitudinal Study of Adolescent Health

After controlling for all nine covariates, delinquency was prospectively related to two of the three measures of suicidal behaviors (ideation, attempt) 1 year later, and to one of the measures (ideation) 7 years later. Specifically, youth classified as delinquent were significantly more likely to have seriously considered suicide 1 and 7 years later, and to have attempted suicide 1 year later compared with their nondelinquent counterparts.

The interaction term representing the cross product of delinquency and gender was significant for thinking about suicide one year later (Wave 2). To interpret the nature of the interaction term, we conducted post hoc probing of the moderational effect. We computed two conditional moderator variables and ran a regression model for each conditional moderator variable. In the first model, males were assigned a score of 0, and in the second model, females were assigned a score of 0. We then derived AORs and 95% CI for the associations between delinquency and suicidal ideation 1 year later. In this way, we were able to determine if delinquency was significantly associated with suicidal ideation for males only, for females only, or for both genders but in differing magnitudes.

Results from these analyses, controlling for all nine covariates, indicated that delinquency was not significantly related to seriously considering suicide for males (AOR 1.21, 95% CI .95–1.55) but it was for females, (AOR 1.71, 95% CI 1.34–2.17). Of note, when only the demographic covariates were included in the model but not the five risk factor variables, delinquency was significantly related to suicidal ideation 1 year later for both males (AOR 1.56, 95% CI 1.23–1.98) and for females (AOR 2.37, 95% CI 1.90–2.95), but the magnitude of effect was greater for females.

Back to Article Outline

Discussion 

Data from a nationally representative sample indicated that delinquent behavior at ages 12–17 years was prospectively associated with suicidal behavior 1 and 7 years later. Specifically, after controlling for demographic variables (age, race, gender, metropolitan status), delinquency was prospectively associated with an increased risk for seriously considering suicide, making a suicide attempt, and requiring medical treatment after an attempt 1 year later, and with an increased risk for ideation and attempt 7 years later. Of particular note, most of these effects persisted even after controlling for depression, self-esteem, problem drinking, impulsivity, and religiosity. Delinquency was still significantly associated with an increased likelihood for seriously considering suicide 1 and 7 years later, and with attempting suicide 1 year later.

Moderator analyses revealed that the associations between delinquency and suicidal behaviors did not vary by race or age, but did vary by gender. Whereas both delinquent boys and girls were significantly more likely to consider suicide 1 year later, this association was stronger for delinquent females than for delinquent males. However, when other risk factor variables were in the model, delinquency was associated with suicidal ideation for girls only. This finding suggests that for boys, the association between delinquency and suicidal ideation may be due to one of the risk factor variables. However, for girls, the association between delinquency and suicidal ideation persisted even after controlling for these other risk factors, suggesting that depression, self-esteem, problem drinking, impulsivity, and religiosity did not account for delinquent girls’ increased risk for ideation.

Although this study had several strengths, including using nationally representative and prospective data, it was limited in its exclusion of youth who were not in school. Only youth on the school rosters were eligible to participate in the Add Health in-home survey. Consequently, a group of youth at high risk for both delinquency and suicidal behavior is not captured in this study.

The findings in this study support and extend findings from a prior study using nationally representative data from the Youth Risk Behavior Surveillance (YRBS) Survey. In the prior study, delinquency was significantly associated with four types of suicidal behavior, three of which paralleled those examined in the current study. However, YRBS data were cross-sectional, and so it was not possible in the earlier study to determine if delinquency predicted suicidal behavior over time.

Although the current study extends the prior findings by showing delinquency was related to suicidal behaviors prospectively, it still does not help explain why delinquent youth are at greater risk for suicidal behavior. Studies are needed to examine the role of mediators or explanatory variables to determine what accounts for the increased risk for suicidal behavior among those who engage in delinquent behaviors. It also is worth noting that delinquency was more strongly associated with suicidal behaviors for girls than boys in both surveys (YRBS, Add Health). Why would delinquent girls be at increased risk for suicidal ideation? One possibility is that delinquency is more normative for boys than girls [27]. In our sample, the percentage of boys classified as delinquents (27%) was significantly higher than the percentage of girls classified as delinquents (16%) (χ² = 308.35). This suggests that girls who engage in delinquent behaviors may differ more from their less delinquent female counterparts on a range of variables indicative of poor functioning than delinquent boys differ from their less delinquent male counterparts. Another possible explanation for the increased suicidal behavior risk associated with delinquency among girls is that the epidemiology of delinquency varies by gender. Although research indicates consistency in the types of delinquent behaviors exhibited by boys and girls, there are differences in predictors of risk [28]. Research with adjudicated youth indicates that female delinquents have significantly higher rates of psychopathology, child maltreatment, other traumatic experiences, and familial risk factors than delinquent boys [29], [30]. Many of these risk factors for delinquency among girls are risk factors for suicidal behaviors in females [31], [32], and this common set of risk factors may confer a trajectory that promotes both delinquency and suicidal behaviors for girls. Future research should attempt to understand why delinquent females are at risk for suicidal ideation. Studies that seek to explain the increased risk for suicidal behaviors among delinquent girls are needed so that interventions can be targeted to address these factors.

Our finding that delinquency is associated with suicidal behaviors has important implications for the prevention of morbidity and mortality associated with suicidal behaviors among youth. Specifically, the findings suggest the need for targeted interventions to prevent suicide with this population. This could be done in juvenile justice settings as well as school settings. Research on gender differences in suicidal behaviors suggests that suicide-related interventions may need to be gender-specific. Given that males report significantly higher levels of risk-taking behaviors than females, and females report significantly higher levels of depressive symptoms than males, interventions for females may be most effective when they focus on gender-specific styles of coping, and interventions for males may be most effective when they focus on learning alternative activities that do not promote risk-taking [33]. Given that suicide was the third leading cause of death for 12–26-year-olds in 2003 [1], and that most suicides are preceded by nonfatal suicidal behaviors [2], it is key that we identify factors that increase the risk for suicidal behaviors and target these factors in intervention programs.

Back to Article Outline

References 

1. Centers for Disease Control and Prevention. Web-Based Injury Statistics Query and Reporting System (WISQARS), 2003 [cited 2006 July 5]. National Center for Injury Prevention and Control, Centers for Disease Control and Prevention. Available from: http://www.cdc.gov/ncipc/wisqars.
   • View In Article
2. Battle A, Battle MV, Tolley EA. Potential for suicide and aggression in delinquents at juvenile court in a southern city. Suicide Life Threat Behav. 1993;23:230–244
   • View In Article
   • MEDLINE
3. Marttunen MJ, Aro H, Lonnqvist JK. Adolescent suicides: endpoint of long-term difficulties. J Am Acad Child Adolesc Psychiatry. 1992;31:649–654
   • View In Article
   • Abstract
   • Full-Text PDF (5740 KB)
   • CrossRef
4. Eaton DK, Kann L, Kinchen S, et al. Youth risk behavior surveillance—United States, 2005. MMWR Surveill Summ. 2006;55(5):1–108
   • View In Article
5. Cole DA. Psychopathology of adolescent suicide: hopelessness, coping beliefs, and depression. J Abnorm Psychol. 1989;9:248–255
   • View In Article
6. Evans E, Hawton K, Rodham K. Factors associated with suicidal phenomena in adolescents: a systematic review of population-based studies. Clin Psychol Rev. 2004;24:957–979
   • View In Article
   • MEDLINE
   • CrossRef
7. Goldsmith SK, Pellmar TC, Kleinman AM. Reducing Suicide: A National Imperative. Washington, DC: National Academics Press; 2002;
   • View In Article
8. King RA, Schwab-Stone M, Flisher AJ, et al. Psychosocial and risk behavior correlates of youth suicide attempts and suicidal ideation. J Am Acad Child Adolesc Psychiatry. 2002;40:837–846
   • View In Article
   • Abstract
   • Full Text
   • Full-Text PDF (135 KB)
   • CrossRef
9. Lewinsohn PM, Rohde P, Seeley JR. Psychosocial risk factors for future adolescent suicide attempts. J Consult Clin Psychol. 1994;62:297–305
   • View In Article
   • MEDLINE
   • CrossRef
10. U.S. Public Health Service. Surgeon General’s Call to Action to Prevent Suicide. Washington, DC: U.S. Public Health Service; 1999;
    • View In Article
11. Woods ER, Lin YG, Middleman A, et al. The associations of suicide attempts in adolescents. Pediatrics. 1997;99:791–796
    • View In Article
    • CrossRef
12. Thompson MP, Kingree JB, Ho C. Associations between delinquency and suicidal behaviors in a nationally representative sample of adolescents. Suicide Life Threat Behav. 2006;36:57–64
    • View In Article
    • MEDLINE
    • CrossRef
13. Durkheim E. Suicide. London, UK: Routledge and Kegan Paul; 1952;
    • View In Article
14. Junger-Tas J. An empirical test of social control theory. J Quant Criminol. 1992;8:9–28
    • View In Article
15. Kempton T, Forehand R. Suicide attempts among juvenile delinquents; the contribution of mental health factors. Behav Res Ther. 1992;30:537–541
    • View In Article
    • MEDLINE
    • CrossRef
16. Morris RE, Harrison EA, Knox GW, et al. Health risk behavioral survey from 39 juvenile correctional facilities in the United States. J Adolesc Health. 1995;17:334–344
    • View In Article
    • Abstract
    • Full-Text PDF (986 KB)
    • CrossRef
17. Penn JV, Esposito CL, Schaeffer LE, et al. Suicide attempts and self-mutilative behavior in a juvenile correctional facility. J Am Acad Child Adolesc Psychiatry. 2003;42:762–769
    • View In Article
    • Abstract
    • Full Text
    • Full-Text PDF (122 KB)
    • CrossRef
18. Rohde P, Seeley JR, Mace DE. Correlates of suicidal behavior in a juvenile detention population. Suicide Life Threat Behav. 1997;27:164–175
    • View In Article
    • MEDLINE
19. Ruchkin VV, Schwab-Stone M, Koposiv RA, et al. Suicidal ideations and attempts in juvenile delinquents. J Child Psychol Psychiatry. 2003;4:1058–1066
    • View In Article
20. Durant RH, Krowchuk DP, Kreiter S, et al. Weapon carrying on school property among middle school students. Arch Pediatr Adolesc Med. 1999;15:21–26
    • View In Article
21. Orpinas PK, Basen-Engquist K, Grunbaum J, et al. The co-morbidity of violence-related behaviors with health-risk behaviors in a population of high school students. J Adolesc Health. 1995;1:216–225
    • View In Article
22. Centers for Disease Control and Prevention. Suicide attempts and physical fighting among high school students—United States, 2001. MMWR. 2004;53:474–476
    • View In Article
23. Swahn MH, Donovan JE. Correlates and predictors of violent behavior among adolescent drinkers. J Adolesc Health. 2004;34:480–492
    • View In Article
    • Abstract
    • Full Text
    • Full-Text PDF (120 KB)
    • CrossRef
24. Aiken LS, West SG. Multiple Regression: Testing and Interpreting Interactions. Newbury Park, CA: Sage; 1991;
    • View In Article
25. Baron RM, Kenny DA. The moderator-mediator variable distinction in social psychological research: conceptual, strategic, and statistical considerations. J Pers Soc Psychol. 1986;51:1173–1182
    • View In Article
    • MEDLINE
    • CrossRef
26. Holmbeck GN. Post-hoc probing of significant moderational and mediational effects in studies of pediatric populations. J Pediatr Psychol. 2002;27:87–96
    • View In Article
    • MEDLINE
    • CrossRef
27. Huizinga D, Loeber R, Thornberry TP, et al. Co-occurrence of delinquency and other problem behaviors. Juvenile Justice Bulletin. Washington, DC: Department of Justice; 2000;
    • View In Article
28. Gorman-Smith D, Loeber R. Are developmental pathways in disruptive behaviors the same for girls and for boys?. J Child Fam Stud. 2005;14:15–27
    • View In Article
29. Grisso T, Barnum R, Fletcher KE, et al. Massachusetts youth screening instrument for mental health needs of juvenile justice youths. J Am Acad Child Adolesc Psychiatry. 2001;40:541–548
    • View In Article
    • Abstract
    • Full Text
    • Full-Text PDF (120 KB)
    • CrossRef
30. McCabe KM, Lansing AE, Garland A, et al. Gender differences in psychopathology, functional impairment, and familial risk factors among adjudicated delinquents. J Am Acad Child Adolesc Psychiatry. 2002;41:860–867
    • View In Article
    • Abstract
    • Full Text
    • Full-Text PDF (116 KB)
    • CrossRef
31. Kaslow NJ, Thompson MP, Okun A, et al. Risk and protective factors for suicidal behavior in abused, African American women. J Consult Clin Psychol. 2002;70:311–319
    • View In Article
    • MEDLINE
    • CrossRef
32. Wunderlich U, Bronisch T, Wittchen H, et al. Gender differences in adolescents and young adults with suicidal behaviour. Acta Psychiatr Scand. 2001;104:332–339
    • View In Article
    • MEDLINE
    • CrossRef
33. Langhinrichsen-Rohling J, Lewinsohn P, Rohde P, et al. Gender differences in the suicide-related behaviors of adolescents and young adults. Sex Roles. 1998;39:839–854
    • View In Article