Skip to main content
. Author manuscript; available in PMC: 2020 Aug 1.
Published in final edited form as: J Behav Med. 2019 Aug 1;42(4):763–810. doi: 10.1007/s10865-019-00048-x

Table 3.

Articles reporting data on risk and protective factors for adolescent firearm carriage

Article Design Sample size Study sample and setting Unique attributes Carriage rates Results Limitations NOS score
Allen and Lo (2012) Cross-sectional
Secondary analysis of data from Sheley(1994) Study
835 juvenile inmates
695
students
Male inmates from 6 correctional facilities and 9th-12th grade students from 10 inner-city public high schools in CA, LA, IL, and NJ Ages 15–19
100% male
Students:
M age 17
73% Black
18% Hispanic Inmates:
M age = 17
46% Black
29% Hispanic
Rates of co-occurring drug trafficking and gun carriage:
Students = 16%
Inmates = 65%
Students: After controlling for age, absentee father, employment, h/o expulsion (OR 2.6), and code-based beliefs (OR 1.2) were positively associated with co-occurring drug dealing and gun carriage. Negative associations were Non-black (OR 0.10) and Hispanic (OR 0.47) race/ethnicity
Inmates: After controlling for age/employment, having an absentee father (OR 3.7), h/o violent behavior (OR 4.4), and code-based beliefs (OR 1.2) were positively associated with drug dealing/gun carriage. Negative associations were non-black (OR 0.1) and Hispanic (OR 0.2) race/ethnicity
Limited generalizability
All-male student and inmate samples from select cities in four states
Cross-sectional data
Self-report data
Questionnaires were not identical for samples.
Note: study sample is from 1991 (published 2012)
6
Apel and Burrow (2011) Cohort
2 waves of data from National Longitudinal Youth Survey
1524 Nationally representative sample of US youth with oversampling of Black/Hispanic youth M age 13 (w1)
52% male
15% Black
12% Hispanic
Lifetime carriage = 7.4% in w1
Past 1-year carriage = 5.8% in wave 2
In cross-sectional model (independent/dependent variables from 1997 data; exposure to violence (i.e., gunshots in neighborhood) was correlated with carriage (OR 2.3), bullying was not correlated after adjusting for demographic, family, school, peer affiliation, and other factors. In longitudinal model (independent variables 1997; dependent variables 1998), exposure to violence was correlated with carriage (OR 2.0), but bullying was not correlated after adjustment Self-report data
Limited to young adolescent sample
Note: study sample taken from 1997 (published 2011)
6
Arria et al. (1995) Cohort
Surveyed annually for 5 years
1714 Students from 43 classrooms within 19 public schools from 5 districts in MD M age (w1) 9
M age (w5) 13
50% Male
73% Black
26% White
Past 1-year carriage = 10% (Male)
1% (Female)
Rates of gun carriage increased across waves of data, starting at < 2% (w1) and increasing to 10% (male) and 1%(female) by w5 data collection; Less lethal weapon carriage (stick, knife) was associated with gun carrying at later waves across multiple waves Limited generalizability
Single city; Younger
Adolescent sample
Missed student drop-outs
Self-report data
6
Beardslee, et al. (2018a) Cohort
Pittsburgh Youth Study; Data is from 6-month surveys (3 years) and then 12-month surveys in ages 10–17
485 Random sub-sample of male youth from youngest cohort of large school-based longitudinal study in Pittsburgh, PA 100% male
56% Black
41% White
Excluded non-White/Black participants
Lifetime gun carriage = 27% (Black)
12% (White)
Youth with higher levels of conduct problems and peer delinquency at earlier childhood waves, as well as those with increases in conduct problems across early childhood waves, were more likely to initiate gun carriage before 18
Examining whether racial differences in carrying behavior were due to differential exposure to risk factors or differential sensitivity, study found more support for differential exposure model, with 60% of the race effect on carriage being attributable to either initial peer delinquency levels or initial levels of conduct problems
Limited generalizability
Limited to male
Black/White youth sample
Single city sample
Self-report data.
Small sample size may limit ability to detect group differences in predictors of carriage
6
Beardslee et al. (2018b) Cohort
Pathway to Desistance Study: Data collected every 6-month for 3 years then every 12-month for 4 years
1170 Male offenders recruited from court system in two counties in AZ (Maricopa) and PA (Philadelphia) Ages 14—19 at baseline
100% male
42% Black
34% Hispanic
19% White
70% on active probation at baseline
Past 6-month Carriage = 15%(w1); 15%(w3); 12%(w5); 12%(w6)
Past 12-month carriage 17% (w7); 10% (w10)
Among youth with recent offending, gun violence exposure (witness/victim of gun violence) were 43% more likely to engage in carriage at the next wave after controlling for time-stable and time-varying (exposure to peers who carried; exposure to peers engaged in other criminal acts, developmental changes, changes in gun carrying from incarceration or institutionalization) covariates
No evidence that non-gun violence exposure conferred same risk
Limited generalizability
Male sample of juvenile offenders in two states
Self-report data Youth offenders likely had fluctuations in carrying over time
6
Cao et al. (2008) Cross-sectional
Nat. Crime Victim. Survey (School Crime Supp. data)
7391 Youth in public or private school. Limited to Black and White (Hispanic and non-Hispanic) youth. Nat. representative sample Ages 12–17
M age 15
52% male
70% White
16% Hispanic
14% Black
91% public school
Past 6-month carriage to school = 0.5% Among adolescents attending public/private school, those carrying a gun for protection were more likely to report recent involvement in physical fighting (AOR 1.1); knowing a peer carrying guns (AOR 2.0), and not being female (AOR = 0.8). Factors not significant in the model included fear of being attacked, avoidance, substance use factors, gangs at school, truancy, security guards, age, race, parental education, rurality, and region Limited generalizability
Low incidence of carriage
Limited ethnic/racial groups
Cross-sectional data
Excluded carriage-related factors (e.g., sell drugs)
Fighting may have been aggression or victimization
6
Carter (2013) Cross-sectional
Flint Youth Injury Study
689 Assault-injured urban youth seeking ED treatment at a Level-1 trauma center in Flint, MI Ages 14–24
M age 20
50% Male
61% Black.
73% low SES
Past 6-month carriage = 10% Male gender (AOR 2.8), higher SES status (AOR 1.5), ilicit drug use (AOR 1.6), serious physical fighting (AOR 1.7), and attitudes favoring retaliatory violence (AOR 1.6) were associated with firearm possession. Age and race were not significant predictors of possession (carriage/ownership) Limited generalizability
Single urban ED sample
Predominantly Black youth
Cross-sectional data
Self-report data
Exclusions may lead to
underestimation of carriage
7
Connolly and Beaver (2015) Cohort
National Longitudinal Survey of Youth, surveyed annually for 8 years
1304 sibling pairs Sibling pairs nested within the nationally representative sample of youth; oversampled Black/Hispanic adolescents Ages 12–18 at start of study; 27 monozygotic (MZ) twins, 932 dizygotic (DZ) twin/full sibling pairs, 75 half-sibling (HS) pairs Lifetime carriage = 9.6%
(among the full sibling sample)
5% (MZ)
10% (DZ)
13% (HS)
Half-siblings reported higher rates of gun carrying than MZ twins, DZ twins, or full siblings. Within-sibling concordance was higher for MZ twins (p < 0.05) than for DZ twins/full siblings (p < 0.05), and half-siblings (p < 0.05). Additive genetic effects explained 27% of variation in gun carriage. Common genetic influences explained 66% of covariance between gun carrying and gang membership. Shared environmental factors didn’t explain variance across models, with non-shared environmental factors explaining remainder of variance not accounted for by genetic factors Limited generalizability
Sample of sibling pairs may not represent non-siblings
Self-report data
Note study sample taken from 1997 to 2005 (published 2015)
6
Cook and Ludwig (2004) Cross-sectional
1995 National Survey of Adolescent Males (NSAM)
1151 Male youth living in US households. Survey was of youth 15–19, but analytic sample restricted to under age 18 Ages 15–17
100% male
38% White
28% Black
31% Hispanic
46% Urban
34% Suburban
20% Rural
Past 30-day carriage = 10% Likelihood of gun carriage among youth was positively associated with rate of robbery (AOR 6.0) and prevalence of gun ownership (AOR 4.9) after controlling for individual and household characteristics
Of note, Black and Hispanic youth were more likely to carry guns than others, although for Hispanics-effect was limited to those in English-speaking homes. Gun carriage was not associated with age, grade, or household SES status
Cross-sectional data.
Self-report data
Used proxy variable (FS/S, suicides committed with guns) for measure of gun availability
7
Cunningham et al. (2010) Cross-sectional 2069 Consecutive sample of adolescents (age 14–18) presenting for any reason for ED treatment at a Level-1 trauma center in Flint, MI Ages 14–18.
45% Male
57% Black
53% low SES
40% seeking ED treatment for an injury
Past year carriage = 7% Gun carriage was associated with Black race (OR 2.4), Male sex (OR 2.4), failing grades (OR 1.5), Marijuana use (OR = 3.3), recent gun victimization (OR 1.9), recent physical fighting (OR = 1.6), group fighting (OR = 3.3), sexual activity (OR = 2.4). Gun carriage frequency was associated with older age (IRR = 1.3), male sex (IRR = 1.3), failing grades (IRR = 1.6), employed (IRR = 1.4), lower SES (IRR = 1.8), binge drinking (IRR = 1.3), fighting resulting in injury (IRR = 1.8), recent gun victimization (IRR = 1.2), serious fighting (IRR = 1.3), and group fighting (IRR = 1.6) Limited generalizability
Urban mostly Black sample
Cross-sectional data
Self-report data
Exclusion of suicidal and sexual assault patients may underestimate gun carriage
7
DuRant et al. (1999) Cross-sectional 2227 Randomly selected 6–8th grade students attending 53 randomly selected public middle schools in NC Ages 11–16
36% age 13
51% male
64% White
28% Black
2% Hispanic
Ever carried to school = 3% RF for carriage included Male sex (AOR 7.1), Minority ethnicity (AOR 3.3), Alcohol use (AOR 4.6), Cocaine use (AOR 3.0), Marijuana use (AOR 3.7), and Smoking frequency (AOR 1.3). Having ever carried a gun, having been threatened with weapon/in a fight, and suicidality were not predictive. Those who smoked daily were 8 times as likely to carry. Controlling for age/sex, earlier age of onset for substance use (i.e., cigarette, alcohol, and marijuana use) was associated with carriage to school Cross-sectional data
Self-report data
Sampling design doesn’t account for potential for clustering of behaviors within schools
7
Freed et al. (2001) Qualitative semi-structured Interviews 45 Male youth incarcerated in a MD juvenile justice facility Ages 14–18
M age 16
100% Male.
67% Black
61% Urban
Lifetime gun carriage = 60% Bivariate comparisons found that those who owned or carried a gun were more likely to be have sold drugs (89% vs 56%, p = 0.05), been victimized by someone with a weapon (81% vs 33%, p < 0.01), and to have lived in city (71% vs 39%, p< 0.05) Limited generalizability
Incarcerated male youth
Qualitative study
Small sample
Self-report data
6
Hayes and Hemenway (1999) Cross-sectional Youth Risk Behavior Survey (YRBS) in MA 3153 Random sample of 9–11th graders from MA public schools Age 15–18
50% Male
71% White
9% Black
8% Hispanic
50% Urban
14% Rural
Past 30-day carriage = 5% Risk factors associated with past 30-day gun carriage included: male gender (OR 5.0), Black race (OR 2.5), older age within class (OR 2.1), gang membership (OR 7.2), missing school out of concern for safety (OR 2.5), seeking medical treatment after a fight (OR 4.5), and fighting without seeking medical treatment (OR 5.7) Cross-sectional data
Self-report data
Misses students not in school
Excluded large % of original sample (entire 12th grade)
7
Hemenway et al. (1996) Cross-sectional
“Hands without guns” study
1192 7th (n = 752) and 10th (n = 440) graders from 12 public schools in Boston, MA and Milwaukee, WI 49% male Past 30-day carriage to school = 3%
Lifetime carriage = 17%
RFs associated with higher likelihood of concealed gun carriage included male sex (OR 5.1), older age/grade (OR 2.1), smoking (OR 5.5), alcohol use/binge drinking (OR 1.8), poor academic performance (OR 1.7), lack of self-efficacy to avoid fighting (OR 2.7), family member victim of shooting (OR 2.3), neighborhood with a lot of shootings (OR 2.9), and no serious discussion with parents about guns (OR 1.5) Limited generalizability
Oversampled minorities
Limited to NE/MW
Cross-sectional data.
Potential underestimate due to excluding truant/dropouts
No socio-demographic measures beyond sex
6
Hemenway et al. (2011) Cross-sectional
Biennial
Boston Youth Survey
1737 9th-12th grade students in Boston, MA recruited from 22 public high schools; Analytic sample restricted to those who answered questions on gun carriage 28% 10th grade
46% male.
41% Black
32% Hispanic
9% White
Past 12-month carriage = 5% Risk factors for carriage included male sex (OR 2.8), alcohol, tobacco, or drug use in past month (OR 2.4), lack of an adult who encouraged them (OR 2.7), having witnessed violence in past month (OR 2.3), recent peer victimization (OR 3.5), gang membership (OR 4.7), and gun access (OR 2.0). In addition, living in neighborhoods where gun carriage rates high (> 8%) predicted carriage (OR 2.2), as did youth overestimates of peers carrying (OR 2.5). On average, youth estimated that peer carriage in their neighborhood was 33% with mean estimates higher than “actual” levels in every neighborhood tested. Among youth who overestimated rates of peer carriage, 69% reported that they would be less likely to carry themselves Limited generalizability
Mainly minority youth sample
Mainly urban high schools
Cross sectional data
Self-report data
No information on non-responders (7%) or those not in school (31%)—potential for underestimate of actual level of gun carriage.
6
Kingery et al. (1996) Cross-sectional 1072 Randomly selected 8th (n = 464) and 10th (n = 608) grade middle and high school students in rural TX 49% Male
70% White
17% Black
Primarily rural population
Past 12-month gun carriage to school = 10% Discriminant analysis found 18 factors related to past year carriage, including exposure to unsafe neighborhoods, cocaine use (lifetime), recent physical fighting, riding on empty buses/trains, belief that carrying is effective to avoid fighting, earlier initiation of cocaine use, being victimized while on the school bus, going outside to sell items door-to-door alone, being threatened but not hurt in past year (outside of school), victim of forced sex, male sex, higher frequency crack use, attitudes favoring gang membership as means of avoiding fighting, higher grade, having possessions stolen by force/threat of force, less instruction in school on means of avoiding fighting/violence, White race, being threatened (but not hurt in past year) at school Limited generalizability
Rural Sample
Predominantly
White sample
Only assessed carriage and motivations for carriage to school
Cross-sectional data
Self-report data
4
Lane et al. (2004) Cross-sectional 223 Black youth living in low-SES neighborhoods (random digit dial sampling of houses) in San Francisco, CA Ages 13–19
M age 16
42% Male
100% Black
3-month intention to carry = 25% (males) and 9% (females) In males, intent to carry was associated with fear of victimization (OR 3.3) and delinquency (OR 14.2).
In females, intent to carry was associated with fear of victimization (OR 4.5) and delinquency (OR 4.1)
Limited generalizability
Small, Black youth sample from low-income neighborhood
Cross-sectional data
Self-report data
Intention to carry gun was proxy measure for carriage
7
Lizotte et al. (2000) Cohort
Rochester Youth Survey given at 6-month intervals (w 1–9) then 2.5 year intervals (w 9–10)
617 Male Youth recruited from Rochester public schools (7th/8th grade);
Oversampled high-crime areas (defined by arrest rates)
100% Male
M age 14 (w2)
M age 20 (w10)
63% Black
20% White
18% Hispanic
Analysis limited to male youth remaining at w10
Past 6-month carriage = 5–6% (w2) and 8–10% in w9–10. Those who carried guns had much higher rate of serious assaults and armed robberies (mean 0.8, p< 0.01). In early adolescence, gang membership is a strong predictor of illegal gun carriage (OR between 3.6 and 8.0 in w2–4). In older adolescence, selling drugs, especially high levels of drugs (OR between 8.0 and 34.7 in w5–10) and illegal peer gun ownership (OR between 2.4 and 23.8 in w5–10) replace gang membership as the primary determinants of illegal gun carriage. Drug use, particularly high drug use, is a significant predictor of carriage at almost every wave. Delinquent values has a sporadic impact on carriage with most of the effect occurring at mid-adolescence. Finally, carriage at a prior wave is a predictor (OR between 3.3 and 8.9) of carriage at nearly every wave (except w7 & 8) Self-reported data
Sample limited to male youth
Oversampled high-crime areas of Rochester, not a suburban or rural sample
6
Loeber et al. (2004) Cohort Develop-mental Trends Study with dyads at 12-month intervals for 13 years 177 Male children living with ≥ 1 biological parent referred from primary care clinic in either PA or GA Ages 7–12 (baseline); followed to 19
30% Black
70% White
53% urban
57% not living with bio-father
41% low SES
Lifetime carriage = 20%
1% carriage (age 12)
12% carriage (age 17)
Gun carriage significantly associated with older age youth (IRR 1.7), violent behavior (IRR 1.1), conduct disorder (IRR 5.2), and maternal psychopathy (IRR 1.1). Youth were less likely to carry firearms if they had been victimized (IRR 0.8). Protective factors included parental monitoring (IRR 1.1). Race/ethnicity, SES, rurality, and anxiety disorder were significant in bivariate model but were not included in the multivariate model Limited generalizability
Small, convenience sample referred from a healthcare clinic
Self-report data
Sample limited to male youth
7
Luster and Oh (2001) Cross-sectional 1997 National Longitudinal Youth Survey (NLYS97) 4619 Representative sample of U.S. male youth from the NLYS97 survey Analytic sample limited to males Separate analysis for those < 15; and those ≥ 15 Age 12–16
100% male
50% White,
25% Black, 21% Hispanic
57% Urban
43% Rural
M Income = 47 K
Past 1-year carriage = 9% Lifetime = 16% <15: past 1-year carriage = 8%
≥ 15 past 1-year carriage = 11%
Predictors of handgun carrying:
Under age 15: White youth (OR 2.9), Problem Behaviors
(OR 1.5), witnessed shooting before age 12 (OR 2.1),
Relatives of friends in a gang (OR 1.7), gang membership (OR 3.0), hearing gunshots in their neighborhood (OR 1.2). Protective factors included parental monitoring (OR 0.9) and high-levels of maternal respect (OR 0.9)
Over age 15: Problem Behaviors (OR 1.5), witnessed shooting before age 12 (OR 2.4), negative peer influence (1.3), and gang involvement (OR 3.2)
Limited generalizability
Oversampled Black and
Hispanic youth
Cross-sectional data
Self-report data
Excluded those who carried a gun previously but not within the past 12-months
6
May (1999) Cross-sectional 8338 High school students from urban and rural counties in MS selected via two-stage process (urban, non-urban) Ages 13–20.
45% Male
53% Black
47% White
Lifetime carriage to school = 8% Factors associated with gun carriage to school included male sex (OR 4.4), Black race (OR 1.6), older age (OR 1.7), higher household income (OR 1.1), gang membership (OR 5.3), perceived neighborhood incivility (OR 1.1), and higher fear index (OR 1.1). Gun carrying to school was less likely in those from two-parent homes (OR 0.8) and with higher social control index scores (OR 0.9) Limited generalizability
Majority Black sample from MS
Non-random selection
Cross-sectional data
Self-report data
Note: study sample taken from 1992 (published 1999)
5
McNabb et al. (1996) Case control 38 cases; 103 controls Youth (< age 19) charged with illegal gun carriage in Jefferson Parish LA & age-, gender- and school-matched controls recruited from public school (3:1 match) Ages 13–18
M age 16
94% Male
46% White
43% Black
Past 30-day carriage
Cases = 71%
Controls = 20%
Lifetime carriage
Cases = 100%
Controls = 29%
School carriage
Cases = 34%
Controls = 4%
Risk factors for firearm carriage included reporting that the school was not safe (AOR 9.0), seen a shooting (AOR 7.0), marijuana use (OR 6.8), and a history of firing a gun (OR 17)
Risk factors for being charged with firearm carriage included the lack of a employed male in households with male parents (AOR = 8.6), marijuana use (AOR = 11.7), and watching TV for more than 6 h/day (AOR 6.5)
Cases significantly more likely to be Black youth than gun-carrying controls
Potential bias of underreported delinquency
Low case response rate (54%) may misrepresent true rate of gun carriage
Self-report data
7
Molnar et al. (2004) Cross-sectional
Project Human Development
1842 Population-based sample of age 9–19-year-old youth from 218 neighborhoods in Chicago, IL Ages 9–19
36% age 9–12
36% age 13–15
29% age 16–19
50% male
30% Black
42% Hispanic
14% White
25% low SES
Lifetime carriage = 3% (4.9% boys, 1.1% girls) After controlling for individual/family factors, neighborhood factors significant in separately tested models included lack of neighborhood safety (OR 5.8), neighborhood social disorder (OR 1.9), and neighborhood physical disorder (OR 2.3). Neighborhood protective factors included collective efficacy (OR 0.3). Significant individual/family factors influencing gun carriage differed by model, but included male gender, older age, presence of guns at home, prior family member shot with gun, witnessed prior violence, and prior victimization by violence. Note, 76% of those who carried a gun had a family member shot by gun. Most gun carriers (63%) were in highest quartile on scale of delinquent & aggressive behaviors (p < 0.05) Limited generalizability
Single urban sample with large Hispanic population
Cross-sectional data
Self-report data
7
Orpinas et al. (1999) Cross-sectional
Students for Peace Study
8865 6th-8th graders from 8 urban middle schools in a large TX school district M age 13
66% Hispanic
19% Black
8% White
4% Asian
Past 30-day carriage = 10% Students with low parental monitoring were significantly more likely to carry a handgun than those who had very high monitoring (Boys OR 19.8, Girls OR 26.3). Students who got along “very bad” with their parents were more likely to carry a handgun than those who get along “very well” (Boys OR 7.9, Girls OR 22.7). No multivariate model predicting gun carriage (only weapon carriage) Limited generalizability
Single urban school district
Mainly Hispanic sample
Cross-sectional data
Self-report data
Missed students not in school
7
Peleg-Oren et al. (2009) Cross-sectional
FL Youth Substance Abuse Survey (FYSAS) & FL YRBS
12,352 Randomly selected 11th and 12th grade students from FL schools. FYSAS data collected 2006 (N = 10,626) and YRBS in 2005 (N = 1726) YRBS
54% Female
57% 11th grade
50% White
24% Hispanic
FYSAS
54% Female
58% 11th grade
61% White
14% Hispanic
YRBS:
Past 30-day carriage to school = 5%
FYSAS
Past 1-year = 5% (outside school)
Past 1-year = 1% (in school)
YRBS: After controlling for socio-demographic factors (sex, race, grades), very early drinkers (vs. early drinkers) [AOR = 3.1] and very early drinkers (vs. non-drinkers) [AOR = 29.4] were predictive of gun carriage.
FYSAS: After controlling for socio-demographic factors (sex, race, grades), very early drinkers (vs. early drinkers) [AOR = 2.5] and very early drinkers (vs. non-drinkers) [AOR = 5.6] were predictive of gun carriage and also gun carriage to school (AOR of 2.6 and 5.1, respectively)
Cross-sectional data.
Self-report data.
Adjusted only for socio-demographic factors
6
Reid et al. (2017) Cohort
Pathways to Desistance Study, given every 6 months for 2 years
1170 Convicted male youth offenders in PA and AZ (recruited from court). Proportion charged with drug-related offenses capped at 15% Ages 14–19,
M age 16
100% male
42% Black
34% Hispanic
19% White
Lifetime carriage at beginning of study = 51% After controlling for study site, race/ethnicity, age, and proportion of time on the streets, global severity index score (i.e., psychological distress measures) was predictive of gun carriage at 3 of 4 time points (12, 18, 24-months) with OR ranging from 1.4 to 2.3
After adding exposure to violence (witnessed and experienced) to the model, the global severity index was no longer a significant predictor of gun carrying. However, exposure to violence remains significant at every wave with increasing odds ratios ranging from 1.4 to 1.6
Limited generalizability
Sample limited to male juvenile offenders from two cities
Self-report data
6
Ruggles and Rajan (2014) Cross-sectional
CDC Youth Risk Behavioral Survey (YRBSS)
88,608
(Six waves of data)
9th-11th grade students drawn from randomized sample of US high schools. Analytic window is YRBSS data from 2001 to 2011 Not reported Past 30-days
5.7% (2001)
6.1% (2003)
5.4% (2005)
5.2% (2007)
5.9% (2009)
5.1% (2011)
43 out of 54 risk behaviors included in the YRBSS were associated with carriage, including strongest associations with alcohol, tobacco, and drug use overall and at school. Gun carriage was also strongly associated with feeling unsafe and being threatened at school. Finally, carriage also had associations with being the victim of sexual assault, to be engaged in disordered eating behaviors, to not wear sunscreen regularly, to have riden in a car with a drunk driver, and mental health factors (e.g., suicidality) Cross-sectional data
Self-report data
Limitations associated with hierarchical clustering methodology
Behaviors could be proxies for broader social issues
7
Sheley (1994) Cross-sectional 758 Male students from urban schools in 5 cities in CA, IL, LA, and NJ. All schools had prior gun incidents M age 16
Range (age 15–17)
72% Black
3% White
19% Hispanic
12% carried routinely (all/most time)
23% carried now/then
Youth who used and sold drugs or only sold drugs but did not use drugs were more likely to report carrying a gun routinely (19% vs. 5%, p < 0.05)
Those who endorsed heavy drug use had higher rates of gun carriage (72 vs 10%, p < 0.05).
Cross-sectional data
Not a representative sample of urban youth (non-random recruitment)
Self-report data
4
Sheley and Brewer (1995) Cross-sectional 418 10 and 11th grade students at 3 of 7 suburban public high schools in Jefferson Parish LA M age 16
48% male
66% White, 21% Black
Current carriage = 17% Gun carriage was associated with male sex (AOR = 1.1), White race (AOR = 1.4), drug activity (AOR = 1.2), and violent criminality (AOR = 1.7). Dangerous environmental exposure is significant overall (i.e., threatened with gun). Environmental findings differend by sex, with having been threatened with gun significant for boys, while fear of being shot by age 25 was significant for girls Limited generalizability
Suburban sample
Cross-sectional data
Self-report data
6
Simon et al. (1998) Cohort 2200 9th and 12th grade students from 6 school districts in San Diego and Los Angeles, CA.
Analytic sample is 12th grade students with complete data
44% Male
30% White
38% Hispanic
8% Black
11% Asian
32% low SES
Lifetime gun carriage
Boys = 22%
Girls = 5%
Psychosocial: Risk-taking (male = AOR 3.1; female = AOR 4.1), depression (male = AOR 1.6; female = AOR 2.4), stress (male = AOR 1.9; female = AOR 2.8), and temper (male = AOR 1.9; female = AOR 2.8) in 9th grade was predictive of carriage in 12th grade for male and female students
Behavioral: 9th grade factors including 3 + days school absence (AOR 2.4), 2 + parties in prior month (AOR 1.9), cigarette (AOR = 2.2), alcohol (AOR = 1.9), and MJ (AOR 2.3) use were predictive of carriage in 12th grade for males. For females, cigarette (AOR 5.1), alcohol (AOR 3.7), and MJ (AOR 3.6) use in 9th grade predicted 12th grade carriage
Perceptions neighborhood crime and SES were associated w/carriage after adjusting for demographics (age, gender)
Limited generalizability
Geographic distribution
Self-report data
Only included students completing survey at both timepoints (may miss school dropouts who may be more likely to carry guns)
7
Tigri et al. (2016) Cohort National Longitudinal Survey of Youth, data from 5 yrs of annual surveys 5018 Nationally representative sample of youth throughout the US. Analysis excludes NLSY oversample of Black and Hispanic youth M age 14 at baseline;
M age 19 at last wave
50% male
70% White
Lifetime carriage = 9% at baseline
Past 1-year carriage = 4% at final wave
After adjusting for age, race, sex, prior carriage, prior peer gang membership, prior gang membership, prior delinquency, current gang membership (OR ranging 2.5–4.2), current peer gang membership (OR ranging 1.6 to 2.2), and delinquency behaviors OR ranging 1.8–2.8) were associated with carriage. Gang membership was strongest RF; delinquency measure most consistent. When examined by sex, gang membership is stronger for both sexes in early adolescence and diminishes in late adolescence (esp for females). Association of gang membership with carriage was inconsistent across race/ethnicity Self-report data
No characterization of gun carrying frequency
No temporal causality given that RF were examined concurrently at time points (although accounted for lagged RF from prior waves of data)
5
Turner et al. (2016) Cohort National Longitudinal Survey of Youth, Sample drawn from 14 years of annual data 6641 Nationally representative sample of youth throughout the US. Analysis excludes NLSY oversample of Black and Hispanic youth. M age 14 at start of study, M age 27 at end of study.
51% Male
70% White 16% Black
14% Hispanic
Lifetime carriage = 28%
Past 1-year carriage = 24%
Past 30-day carriage = 13%
Past 30-day carriage to school = 1%
Repeat bullying victims were more likely (than non-victims) to carry during past 1-year (OR 1.3) and 30-days (OR 1.2) after controlling for prior carriage and socio-demographic factors, peer gang membership, neighborhood gang presence, gang membership. When examined by age, only childhood bullying victims remained significantly associated with carriage; adolescent victims and victims during childhood & adolescence were not associated with carriage. Repeat bullying and childhood repeat bullying predicted gun carriage in propensity score matching analysis Self-report data
Recall bias from retrospective measure of bullying
Unclear cross-over effect (i.e., proportion of bullying victims were also aggressors)
6
Vaughn et al. (2012) Cross-sectional
2008 National Survey on Drug Use and Health (NSDUH)
17,842 Multistage probability sample weighted to create a nationally representative sample of US youth (age > 12); Sample restricted to age 12–17 y/o Mean age 15
51% Male
59% White
18% Hispanic
14% Black
16% low SES
Past year carriage = 3% Factors associated with carriage included male sex (AOR 4.5), prior incarceration (AOR 3.8), selling drugs (OR 16.1), stealing > $50 (AOR 7.5), aggression (AOR 7.0), serious fighting at school (AOR 3.8), ecstasy (OR 5.3), hallucinogen (AOR 4.1), cocaine/crack (AOR 7.7), MJ (AOR 4.1), or heroin (AOR 5.3) use, danger seeking (AOR 4.4), and risk taking behavior (AOR 4.7). Across a range of parental involvement and monitoring behaviors, all were significantly associated with lower likelihood of carrying a gun. Youth who were exposed to violence prevention and drug prevention programming/messaging outside of school were associated with lower likelihood of gun carriage Cross-sectional data
Self-report data
7
Vaughn et al. (2016) Cross-sectional
National Survey on Drug Use and Health (NSDUH)
197,313 Multistage probability sample weighted to create a nationally representative sample of US youth (age > 12); Sample restricted to age 12–17 y/o Ages 12–17
51% male
66% White
15% Black
19% Hispanic
Past year carriage = 3% Carriage is examined by race/ethnicity (White, Black, Hispanic). Across all groups, male gender and history of violence/delinquency (i.e., fighting at school/work, violent aggression, selling drugs, stole > $50) were associated with carriage. Black youth with household income < $20 K were more likely to carry (OR 2.2). Hispanic youth engaged in binge drinking were more likely to carry (OR 1.8). Marijuana use (OR 1.5) and having substance using friends (OR 1.3) increased likelihood of carriage in Black youth. High parental affirmation protected against carriage for White (OR 0.7) and Hispanic (OR 0.6) youth, while parental control was protective for White (OR 0.9) youth only Self-report data
Cross-sectional data
May have missed adolescents most likely to carry a gun (i.e., not in school)
Does not investigate specific contextual factors that relate to gun carrying
7
Watkins et al. (2008) Cross-sectional 967 338 males in juvenile detention (< age 17); 629 incarcerated adult male; Recruited from St. Louis, MO Correctional Facilities Adolescents
M age 15
100% Male
94% Black
Adults
M age 31
100% Male
87% Black
Past 12-month carriage 46% most/all of time 41% seldom Among subsample of juvenile males in detention who also endorsed gun possession (N = 202), factors predicting an increased frequency of gun carriage included age, black race, and gang membership. Among perceptual factors, you who perceived an increase in the prevalence of guns over the past year were less likely to report gun carriage
Deterrent measures, such as perceptions about gun use penalties and increased risk of arrest were not significant and appeared to have no effect on gun carriage
Limited generalizability
Small sample of juvenile arrestees from single city
Mostly Black sample
Cross-sectional data
Self-report
6
Webster et al. (1993) Cross-sectional 294 7th and 8th grade students at two inner-city public high schools in low SES/high-crime areas of Washington, D.C. Ages 11–16
M age 14
(School A)
M age 13
(school B)
100% Black
Past 2 wk carriage = 16%
(carriers)
Lifetime carriage
Males = 23% (A)
Males = 40% (B)
Females = 4%(A)
Females = 5% (B)
Factors predicting gun carriage (for protection or use in a fight) among male youth included prior arrest (OR 16.1), knowing more victims of violence (OR 1.1), history of initiating fights (OR 51.5), perceptions about peer acceptability of violence behaviors (norms) (OR 1.2), and more willingness to endorse views that there are justifiable reasons to shoot someone (OR 1.6). Of note, all males who had a history of arrest for drug-related charges reported carrying a firearm in the sample Limited generalizability
Small geographic
Distribution
Analysis restricted to male
and Black youth
Cross-sectional data
Self-report data
Convenience sample
4
Wilcox et al. (2006) Cohort
Rural Substance Abuse and Violence Project (RSVP)
3968 7th-9th grade students from 113 public middle/high schools in KY followed annually for 4 years (2001–2004) 48% Male
89% White
Analysis restricted to first three waves of data
Past yr school gun carriage reported as ordinal scale (1–5) with 1 = never; 5 = daily W1 M = 1.04
W2 M = 1.04
Using SEM, authors found support for the “triggering” over the “fear and victimization” hypothesis around carriage. They found that the frequency of carriage in 8th grade was positively associated with 9th grade fear, risk perception, victimization, and offending (supporting the triggering hypothesis). Authors also found that fear and victimization in 7th grade was not related to carriage in 8th grade (non-significant) and that 7th grade risk perception was negatively related to carriage (contradicting the fear and victimization hypothesis). Further, previous carriage and gun ownership were strong predictors related to gun carriage. Limited generalizability
KY sample
Sample size limited by need for parental consent (low response rate, potentially biasing sample)
Self-report data
Large number of missing data cases
5
Williams et al. (2002) Cross-sectional 21,981 Representative sample of 6th, 8th, and 10th grade public school students across 27 IL communities (ages 10–19) Ages 10–19
M age 14
50% Male.
65% White
13% Black
12% Latino
64% urban or suburban
32% low SES (eligible for school lunch)
Past year carriage = 5%
Past year carriage to school = 1%
Demographic/Hangun Model: Across models, male gender, being seen as being cool if carrying a gun, less parental monitoring, and increased number of peers carrying guns was predictive of having ever carried, having ever carried to school, and having ever carried and having carried to school. SES was inconsistently associated. Attitudes favoring not taking a gun to school were protective.
Violence/Delinquency Model: Across models, increased freq of aggression, gang membership, prior arrest, and substance use were associated with having ever carried, carrying to school, & having ever carried/carrying to school.
Family, School, Community Models: Across models, no consistent findings regarding family, school, or community variables with regards to having ever carried, carrying to school, and having every carried/carrying to school
Cross-sectional data
Self-report data
Survey distributed by untrained teachers rather than research staff
No measures assessed victimization
7
Xuan and Hemenway (2015) Cross-sectional
Data was from the 2007, 2009, and 2011 Youth Risk Behavior Survey (YRBS)
228,904 Nationally representative sample of 9th-12th graders drawn from 38 U.S. states Not reported Past 30-day gun carriage = 7% Gun carriage among youth in 19 states with stronger gun laws was 5.7% compared to youth in the 19 states with weaker laws where carriage was 7.3%. A 10 point increase in firearm law score (i.e., strength of the gun laws for 5 areas—curbing firearm trafficking, stronger background checks, child safety laws, military assault-style weapon ban, and restricting guns in public places) was associated with a 9% reduction in the odds of youth gun carriage (AOR 0.9). Adult firearm ownership mediated the association between the state gun law score and youth gun carriage (AOR 0.9) with 29% attenuation of the regression coefficient Cross-sectional data
Self-report data
Non-validated scoring for strength of state firearm laws
Local policies were not included in the weighted firearm law scores
5