To examine the association between carrying a weapon at school and the age of onset of substance use, other indicators of violence, and other health risk behaviors among middle school students.
In 1995, a modified version of the Centers for Disease Control and Prevention Youth Risk Behavior Survey was administered to 2227 students (49% were female) attending 53 (of 463) randomly selected middle schools in North Carolina. Weapon carrying on school property during school hours was measured with 2 questions assessing carrying a gun and carrying other weapons such as knives or clubs. The Youth Risk Behavior Survey also assessed other indicators of violence, drug use, suicide plans and attempts, and being threatened with a weapon at school. Variables significantly (P≤.001) associated with gun and other weapon carrying by χ2 tests were analyzed with stepwise logistic regression using the likelihood ratio approach. Odds ratios (ORs) were adjusted for all other variables in the model and 95% confidence intervals (CIs) were computed.
Our study showed that 3% of students had carried a gun and 14.1% had carried a knife or club to school. Gun carrying was associated with increased age (OR, 1.57 [95% CI, 1.15-2.14]); male sex (OR, 5.62 [95% CI, 2.42-13.03]); minority ethnicity (OR, 3.30 [95% CI, 1.55-5.05]); and earlier age of onset of cigarette (OR, 0.85 [95% CI, 0.74-0.97]), alcohol (OR, 0.81 [95% CI, 0.71-0.94]), marijuana (OR, 0.81 [95% CI, 0.71-0.92]), and cocaine use (OR, 0.73 [95% CI, 0.62-0.86]). Knife or club carrying was associated with age (OR, 1.32 [95% CI, 1.14-1.53]); male sex (OR, 2.39 [95% CI, 1.77-2.32]); and earlier age of onset of cigarette (OR, 0.88 [95% CI, 0.84-0.94]), alcohol (OR, 0.81 [95% CI, 0.76-0.86]), and marijuana use (OR, 0.77 [95% CI, 0.72-0.83]). Gun carrying was also associated with frequency of cigarette (OR, 1.34 [95% CI, 1.14-1.57]), alcohol (OR, 4.59 [95% CI, 1.27-16.58]), cocaine (OR, 2.96 [95% CI, 1.29-6.82]), and marijuana use (OR, 3.66 [95% CI, 1.67-8.06]) after adjusting for male sex and minority ethnicity. Carrying a knife or club was associated with carrying a gun (OR, 1.83 [95% CI, 1.31-2.55]); being threatened with a weapon at school (OR, 1.65 [95% CI, 1.10-2.49]); fighting (OR, 4.62 [95% CI, 2.56-8.37]); frequency of alcohol (OR, 2.91 [95% CI, 1.88-4.50]) and cigarette use (OR, 1.20 [95% CI, 1.10-1.31]); and a suicide plan (OR, 1.54 [95% CI, 1.07-2.20]).
Middle school students are more likely to carry a knife or club (14.1%) than a gun (3%) to school. Young adolescents who initiate substance use early and engage in it frequently are more likely to carry guns and other weapons to school, after adjusting for age, sex, and ethnicity. Being threatened with a weapon at school and fighting were only associated with knife or club carrying at school. These findings suggest that school-based prevention programs targeting both violence prevention and substance use should be introduced in elementary school.
VIOLENCE IS a major public health problem in the United States that affects youth from all racial and socioeconomic grops.1- 12 We expect schools to provide a safe environment so that effective learning can take place; however, the prevalence of violence in schools is increasing.7,13- 19 The Centers for Disease Control and Prevention reported that in 1993, 11.8% of high school students had recently carried a weapon to school and 16.2% had been in a fight at school.20 DuRant et al7 found that 15% of male students and 5% of female students in grades 9 through 12 in Massachusetts had carried a weapon, including guns, knives, or clubs, to school in the 30 days prior to the survey. The frequency of weapon carrying at school was associated with the frequency of fighting, alcohol and cigarette use, and threat and injury with a weapon at school, and being offered, sold, or given drugs on school property. This study indicates that weapon carrying in high schools is prevalent and that the clustering of other behaviors among high school students who carry weapons places many students at risk of injury or death.
Less is known about weapon carrying and other violence in middle schools. In a study of 744 students attending 2 middle schools in predominantly low-income, African American neighborhoods in North Carolina, Cotton et al14 reported that 19% had carried a weapon to school, 37% had fought at school, and 18% had been suspended from school for fighting. Weapon carrying at school was associated with male sex and higher scores on an aggression scale, suggesting that weapons were taken to school for aggressive purposes. Arria et al21 recently reported that 17% of middle school–aged adolescents in Baltimore had carried a lethal weapon for protection or defense during a 1-year interval. Defensive weapon carrying was also associated with self-reported fears, worry, and deviant peer affiliations; however, carrying a weapon to school for protection was not examined. In a study of 1503 seventh and eighth grade students in Illinois, Bailey et al13 found that 15% had brought a weapon to school in the previous month, but the types of weapons they brought to school were not studied. Weapon carrying was associated with male sex, single parent family, drinking heavily, fighting, damaging school property, and the belief that other students brought weapons to school. Unlike the study of DuRant et al7 of high school students, Bailey et al13 found that weapon carrying was not associated with victimization and fear for safety in school.
Additional research is needed to describe the characteristics of younger adolescents who bring weapons to middle schools and to determine if factors associated with bringing guns to school differ from those associated with carrying less lethal weapons, such as knives and clubs. The purpose of our study was to examine the associations between carrying a gun, knife, or club to school and fighting, victimization at school and use of tobacco, alcohol, and other drugs. We also analyzed whether having made a suicide plan was associated with carrying a weapon to school.22 In addition, based on previous research,23 we tested the hypothesis that an earlier age of onset of drug use would be associated with carrying a weapon to school.
The Centers for Disease Control and Prevention developed the Youth Risk Behavior Survey (YRBS) to assess the prevalence of health risk behaviors among ninth through 12th grade students across the United States.20,24 In the spring of 1995, a modified and shortened version of the YRBS was administered to 2227 randomly selected sixth through eighth grade students attending 53 randomly selected public middle schools in North Carolina.24 This represented the 463 middle schools and 261,309 public middle school students in the state. The response rate for schools was 74% and the response rate for students was 86%. Student participation was voluntary, and students were assured of confidentiality during administration of the survey. Demographic data on age, grade, sex, ethnicity, family type, and students' self-evaluations of type of student are given in Table 1.
Weapon carrying at school was assessed with 2 questions: "Have you every carried a gun on school property during school hours?" and "Have you ever carried any other weapons (such as a knife or club) on school property during school hours?" In addition, students were asked "Has anyone ever threatened or injured you with a weapon such as a gun, knife, or club on school property?"
Using standard YRBS questions,7,20,22,23 the North Carolina middle school YRBS also measured gun and other weapon carrying in general; fighting; frequency of cigarette, alcohol, marijuana, cocaine, crack, inhalant, anabolic steroid, and injection drug use; and age of onset of cigarette, alcohol, marijuana, and cocaine use. Three questions were asked about suicide. Because younger adolescents were more likely to consider suicide than attempt it, we used the question "Have you ever made a plan to kill yourself?" instead of the question assessing previous attempts.22
Demographic variables included age, sex, school grade, ethnicity, and family composition. The last 2 variables were recoded as white vs minority racial or ethnic group and living in a home with 1 or 2 parents or adult guardians. Students were also asked what kind of student they were on a 7-point scale ranging from "one of the best" to "near the bottom." Over 90% of students rated themselves between "in the middle" to "one of the best."
Investigators from the Centers for Disease Control and Prevention and Westat (Westat Inc, Rockville, Md) published the results from a test-retest reliability study of the 1992 YRBS.25 They administered the YRBS questionnaire twice, 14 days apart, to 1679 students from grades 7 to 12.25 The authors computed a κ statistic for the 53 self-reported items and compared group prevalence estimates from 2 testing occasions. The κ for the entire test ranged from 0.145 to 0.911, with 71.7% of the items considered to have good to excellent reliability (κ=0.61-1.00)
Weighting procedures were used to correct for the sampling scheme in the data collection. Westat (Westat Inc) performed the data cleaning and the computation of the weights under contract by the Centers for Disease Control and Prevention. Weighting compensated for nonresponse and reflected the likelihood of sampling for each student. The weight used for estimation is given by the following: W indicates W1*W2*f1*f2*f3; W1, inverse of the probability of school selection; W2, inverse of the probability of classroom selection; f1, a school-level nonresponse of adjustment factor calculated by school size; f2, a student-level nonresponse adjustment factor calculated by school size; and f3, a poststratification adjustment factor by sex and grade.
Pairwise associations between categorical variables were examined using Pearson χ2 tests, followed by the Cramer V statistic. Bivariate analyses of differences in students who carry a gun vs students who do not carry a gun, students who carry other weapons on school property vs students who do not carry weapons on school property, and ordinal variables were tested with Kruskal-Wallis analysis of variance tests. Using methods described in previous studies,7,22 we controlled for the effect that within-school clustering of behaviors would have on the P values by lowering the P value for the bivariate analyses to .001. Variables found to be significantly (P<.001) associated with carrying a gun, knife, or club to school were entered into a forward stepwise, multiple logistic regression model using the likelihood ratio approach. The models are built so that variables are only added to the model if they can make significant contributions to the model, after adjusting for shared variation between variables already in the model. This approach avoids the problem of multicollinearity when highly correlated, independent variables are forced into a model. The odds ratios and 95% confidence interval are adjusted for all other variables in the model.
Carrying a gun to school occurred more often among adolescent boys (5%) than adolescent girls (0.8%) (Table 2). Carrying a gun to school was also associated with minority ethnicity and living in a single parent home. There was no significant relationship between school grade and carrying a gun to school. Similarly, a higher percentage of adolescent boys (20.2%) reported having carried a knife or club to school than adolescent girls (7.7%). Eighth grade students were more likely to carry a knife or club to school than sixth and seventh grade students (Table 2). There was no relationship between race/ethnicity and family composition and carrying other weapons to school.
There was a significant (P<.001) but weak association (Cramer V, 0.27) between carrying a gun and carrying other weapons to school. Two thirds of students who had carried a gun to school also carried other weapons on school property during school hours. Only 14.2% of students who had carried other weapons at school had also carried a gun at school.
Carrying a gun or other weapons at school was weakly associated with older age, a lower self-assessment of academic achievement, having carried a gun, knife, or club in general, being threatened or injured with a weapon at school, and fighting (Table 3). Carrying a gun and carrying a knife or club at school were also associated with tobacco, alcohol, and other substance use (Table 4). The strongest associations were found between tobacco and marijuana use and carrying either a gun or other weapon on school property. Weapon carrying at school was weakly associated with suicidal behaviors (Table 5).
When analyzed with multiple logistic regression, the number of days the students had smoked during the previous 30 days had the strongest association with carrying a gun on school property (Table 6). While students who had only smoked 1 or 2 days during the previous month were 1.34 times more likely than nonsmokers to carry a gun to school, students who smoked every day were 8 times more likely than nonsmokers to carry a gun to school. Adolescent boys were 7 times more likely than adolescent girls to report carrying a gun on school property. In order of magnitude of the adjusted ORs, carrying a gun to school was also associated with previous alcohol and marijuana use, minority racial/ethnic group, and cocaine use.
Smoking also had the strongest correlate with carrying a knife or club on school property, but the relationship was weaker than that of carrying a gun to school (Table 6). Students who smoked during all 30 days prior to the YRBS were 7.2 times more likely than nonsmokers to carry a knife or club to school. Unlike what was found for carrying a gun, students reporting that they had previously been in a fight were 4.62 times more likely than nonfighters to have carried a knife or club to school. In order of magnitude of the adjusted ORs, other weapon carrying at school was associated with previous alcohol use, male sex, carrying a gun, having been threatened or injured with a weapon at school, and having previously made a suicide plan (Table 6).
Carrying a gun and carrying a knife or club to school were associated with an earlier age of onset of cigarette, alcohol, marijuana, or cocaine use (Table 7). When analyzed with multiple logistic regression, after adjusting for the effects of age, sex, and ethnicity, age of onset of cigarette, alcohol, marijuana, and cocaine use remained significant, independent correlates with carrying a gun to school (Table 8). After adjusting for the significant effects of age and sex, age of onset of cigarette, alcohol, and marijuana use were significant independent correlates of carrying a knife or club to school. Age of onset of cocaine use was no longer significant after the effects of these other 3 substances, age, and sex were taken into account.
Three previous studies of middle school students have reported the prevalence of weapon carrying on school property as 19% for ever carrying a weapon to school,14 17% for carrying a weapon for protection during the previous 12 months,21 and 15% for carrying a weapon during the previous month.13 In 2 of these studies,14,21 the students were predominantly low income and from minority racial or ethnic groups. The third study was designed to represent all seventh and eighth grade students in Illinois.13 In a representative sample of sixth, seventh, and eighth grade students in North Carolina, we found that 3% of students had carried a gun on school property and 14.1% had carried a knife or club to school. There were significant sex differences in weapon carrying at school with 5% of adolescent boys vs 0.8% of adolescent girls reporting to have carried a gun to school and 20.2% of adolescent boys vs 7.7% of adolescent girls having carried a knife or club on school property. Surprisingly, the proportion of middle school students in North Carolina carrying weapons to school was higher than DuRant et al7 found among high school students in Massachusetts. Responding to the question, "During the past 30 days, on how many days did you carry a weapon such as a gun, knife, or club on school property?" 15% of males and 5% of females reported carrying a weapon one or more times. These differences may be because of the different time frames for questions measuring weapon carrying on the Massachusetts YRBS and the North Carolina YRBS; however, differences in the availability of guns and other weapons in North Carolina vs Massachusetts may also account for these findings.
Previous studies have reported that weapon carrying, use of violence, illicit drug use, and other risk and problem behaviors cluster among adolescents.5- 7,12,17- 19,22,23 DuRant et al7 recently reported that among high school students, male sex and the frequencies of fighting, being a victim of a threat or injury, alcohol and cigarette use, and being offered or sold an illicit drug while on school property accounted for 21% of the variance in the frequency of weapon carrying in school. When weapon carrying in school was dichotomized and analyzed with logistic regression, a model containing age, male sex, lower academic achievement, days not attending school because of fear, times threatened or injured with a weapon at school, frequency of fighting at school, cigarette and alcohol use, and being offered or sold illicit drugs on school property correctly classified 91.8% of students who did and did not carry weapons on school property. Less is known concerning the characteristics of youth who carry weapons to middle schools. In a sample of 1247 seventh and eighth grade students from 36 schools in Illinois, multiple logistic regression analysis found that bringing something to school for protection was associated with male sex, not living with both parents, not feeling close to parents, drinking heavily, participating in fights, damaging school property, and perceiving that at least a few other students brought weapons to school.13 Unlike DuRant et al,7 victimization and fear for safety in school were not significantly associated with weapon carrying.13 In contrast to Bailey et al,13 the 1995 North Carolina YRBS of middle school students had separate questions for carrying a gun and carrying a knife or club on school property. After adjusting for sex and race or ethnicity, the more dangerous of the 2 behaviors, carrying a gun, was associated with frequency of cigarette, alcohol, marijuana, and cocaine use. Although variables such as family composition, academic status, fighting, and being threatened or injured with a weapon at school were significantly associated with gun carrying on school property during bivariate analyses, during multivariate analyses they did not remain significant after accounting for illicit substance use. When we further examined the relationship between substance use and carrying a gun, the younger the age at which these students first used cocaine, marijuana, alcohol, and cigarettes, the more likely they were to have also carried a gun to school.
While a knife or club are less lethal than guns, a higher percentage of North Carolina middle school students reported to have carried them to school. In contrast to carrying a gun at school, carrying other weapons was associated with having carried a gun in general, having been threatened or injured with a weapon on school property, having fought, having used tobacco and alcohol, and having planned a suicide. Similar to gun carrying, the younger the age at which these students first smoked cigarettes, drank alcohol, or smoked marijuana, the more likely they were also to have carried a knife or club to school.
There are several limitations of this study that need to be considered when interpreting the findings. First, the YRBS is a cross-sectional survey of self-reported data and does not contain direct measures of behaviors. Some question the reliability of self-reported behaviors by adolescents. Because the YRBS is anonymous and not administered in the adolescent's home, the likelihood of reliable reporting of information is increased. This is supported by the high 2-week test-retest reliabilities found with the YRBS.25 In addition, the cross-sectional design of the YRBS prevents any assumptions of cause and effect between the independent and dependent variables. The correlation and logistic regression analyses should only be viewed as covariational relationships. Finally, statistical software, such as the Statistical Package for the Social Sciences (SPSS Inc, Chicago, Ill) and the Statistical Analysis System (SAS Institute, Cary, NC) assume that the data are collected with simple random sample designs. The YRBS uses a complex survey design. The within-school clustering of behaviors could result in small artificial reductions in the P values.7,22 To correct for the violation of this assumption, we tested our hypotheses at the .001 probability level instead of the traditional .05 level. While this is an overly conservative correction of the P values, we had sufficient power to find very small bivariate associations significant at the P<.001 level. Thus, the likelihood of committing a type I error was very small. The use of other statistical software, such as SUDAAN (Research Triangle Institute, Research Triangle Park, NC) or Westvar (Westat Inc), each of which uses different methods from the Statistical Package for the Social Sciences to compute error terms, would not have changed the results of the study.7
No single method has been found to prevent the use of violence by adolescents. Jessor26 argued that owing to the clustering of health risk and problem behaviors, interventions should target multiple risk behaviors and address multiple risk factors and protective domains. Findings from the present study have implications for violence prevention efforts. While DuRant et al11 have previously recommended that skills-building violence prevention and conflict resolution programs be added to health education curricula for middle schools, our data suggest that similar prevention programs need to be introduced in elementary schools. In addition, substance use prevention should be coupled with violence prevention programs in both elementary and middle schools. Pediatricians and other child and adolescent health care providers have a high level of credibility with educators and parents of school-aged children, and therefore should encourage school officials to add effective violence prevention and substance use prevention programs to health education curricula.11,27
From a clinical perspective, since the frequency and severity of health risk behaviors increases with age, screening during early adolescence is vital. Health care providers should also be aware that patients who report early experimentation with tobacco, alcohol, marijuana, and cocaine are at greater risk for engaging in dangerous behaviors, such as carrying guns or other lethal weapons to school. Therefore, in addition to determining if a patient has experimented with or is using tobacco and alcohol and other drugs, the health care provider should also determine (1) if the adolescent has access to a gun or other lethal weapons; (2) if he or she carries weapons for protection both in and out of school; (3) the frequency that the patient has been involved in fights in the past; (4) the exposure to violence and victimization; and (5) the degree to which the patient fears victimization at school or in traveling to and from school. The earlier that high-risk children and adolescents can be identified and interventions provided, the more likely that injury or death from violence can be prevented.
Accepted for publication July 1, 1998.
Presented at the Society for Adolescent Medicine Scientific Meeting, Atlanta, Ga, March 5, 1998.
Corresponding author: Robert H. DuRant, PhD, Department of Pediatrics, Wake Forest University School of Medicine, Medical Center Boulevard, Winston-Salem, NC 25157-1081 (e-mail: email@example.com).
Editor's Note: I think school-based prevention programs targeting substance abuse and violence prevention should begin in nursery school and sooner, in the home.—Catherine D. DeAngelis, MD
DuRant RH, Krowchuk DP, Kreiter S, Sinal SH, Woods CR. Weapon Carrying on School Property Among Middle School Students. Arch Pediatr Adolesc Med. 1999;153(1):21-26. doi:10.1001/archpedi.153.1.21