Age-Within-School-Class and Adolescent Gun-carrying
D. Neil Hayes, MD, MPH*, and David Hemenway, PhD‡
ABSTRACT. Background. Intentional injuries (sui-cide and homi(sui-cide) are a leading causes of morbidity and mortality in the United States. Firearms cause ;70% of these fatal intentional injuries. Risk factors associated with gun-carrying in adolescent populations include male gender, smoking, alcohol use, drug use, and num-ber of sexual partners. Current knowledge of these and other risk factors has provided limited benefit because many are no more obvious to the clinician a priori than is the tendency to carry guns. Increasing relative age of a student within school class is an easily measured param-eter that has been associated with behavioral problems, absenteeism, negative self-image, and high dropout rates.
Objective. To characterize the association between relative student age-within-class and tendency to carry firearms.
Design. The Massachusetts Youth Risk Behavior Sur-vey, which collects data on demographic characteristics, risk behaviors, and health outcomes.
Participants. A randomly selected group of 3153 Massachusetts students in grades 9 through 11.
Primary Outcome Measure. The odds of firearms-car-rying comparing older to average-age and younger stu-dents.
Results. Using multivariate logistic regression, seven risk factors predicted gun-carrying with statistically sig-nificant results: older age-within-class (OR: 2.12; 95% CI: 1.09 – 4.12), male gender (OR: 4.95; 95% CI: 3.01– 8.15), black race (OR: 2.49; 95% CI: 1.20 –5.14), gang member-ship (OR: 7.22; 95% CI: 4.51–11.56), missing school out of concern for safety (OR: 2.50; 95% CI: 1.30 – 4.80), seeking medical treatment after a fight (OR: 4.47; 95% CI: 2.56 – 7.78), and fighting without seeking medical treatment (OR: 5.73; 95% CI: 3.09 –10.60).
Conclusion. Older 9th-, 10th-, and 11th-grade stu-dents are more likely than their classmates to carry fire-arms. This information may prove helpful in identifying high-risk students and targeting prevention strategies.
Pediatrics 1999;103(5). URL: http://www.pediatrics.org/ cgi/content/full/103/5/e64; adolescent, weapon(s), fire-arm(s), gun(s), age, grade, race, gender, gang, fighting, safety, truancy.
ABBREVIATIONS. MYRBS, Massachusetts Youth Risk Behavior Survey; CDC, Centers for Disease Control and Prevention.
I
ntentional injuries rank among the leading causes of morbidity and mortality in the United States.1 For each year in the 1990s, there were ;55 000 fatal intentional injuries (;25 000 homicides and 30 000 suicides). Approximately 70% of these fatal intentional injuries were caused by a firearm. Be-tween 1985 and 1995, intentional deaths involved children and adolescents in higher numbers as both victims and perpetrators.2– 6The peak age for firearm homicide offenders is age 17 years, whereas the peak range for victims is 16 to 19 years.3Studies indicate that the presence of guns in the home increases the risk of adolescent suicide and homicide.2,4,7–9 However, most youth homicides oc-cur outside the home, and the vast majority are caused by gunshot wounds. Many studies have char-acterized youth who carry firearms.2,4,5,7,9 –25 Most show that 10% to 20% of adolescents (age 9 to 19 years) have carried a gun in the recent past; almost all this gun-carrying is illegal. The high prevalence of gun-carrying is not limited to poor or urban commu-nities. White suburban adolescent populations have high rates of gun-carrying, which increased .130% over the course of a decade.21,22,26
Risk factors associated with gun-carrying include male gender,11,19,22,23 smoking,15 alcohol use,15,17,19,23 drug use,11,17,21,25 multiple sex partners,23 poor aca-demic performance,15,23 criminal activity,11 fight-ing,14,19,24 lack of confidence staying out of fights,15,21 arrests,14fear,17,18,20,21threatened with a gun,11 shoot-ing in neighborhood,15friend or relative who carries a gun,11no discussions with parent about guns,15and not feeling close to parents.19
Knowledge of these risk factors has provided lim-ited benefit in the clinical setting despite a growing body of work showing the effectiveness of violence-prevention strategies.27–30Many of the risk factors are no more obvious to the clinician a priori than is the tendency to carry guns. For example, the clinician may have no knowledge of the adolescent’s sexual or drug habits.13 In fact, of the risk factors cited, male gender is the only objectively measurable character-istic that can be assessed easily and that has proved consistently to be associated with gun-carrying. More objective and readily accessible characteristics are needed to target high-risk groups to help clini-cians engage in prevention strategies. Although di-rect questioning regarding gun-carrying may be the most efficient means of assessing the adolescent’s behavior, issues such as patient trust or time con-straints increase the need for alternate means of eval-uation.
The parameters of age and grade are easily mea-From the *Department of Internal Medicine, Boston University School of
Medicine; and the ‡Department of Health Policy and Management, Har-vard School of Public Health, Boston, Massachusetts.
Received for publication Jul 9, 1998; accepted Oct 19, 1998.
Address correspondence to David Hemenway, PhD, Harvard School of Public Health, 677 Huntington Ave, Boston, MA 02115.
sured and have been evaluated separately regarding their association with gun-carrying, with inconclu-sive results. Some populations showed an increase in weapon-carrying with increasing grade15; in others, decreases in weapon-carrying with grade were not-ed.16,23The discrepancies could be attributed in part to differences in the populations under study and, for school-based studies, the tendency for high-risk students to drop out as they age.31,32 Researchers have not examined age and grade together when describing gun-carrying.
In the field of education, researchers use median age for grade as an index to isolate the effect of age from that of grade.33–39 Increasing relative age of a student within the class has been associated with increasing behavioral problems,36 absenteeism, neg-ative self-image,34and increased rate of dropout.37In contrast, the younger students in a grade have a lower rate of problem behavior and better self-im-age.40 Class age range stems from birth date within the calendar year, parental decision to delay the start of school, illness, grade retention (“failing a grade”), and other factors. Older students within a class are more likely to be nonwhite than white, male than female,39,41 and immigrant than native-born.36 Be-cause older students are more likely to have failed a grade than are average-age students, it is sometimes assumed that increasing age within a class is a proxy for poor academic performance.36,37Recent research has shown that older students within a class are at risk for dropout, absenteeism, and behavior prob-lems, independent of academic achievement.36,37 Poor self-esteem and social isolation from peers have been suggested as potential causes.36,37
This article uses data from the Massachusetts Youth Risk Behavior Survey Risk Behavior Survey (MYRBS) to characterize relative age-within-class as a novel risk factor for gun-carrying. We examined risk factors for being a year above the median age in class. We also re-examine previously established or suspected risk factors for gun-carrying, including gender, race, fear, fighting, and deviant behavior, to provide a clinical screening assessment of a student’s risk for firearm-carrying.
METHODS
Study Base and Sampling Technique
The MYRBS is described in detail elsewhere.16,42– 45Briefly, the
MYRBS was administered to a randomly selected sample of Mas-sachusetts public high school students in grades 9 through 12 from March to June 1995. A multistage clustering sample design was used to select a random cohort of schools and students. In the first stage, 63 of 299 eligible schools were randomly selected. Fifty-nine (94%) agreed to participate. Reasons for nonparticipation included lack of time and concurrent participation in other research projects. In the second stage, 5370 students were selected for participation (3 to 5 classrooms/school). Of students selected, 4166 surveys were collected, a student response rate of 78%, which reflects student attendance on the day of survey administration. Overall response rate for the 1995 MYRBS was 73%, the product of 78% of students responding multiplied by the 94% of schools that agreed to participate (0.9430.7850.73). Parents were informed of their child’s selection for the survey and asked to inform the school if they wished that the student be exempted from partici-pation. Nine parents refused.
Survey Instrument
The survey consisted of 91 questions on demographic charac-teristics, risk behaviors, and health outcomes. For this analysis, information from 10 questions was used: 1) gender (male/female); 2) race (six categories); 3) grade (9 through 12 or ungraded for special education or technical programs); 4) type of community (suburban, rural, urban); 5) age in years (12 through 17, older than 18); 6) gang membership (yes/no/blank response); 7) number of absences from school in last 30 days because of concern for per-sonal safety (five categories); 8) number of times treated by doctor or nurse after fight in last 12 months (five categories); 9) number of fights in last 12 months (eight categories); and 10) whether the student had carried a gun in the 30 days preceding the survey.
Exclusion Criteria and Model-building
For the analysis, the following transformations to the data were made. Students reporting their race as Native-American were consolidated into the “other” category because of their small number. To eliminate colinearity between the covariants mea-sured by questions 8 and 9, the number of times a student had been treated by a nurse or physician after fighting, was subtracted from the number of fights in the last year. The median age in years was calculated for grades 9 through 12 (9th515, 10th516, 11th5
17, 12th 5 18). Because the MRYBS collected information on student age to the nearest year only, subjects were assigned to three categories based on relationship to the median age: 1 year below median, median age for class, or 1 year older than median. The MYRBS characterized all students older than age 17 as “eigh-teen years old or older,” thus, it was not possible to distinguish 12th-grade students who were older than the mean from those who were at the mean age for the class. Because the effect of age distribution within a class was the primary exposure of interest, 12th-grade students were excluded from the analysis (n5924). Of the remaining subjects, 99% fell within 1 year of the median age. Of the 1% who did not, the range of relative age difference was from 6 years younger to 3 years older than the median. Because the vast majority of students fell in the range of 1 year older to 1 year younger than the median, it was decided to examine the effect of age as 2 binary variables, each corresponding to 1-year relative age changes from the median. Students$2 years above the median (n536) and#2 years below the median (n511) were excluded from the analysis. Alternate models (results not shown) that included these 47 subjects did not significantly changes the results of the analysis. Students responding “ungraded” to ques-tion 3 (n56) and students who did not volunteer information on grade or age (n536) also were excluded. Because of a relatively large number of missing data for the question addressing gang membership (n5267), those with blank responses were retained in a “missing” category for the final analysis. Exclusion of the blank responses did not alter the significance of effect estimates. All ordinal variables (questions 3, 5, 7, 8, and 9) were modeled as linear, categoric (where appropriate), and binary terms in the logistic regression without any change in significance of the study results.
The Woolf approximation of 95% CI was used for tabular analysis. Clustering of responses within schools and weighing of the data (using a weighing term provided by the Centers for Disease Control and Prevention (CDC) reflecting the likelihood of having been sampled and a factor compensating for different patterns of nonresponse)16were accounted for using survey
logis-tic regression software. Logislogis-tic regression not accounting for clustering or weighing factors did not produce significantly dif-ferent results.
RESULTS
personal safety, gang membership, and seeking med-ical care from a physician or nurse all were associ-ated with a doubling of the odds of being older than other students within the same grade. Tenth and 11th-graders were no more likely to be above the median age than were 9th-graders. Having fought in the last 12 months was not associated with an in-crease in the odds of older-student status.
In the multivariate logistic analysis, only nonwhite race and living in an urban community were inde-pendent predictors of significantly elevated relative risk for increased age within a class. Male gender, gun-carrying, and missing school attained marginal significance. Gang membership and rural living were not independent predictors of an increased age within a class.
In the bivariate analyses, males had a relative risk eight times that of females for having carried a gun in the last 30 days (Table 2). Compared with whites, both blacks (OR: 2.48; 95% CI: 1.61–3.82) and Hispan-ics (0R: 1.74; 95% CI: 1.07–2.85) were more likely to have carried a gun. It should be noted, however, that because relatively few of the subjects were black or Hispanic, most of the guns were carried by white students. Eleventh-grade students were less likely to
carry guns than were 9th-graders (OR: 0.57; 95% CI: 0.39 – 0.84), with 10th-graders having intermediate but statistically insignificant odds of gun-carrying (OR: 0.87; 95% CI: 0.61–1.25). The odds that an urban student carried a gun in the last 30 days were 75% greater than for suburban or rural students (OR: 1.75; 95% CI: 1.23–2.49). Relative age-within-grade also was predictive of gun-carrying, with the median-age students more than twice as likely to have carried a gun as the youngest students (OR: 2.33; 95% CI: 1.54 –3.54). The eldest students within a class were the most likely to have carried a gun in the last 30 days (OR: 4.70; 95% CI: 2.90 –7.61). Gang member-ship was the single best predictor of gun-carrying (OR: 15; 95% CI: 10.50 –21.44). Again, because rela-tively few students admitted to being members of a gang, the majority of all gun-carrying was done by nongang members. Having missed school out of con-cern for personal safety (OR: 6.02; 95% CI: 4.00 –9.06), having sought medical care related to a fight (OR: 11.45; 95% CI: 7.56 –17.35), and having been in a fight without seeking treatment (OR: 11.76; 95% CI: 7.37– 18.75) all were strongly associated with increased odds of gun-carrying.
In multivariate logistic regression, seven risk fac-TABLE 1. Distribution of Students Above Median Class Age From Population of 3153 Massachusetts High School Students (Grades 9 –11) by Selected Demographic Characteristics
Demographic Characteristic
N* (All Eligible
Cases)
Percentage of Students
Above Median Class Age
Crude OR 95% CI
(Woolf)
Adjusted OR† Adjusted 95% CI
Total 3153 12.2
Sex
Female 1567 8.4 1.00 — 1.00 —
Male 1583 16.0 2.07 1.65–2.59 1.60 0.97–2.66
Race
White 2230 8.7 1.00 — 1.00 —
Black 281 23.1 3.17 2.32–4.34 1.79 1.15–2.78
Hispanic 267 19.1 2.49 1.78–3.50 1.98 1.17–3.34
Asian 163 22.1 2.99 2.01–4.46 2.68 1.44–4.98
Other 198 18.7 2.43 1.65–3.57 1.93 1.17–3.20
Grade
Grade 9 1009 12.9 1.00 — 1.00 —
Grade 10 1028 11.3 0.87 0.67–1.13 0.89 0.60–1.34
Grade 11 1116 12.4 0.95 0.74–1.23 1.03 0.67–1.58
Kind of community
Suburban 1125 6.4 1.00 — 1.00 —
Urban 1586 17.0 3.00 2.29–3.94 2.15 1.42–3.27
Rural 442 9.7 1.58 1.06–2.34 1.78 0.90–3.50
Carried gun last 30 d
No 2936 11.2 1.00 — 1.00 —
Yes 171 25.1 2.64 1.83–3.80 1.60 0.97–2.66
Member of a gang
No 2649 11.6 1.00 — 1.00 —
Yes 237 20.7 1.98 1.42–2.78 1.18 0.77–1.83
Blank 267 10.9 0.92 0.62–1.39 0.75 0.36–1.57
Missed school last 30 d out of fear for safety
No 2964 11.5 1.00 — 1.00 —
Yes 176 23.2 2.33 1.61–3.36 1.58 1.00–2.49
Treated by MD or nurse after fighting in last 12 mo
No 2958 11.5 1.00 — 1.00 —
Yes 131 19.8 1.91 1.22–2.97 1.17 0.69–2.02
Fought without seeking treatment for injuries
No 1891 11.1 1.00 — 1.00 —
Yes 1155 12.6 1.15 0.92–1.44 0.96 0.70–1.31
* Totals do not equal 3153 in all demographic categories because of missing data.
tors for gun-carrying remained significant: male gen-der (OR: 4.95; 95% CI: 3.01– 8.15), black race (OR: 2.49; 95% CI: 1.20 –5.14), older age within class (OR: 2.12; 95% CI: 1.09 – 4.12), gang membership (OR: 7.22; 95% CI: 4.51–11.56), missing school out of concern for safety (OR: 2.50; 95% CI: 1.30 – 4.80), seeking medical treatment after a fight (OR: 4.47; 95% CI: 2.56 –7.78), and fighting without seeking medical treatment (OR: 5.73; 95% CI: 3.09 –10.60). Median class age was mar-ginally significant for an increase in odds for gun-carrying (OR: 1.64; 95% CI: 0.98 –2.74), as was urban community (OR: 1.68; 95% CI: 0.98 –2.89).
DISCUSSION
To date, knowledge about adolescent firearms-car-rying behaviors has received limited clinical atten-tion. Risk factors such as tendency toward illicit ac-tivities, substance abuse, or sexual promiscuity are not clinically obvious and often difficult to deter-mine. Objective determinants of risk such as age or grade are associated inconsistently with increasing and decreasing risk. This analysis focuses on easily obtainable characteristics from clinical history that can provide a useful assessment of the odds that an
adolescent carries a firearm. It should be noted, how-ever, that although age-within-school class may prove a useful objective risk factor for weapon-car-rying, a careful psychological assessment still is re-quired for patient evaluation.
The primary outcome of interest was the effect of age and grade on weapon-carrying. The finding that older students within a class carried weapons with greater frequency is consistent with research show-ing associations between increasshow-ing relative age and other behavioral, self-esteem, and school perfor-mance problems. A linear trend is apparent with the youngest students having the lowest odds, average-age students having intermediate odds, and oldest students at greatest odds for gun-carrying. At the same time, increasing grade tended to decrease the odds of having carried a gun, although effect esti-mates were statistically significant only in the biva-riate analysis comparing 9th-graders with 11th-grad-ers. One plausible explanation for a decrease in firearm-carrying with increasing grade is that stu-dents carrying guns in the lower grades drop out before advancement to higher grades.
As in results of other studies, male gender was TABLE 2. Distribution of Gun-carrying Behavior by Known or Suspected Risk Factors
Known or Suspected Risk Factor
n* Percent
Carrying Gun in Last 30 Days
Crude OR for Carrying
Gun
95% CI (Woolf)
Adjusted OR† Adjusted 95% CI
Total 3107 5.5
Sex
Female 1551 1.4 1.00 — 1.00 —
Male 1553 9.7 7.79 4.91–12.37 4.95 3.01–8.15
Race
White 2208 4.7 1.00 — 1.00 —
Black 266 10.9 2.48 1.61–3.82 2.49 1.20–5.14
Hispanic 264 8.0 1.74 1.07–2.85 0.96 0.44–2.10
Asian 163 4.3 0.91 0.42–1.99 0.89 0.31–2.54
Other 192 5.2 1.11 0.57–2.16 0.88 0.41–1.92
Grade
Grade 9 991 6.8 1.00 — 1.00 —
Grade 10 1010 5.9 0.87 0.61–1.25 0.69 0.40–1.21
Grade 11 1106 4.0 0.57 0.39–0.84 0.66 0.39–1.11
Kind of community
Suburban 1118 4.1 1.00 — 1.00 —
Urban 1548 7.0 1.75 1.23–2.49 1.68 0.98–2.89
Rural 441 3.9 0.93 0.53–1.65 1.05 0.45–2.43
Relative age of student
Younger than median 1115 2.7 1.00 — 1.00 —
Median age for class 1618 6.1 2.33 1.54–3.54 1.64 0.98–2.74
Older than median 374 11.5 4.70 2.90–7.61 2.12 1.09–4.12
Member of a gang
No 2628 3.1 1.00 — 1.00 —
Yes 221 32.6 15.00 10.50–21.44 7.22 4.51–11.56
Blank 258 6.6 2.19 1.28–3.75 1.27 0.59–2.72
Missed school last 30 d out of fear for safety
No 2933 4.4 1.00 — 1.00 —
Yes 165 21.8 6.02 4.00–9.06 2.50 1.30–4.80
Treated by MD or nurse after fighting in last 12 mo
No 2928 4.0 1.00 — 1.00 —
Yes 127 37 11.45 7.56–17.35 4.47 2.56–7.78
Fought without seeking treatment for injuries
No 1879 1.1 1.00 — 1.00 —
Yes 1134 11.7 11.76 7.37–18.75 5.73 3.09–10.60
* Complete case analysis (n53107) includes 3153 students from Table 1 minus 46 students with missing data on gun-carrying in the last 30 days. Totals do not equal 3107 in all categories because of missing data.
strongly associated with firearm-carrying, as was black race and living in an urban environment. It is unlikely that these students were carrying guns for hunting or target practice. Gang membership was the single best predictor of gun-carrying; it also was the question most likely to be left blank. The large number of nonresponses may be attributable not only to student concerns for anonymity, but also to uncertainty about whether they are actually mem-bers of a gang. Students who missed school out of concern for safety were more likely to carry guns, consistent with considerable research linking fear to weapon-carrying.15,17,18,20,21Finally, fighting was asso-ciated with a dramatic increase in the odds of carry-ing a gun. Among students who had not fought in the last year, only 1% carried a firearm in the last 30 days. Among those who sought medical attention after a fight, 37% had carried a gun. This situation seems opportune for a discussion with the patient about firearms.
Effect estimates from the analysis are presented in two forms, as crude ORs and ORs obtained from multivariate regression models. Both the crude and adjusted OR have specific interpretation for those interested in the students who carry guns. Practitio-ners who encounter patients in circumstances such as in the emergency department often are limited in their ability to fully assess relevant risk factors, es-pecially those drawn from sensitive aspects of the patient’s history. A physician treating a child for injuries sustained during a fight may know little else about the patient. Under these circumstances, the crude OR may be the most useful assessment of the child’s risk for carrying a firearm. However, the crude OR is inaccurate in assessing the causality of each risk factor on weapon-carrying, but it is more accurate in assessing overall risk. When the oppor-tunity to define the adolescent’s risk profile presents itself more fully, the adjusted ORs should prove more accurate. Interventions then can be focused with attention to the potential causal role attributed to individual components making up total risk.
Although the MYRBS uses a large population-based dataset, there remains the potential for biased results. Students absent on the date the survey are not a random subset of the overall population. Ad-olescents who do not attend school typically have higher rates of risky behavior than those who do.31 Another problem is that the MYRBS relies entirely on self-reports. Although self-reports of criminal activ-ity are considered reliable,46 informal analyses also were performed to evaluate the possibility that de-liberate misrepresentation contributed greatly to the results. Analyses included eliminating respondents who reported engaging in all or almost all risky behaviors listed in the survey, engaged in the highest levels of the risky behaviors, and reported inconsis-tent results (for example, where the number of times seeking medical treatment after a fight was greater than the number of total times fought). Inconsistent responses and reports of extremely high risk-taking behavior made up,0.5% of all surveys included in the analysis; their elimination did not change the results appreciably.
Age is only measured to the nearest year; a student age 17 might be from 1 day to 730 days older than a student age 16. This type of error in measuring stu-dents age is a random error that depends only on the student’s birthday and not on any characteristic of the survey. Measurement errors in the exposure vari-able (such as relative age in this case) have been shown to weaken the association of interest, as long as the measurement errors are random.47 A better measurement of age might demonstrate a larger OR for the association.
Uncontrolled confounders pose an additional con-cern in the interpretation of the study results. Lim-ited data were available to describe socioeconomic, demographic, and academic performance. Socioeco-nomic factors likely play a role in such associations as that seen between black race and weapon-carry-ing. The association between race and homicide rates sometime disappears when factors related to poverty are included in the analysis.48The focus of this article was on known or suspected risk factors for weapon-carrying. It is possible that additional confounding by such variables as alcohol or drug use or other risk-taking behavior might account for the relation-ship between relative age and weapon-carrying.
The risk factors presented provide information about the association, not causality. For example, it is possible that students who are older than their class-mates feel isolated and have low self-esteem, which lead to gun-carrying. It also is possible that carrying handguns creates a sense of isolation, which subse-quently causes the student to fall behind in school. Similar statements could be made for most of the associations noted in the analysis. Despite the statis-tically significant relationship, relative age alone is a limited screening tool for weapon-carrying (sensitiv-ity, 25%; specific(sensitiv-ity, 89%). In other words, only 25% of students older than the median age reported car-rying a weapon in the last 30 days. Of students at the median age or younger, 89% denied carrying a weapon in the last 30 days. Predictive values, which are functions of prevalence, illustrate that relative class age has different interpretation depending on the population of adolescents. When applied to the overall population of survey respondents, the pro-portion of students older than the class median that carried weapons was 12% (positive predictive value). In the same group of students, 95% of those at the median age or younger did not carry weapons (neg-ative predictive value). When applied to the popula-tion of students who received medical treatment af-ter fighting, the proportion of students older than the class median that carried weapons was 57% (positive predictive value). Among students receiving medical care for fighting related injuries, 67% of those at the median age or younger did not carry weapons (neg-ative predictive value).
informa-tion can be useful to clinicians seeking to help high-risk adolescents.
REFERENCES
1. Compressed Mortality Files. National Center for Health Statistics. Centers for Disease Control and Prevention. http://www.wonder.cdc.gov/. 1998
2. Clarke RV, Jones PR. Suicide and increased availability of handguns in the United States.Soc Sci Med.1989;28:805– 809
3. Rivara FP, Farrington DP. Prevention of violence.Arch Pediatr Adolesc Med.1995;149:421– 429
4. Brent DA, Perper JA, Moritz G, Baugher M, Schweers J, Roth C. Fire-arms and adolescent suicide.Am J Dis Child.1993;147:1066 –1071 5. Kellermann AL. Gunsmoke, changing public attitudes toward smoking
and firearms.Am J Public Health.1997;87:910 –913
6. Hung K.Firearms Statistics. Research and Statistics Division, Depart-ment of Justice, Canada. October 1997
7. Kellermann AL, Rivara FP, Somes G, et al. Suicide in the home in relation to gun ownership.N Engl J Med.1992;327:467– 472
8. Rivara FP, Grossman DC, Cummings P. Injury prevention. N Engl J Med.1997;337:543–547
9. Sloan J, Kellermann AL. Handgun regulations, crime, assaults, and homicide.N Engl J Med.1988;319:1256 –1262
10. Ginsberg C. Violence-related attitudes and behaviors of high school students, New York City 1992.MMWR.1993;43:773–777
11. Sheley JF, Brewer VE. Possession and carrying of firearms among suburban youth.Public Health Rep.1995;110:18 –26
12. Orgog GJ, Wasserberger J, Schatz I, et al. Gunshot wounds in children under 10 years of age.Am J Dis Child.1988;142:618 – 622
13. Sheley JF, McGee ZT, Wright JD. Gun related violence in and around inner-city schools.Am J Dis Child.1992;146:677– 682
14. Webster DW, Gainer PS, Champion HR. Weapons carrying among inner-city junior high school students: defensive behavior vs. aggressive delinquency.Am J Public Health.1993;83:1604 –1608
15. Hemenway D, Prothrow-Stith D, Bergstein JM, Ander R, Kennedy BP. Gun carrying among adolescents.Law Contemp Problems.1996;59:39 –54 16. Kessel SM.Massachusetts Youth Risk Behavior Survey Results. Malden,
MA: Massachusetts Department of Education; 1996
17. Presley CA, Meilman PW, Cashin JR. Weapon carrying and substance abuse among college students.J Am Coll Health.1997;46:3– 8 18. Arria A, Borges G, Anthony JC. Fears and other suspected risk factors
for carrying lethal weapons among urban youths of middle-school age. Arch Pediatr Adolesc Med.1997;151:555–560
19. Bailey SL, Flewelling RL, Rosenbaum DP. Characteristics of students who bring weapons to school.J Adolesc Health.1997;20:261–270 20. Martin SL, Sadowski LS, Cotton NU, McCarraher DR. Response of
African-American adolescents in North Carolina to gun carrying by their school mates.J School Health.1996;66:23–26
21. Kingery PM, Pruitt BE, Heuberger G. Profile of rural Texas adolescents who carry handguns to school.J School Health.1996;66:18 –22 22. Arria AM, Wood NP, Anthony JC. Prevalence of carrying a weapon and
other related behaviors in urban schoolchildren.Arch Pediatr Adolesc Med.1995;149:1345–1350
23. Orpinas PK, Basen-Engquist K, Grunbaum JA, Parcel GS. The co-morbidity of violence-related risk behaviors in a population of high school students.J Adolesc Health.1995;16:216 –225
24. Sosin DM, Koepsell TD, Rivara FP, Mercy JA. Fighting as a marker for multiple problem behaviors in adolescents.J Adolesc Health.1995;16: 209 –215
25. Sheley JF. Drugs and guns among inner-city high school students.J
Drug Educ.1994;24:303–321
26. Deaths resulting for firearm- and motor-vehicle-related injuries— United States, 1968 –1991.MMWR.1994;43:37– 42
27. Irwin C, Millstein S. Biopsychosocial correlates of risk-taking behaviors during adolescence: can the physician intervene?J Adolesc Health Care. 1986;7(suppl 6):82–96
28. Grossman D, Neckerman H, Koepsell T, et al. Effectiveness of a violence prevention curriculum among children in elementary school.JAMA. 1997;277:1605–1611
29. DeVos E, Stone D, Goetz M, et al. Evaluation of a hospital-based youth violence intervention.Am J Prev Med.1996;12:101–108
30. Powell K. Prevention of youth violence: rationale and characteristics of 15 evaluation projects.Am J Prev Med.1996;12(suppl 2):3–12 31. Health risk behaviors among adolescents who do and do not attend
school, United States, 1992.MMWR.1994;43:37– 42
32. Massachusetts Drop Out Rates. Malden, MA: Massachusetts Department of Education. http://www.doe.mass.edu/doedata
33. Bickel DD, Zigmond N, Strayhorn J. Chronologic age at entrance to first grade: effects on elementary school success.Early Childhood Res Q. 1991;6:105–117
34. Walker E. Changing self esteem: the impact of self esteem changes on at-risk student achievement. Newark, NJ: Newark Board of Education Office of Research, Evaluation, and Testing; 1991
35. Williams J. Students at risk: a comprehensive approach to dropout prevention. Dayton, OH: Dayton City Schools; 1991
36. Roderick M. Grade retention and school dropout: policy debate and research questions.Research Bulletin. Bloomington, IN: Phi Delta Kappa Center for Education, Development, and Research. 1995;15:3– 8 37. Anderson VH.The Overage Student. Candidate for School Failure.Portland,
OR: Portland State University, 1990. Dissertation
38. May DC, Kundert DK, Brent D. Does delayed school entry reduce later grade retention and use of special education services?Remedial Special Educ.1995;16:228 –294
39. McArthur EK, Bianchi SM.Characteristics of Children Who are “Behind” in School. Final Report. Washington, DC: Bureau of the Census, National Center for Education Statistics; 1993
40. Zafirau JS.A Study of the Flow-through in the Cleveland City Schools: 1985–1989. Cleveland, OH: Cleveland Public Schools; 1990. Report 41. Sayler MF, Brookshire WK. Social, emotional, and behavioral
adjust-ment of accelerated students, students in gifted classes, and regular students in eighth grade.Gifted Child Q.1993;37:150 –154
42. Brener N, Collins J, Kann L, et al. Reliability of the Youth Risk Behavior Survey Questionnaire.Am J Epidemiol.1995;141:575–580
43. DuRant R, Kahn J, Hayden P, et al. The association of weapon carrying and fighting on school property and other health risk and problem behaviors among high school students.Arch Pediatr Adolesc Med.1997; 151:360 –366
44. Middleman A, Faulkner A, Woods E, et al. High-risk behaviors among high school students in Massachusetts who use anabolic steroids. Pedi-atrics.1995;96:268 –272
45. Shrier L, Emans S, Woods E, et al. The association of sexual risk behaviors and problem drug behaviors in high school students.J Adolesc Health.1996;20:377–383
46. O’Brien R.Crime and Victimization. Beverly Hills, CA: Sage Publications Inc; 1985
47. Wacholders S. When measurement errors correlate with truth: surpris-ing effects of nondifferential misclassification. Epidemiology. 1995;6: 157–161
DOI: 10.1542/peds.103.5.e64
1999;103;e64
Pediatrics
D. Neil Hayes and David Hemenway
Age-Within-School-Class and Adolescent Gun-carrying
Services
Updated Information &
http://pediatrics.aappublications.org/content/103/5/e64
including high resolution figures, can be found at:
References
http://pediatrics.aappublications.org/content/103/5/e64#BIBL
This article cites 35 articles, 1 of which you can access for free at:
Subspecialty Collections
http://www.aappublications.org/cgi/collection/firearms_sub
Firearms
icine_sub
http://www.aappublications.org/cgi/collection/adolescent_health:med
Adolescent Health/Medicine
http://www.aappublications.org/cgi/collection/for_your_benefit
For Your Benefit following collection(s):
This article, along with others on similar topics, appears in the
Permissions & Licensing
http://www.aappublications.org/site/misc/Permissions.xhtml
in its entirety can be found online at:
Information about reproducing this article in parts (figures, tables) or
Reprints
http://www.aappublications.org/site/misc/reprints.xhtml
DOI: 10.1542/peds.103.5.e64
1999;103;e64
Pediatrics
D. Neil Hayes and David Hemenway
Age-Within-School-Class and Adolescent Gun-carrying
http://pediatrics.aappublications.org/content/103/5/e64
located on the World Wide Web at:
The online version of this article, along with updated information and services, is
by the American Academy of Pediatrics. All rights reserved. Print ISSN: 1073-0397.
