Measured by the ratio of firearm-related suicide divided by all suicide (FS/S ratio). Youth gun carrying data from Youth Risk Behavior Surveillance Surveys, 2007, 2009, and 2011.13
eFigure. State prevalence (%) of youth gun carrying and associated gun law score quartiles. Youth gun carrying data from Youth Risk Behavior Surveys, 2007, 2009 and 2011 (N=38 states).
Xuan Z, Hemenway D. State Gun Law Environment and Youth Gun Carrying in the United States. JAMA Pediatr. 2015;169(11):1024-1031. doi:10.1001/jamapediatrics.2015.2116
Gun violence and injuries pose a substantial threat to children and youth in the United States. Existing evidence points to the need for interventions and policies for keeping guns out of the hands of children and youth.
(1) To examine the association between state gun law environment and youth gun carrying in the United States, and (2) to determine whether adult gun ownership mediates this association.
Design, Setting, and Participants
This was a repeated cross-sectional observational study design with 3 years of data on youth gun carrying from US states. The Youth Risk Behavior Survey comprises data of representative samples of students in grades 9 to 12 from biennial years of 2007, 2009, and 2011. We hypothesized that states with more restrictive gun laws have lower rates of youth gun carrying, and this association is mediated by adult gun ownership.
State gun law environment as measured by state gun law score.
Main Outcomes and Measures
Youth gun carrying was defined as having carried a gun on at least 1 day during the 30 days before the survey.
In the fully adjusted model, a 10-point increase in the state gun law score, which represented a more restrictive gun law environment, was associated with a 9% decrease in the odds of youth gun carrying (adjusted odds ratio [AOR], 0.91 [95% CI, 0.86-0.96]). Adult gun ownership mediated the association between state gun law score and youth gun carrying (AOR, 0.94 [ 95% CI, 0.86-1.01], with 29% attenuation of the regression coefficient from −0.09 to −0.07 based on bootstrap resampling).
Conclusions and Relevance
More restrictive overall gun control policies are associated with a reduced likelihood of youth gun carrying. These findings are relevant to gun policy debates about the critical importance of strengthening overall gun law environment to prevent youth gun carrying.
From 1999 to 2013, an average of some 15 000 teenagers 12 to 19 years old died annually in the United States. The 3 leading causes of death among teenagers were unintentional injuries (45%), homicide (14%), and suicide (13%).1 Among these fatal youth injuries, most homicides were gun-related (83%), and about half of suicides involved a gun (45%).2 When adolescents resolve their interpersonal conflicts with guns, serious or fatal injuries are likely consequences.3
There is a strong consensus in the United States that unsupervised youth should not carry guns in public. The federal Gun Control Act of 1968 prohibits gun dealers from selling handguns to youth (<21 years) and selling rifles and shotguns to those younger than 18 years. A 1994 amendment further determined 18 years as the national minimum age for purchasing or possessing handguns or handgun ammunition. Many state laws include age-specific restrictions on the purchase, possession, or storage of firearms.
Research on youth-focused gun laws is limited and has shown mixed results. A study4 showed that state child access prevention (CAP) laws were associated with a modest reduction in suicide rates among youth aged 14 to 17 years, whereas minimum age restrictions for the purchase and possession of firearms do not seem to reduce overall rates of suicide among youth. Even for CAP laws, a study found that the aggregate benefits across multiple states on reducing deaths among youth 15 years or younger from unintentional shootings were mainly attributable to a single state law.5 Another study6 found no support of laws prohibiting gun possession by youth in reducing youth being killed by guns or with their use of a gun to commit homicide.
The limited impact of youth-focused gun laws may be explained by the wide prevalence of gun ownership. The presence of a household gun is associated with increased risk of suicide among adolescents,7- 9 and most guns used among suicide attempters and completers are stored in the residence of the suicide attempter or completer, friend, or relative10; most are family guns.11 Thus, the effect of gun laws on youth suicide should focus on the overall gun law environment and not just laws directly focusing on youth.
The study of the state gun law environment is limited. A recent study12 used the state enactment of up to 28 laws to determine the degree of firearm legislation in the state and found that more firearm laws in a state was associated with lower firearm-related fatalities. The summative approach assumed that each gun law had an equal impact; in reality, the effect of each gun law on gun availability is likely unequal. An alternative approach, weighting the relative strength of each law, may provide a better way of characterizing the gun law environment.
We could not find any study that assessed the association between overall gun laws and youth gun carrying or whether adult gun ownership mediates the association between gun laws and youth gun carrying. According to a report13 from the Centers for Disease Control and Prevention (CDC) in 2011, there was considerable state variation of gun carrying prevalence among high school youth.
The objectives of this study were to (1) assess the association between the gun law environment and youth gun carrying and (2) determine whether adult gun ownership mediates this association.
The purpose of the study was to investigate the association between state gun law environment and youth gun carrying in the United States, and whether this association is mediated by adult gun ownership.
Among 38 states in our study, 5.7% of high school students living in the 19 states with stricter gun laws carried a gun in past 30 days while 7.3% of students living in states with the weaker gun laws carried a gun.
A 10-point increase in the strictness of the state gun law score was associated with a 9% decrease in the odds of youth gun carrying.
Across states, restrictive gun laws may reduce youth gun carrying by limiting adult gun ownership.
To correspond to the years of available state gun law data, we constructed 3 years of repeated cross-sectional nationwide surveys for 2007, 2009, and 2011 on individual-level youth gun carrying from the state-based Youth Risk Behavior Survey (YRBS).13 Data from states with response rates of greater than 60% were weighted to be representative of that state’s population of students in grades 9 to 12.14 For the study period from 2007 to 2011, the total number of states with weighted data on youth gun carrying was 31 states in 2007, 31 states in 2009, and 27 states in 2011; a total of 228 904 respondents answered the youth gun carrying question from 89 state-year strata representing 38 states.
To characterize the gun law environment, we used data from the Brady Center to Prevent Gun Violence,15 where state gun law scorecards have been released annually since 2007. The scorecards divided gun laws into 5 subgroups based on targeted content domains: curb firearm trafficking, strengthen background checks, ensure child safety, ban military-style assault weapons, and restrict guns in public places. Within these scorecards, each state was given points depending on the enactment of up to 28 laws from these 5 domains. The Brady Center scorecards developed a weighted method providing additional weight to laws considered to be of greater importance, then rescaled the total scores to a possible range from 0 to 100 points, with greater value representing a more restrictive gun control environment. Detailed descriptions of each law, content domains, and the weighted algorithm are available from the Brady Center.15
Youth gun carrying was defined as a binary variable of having carried a gun on at least 1 day during the 30 days before the survey. The reliability κ for the survey measure (κ = 50.8%), calculated using 1999 data, was classified as between “moderate” and “substantial.”16,17
For our study, adult gun ownership is hypothesized as the potential mediator for the association between the gun law environment and youth gun carrying. Unfortunately, during the study period, 2007 to 2011, there was no annual national survey that estimated the prevalence of adult gun ownership in all 50 states; therefore, we used a well-documented proxy measure: the ratio of suicides committed with a gun to all suicides (FS/S ratio).18- 24 The FS/S ratio has been shown to correlate highly with survey measures of gun ownership18,25 and has been the most widely used proxy measure for state-level gun availability.19,20,24- 30 A recent study31 further proposed a new proxy measure based on a weighted average of the state-level FS/S ratio and the hunting license rate. For the current study, we used the FS/S ratio among adults at least 21 years old as the primary proxy measure for adult gun ownership and conducted a sensitivity analysis with the weighted average measure for adult gun ownership from the recent study.
The most current national survey on state-level adult gun ownership comes from the CDC’s Behavioral Risk Factor Surveillance System survey in 2004, which measured the prevalence of adult gun ownership in all 50 states.32 For the current study, we found the Spearman correlations between 2004 measure of state adult gun ownership and the adult FS/S ratio to be highly correlated across all 50 states (ρ = 0.75, P < .01 for 2007; ρ = 0.78, P < .01 for 2009; ρ = 0.82, P < .01 for 2011).
State-level covariates include the proportions (percentages) of adults at least 21 years old; male; white, black, and other racial group; percentage living in urban areas; median household income; percentage unemployed; number of police officers per 1000 residents, and regions. Year is treated as a categorical variable. Individual-level covariates include age, sex (male vs female), race/ethnicity (non-Hispanic white, non-Hispanic black, non-Hispanic other, and Hispanic), and grade level.
For the state-level analysis, Spearman correlation was used to examine the correlations among the state gun law score, adult gun ownership, and the state prevalence of youth gun carrying. We further used simple linear regression to regress the outcome of the weighted state-level youth gun carrying prevalence (ie, weighted by state population) to the state gun law score. For the individual-level analysis, we used logistic regression to examine the odds of youth gun carrying. Individual-level odds ratios were based on a 10-point increase in the state gun law score using generalized estimating equations, adjusting for the clustering of individuals within the sampling units. Because multiple years of surveys were included, survey weights from the YRBS were recalculated with multiple-year adjustment. We further stratified on the basis of sex, grade level, and race/ethnicity. Finally, mediation analyses tested whether adult gun ownership (FS/S ratio) mediated the association between state gun law environment and youth gun carrying.33 We used the following procedures to evaluate the mediation effect. First, the state gun law score had to be associated with adult gun ownership. Second, the state gun law score had to be associated with youth gun carrying in the absence of adult gun ownership. Third, the adult gun ownership had to be associated with youth gun carrying in the absence of the state gun law score. Fourth, the association between state gun law score and youth gun carrying was attenuated with adjustment for adult gun ownership. Finally, we ascertained that there was no statistical interaction between state gun law score and adult gun ownership. We used bootstrap resampling (N = 500 replicates) to obtain estimates to evaluate the mediating effect. In addition, we used Sobel test to assess the statistical significance of the mediator.34,35 All analyses were conducted with SAS statistical software (version 9.3; SAS Institute Inc).
There is substantial variation of state-level gun law score across the 50 states, with average scores across 2007, 2009, and 2011 ranging from 1.3 (Utah) to 79.7 (California) (Table 1). The adult gun ownership as measured by FS/S ratio ranged from 20.0% (Massachusetts) to 70.9% (Mississippi), with an average FS/S ratio of 52.0%. Among the 38 states with data on youth gun carrying, the average aggregate prevalence of youth gun carrying was 6.7%, ranging from 1.4% (New Jersey) to 11.0% (Wyoming). The eFigure in the Supplement shows the state prevalence of youth gun carrying and associated gun law score in quartiles based on the mean score across 2007, 2009, and 2011.
The Figure shows Spearman correlations among state gun law scores, adult gun ownership, and prevalence of youth gun carrying across 3 survey years among 38 states with available youth gun carrying data.
Across the 89 state-year strata, a 10-point increase in the gun law score was associated with a 0.39 (P = .049) absolute percentage point decrease in youth gun carrying prevalence, controlling for state-level covariates, regions, and year (Table 2). The Northeastern, Midwestern, and Southern regions showed significantly lower prevalence of youth gun carrying relative to the West region, while other state-level covariates were not statistically significant.
For the individual-level analysis, a higher gun law score was associated with a lower likelihood of youth gun carrying in both bivariate and adjusted models (Table 3). Those who were 16 years or older, male, or non-Hispanic others had greater odds of gun carrying. Those who were in Midwestern or Southern regions were associated with lower odds of gun carrying. While there was a positive association between the state proportion of non-Hispanic black and the odds of youth gun carrying, other state-level covariates were not statistically significant (Table 3).
In the fully adjusted model on individual-level outcome, a 10-point increase in the gun law score was associated with 9% lower odds of youth gun carrying (adjusted odds ratio [AOR], 0.91 [95% CI, 0.86-0.96]) among the overall sample (Table 4). Among demographic subgroups, the reduced odds of youth gun carrying was significant and similar based on both sex groups, the group of 11th and 12th graders, and among non-Hispanic white and non-Hispanic other races. Among non-Hispanic black and Hispanic youth, and among the group of ninth and 10th graders, the associations were not significant.
With respect to state-level analysis, we performed 5 steps in ascertaining mediation effect. First, adjusting for state-level covariates, years and regions across the 89 state-year strata, the state gun law score was inversely associated with the adult gun ownership (β = −3.49; P < .01). Second, as shown in adjusted model 2 in Table 2, the state gun law score was inversely associated with the prevalence of youth gun carrying (β = −0.39; P = .049). Third, higher adult gun ownership levels were associated with a higher prevalence of youth gun carrying (β = 0.08; P < .01). Fourth, the adjusted association between the state gun law score and the prevalence of youth gun carrying was no longer significant after adjustment for adult gun ownership (β = −0.17 [95% CI, −0.59-0.25] based on 500 bootstrap resampling replications). Finally, there was no statistical interaction between the gun law score and adult gun ownership (β = −0.01 for the interaction term; P = .56). We observed approximately 56% attenuation (ie, from coefficient of −0.39 to −0.17) of the association between state gun law score and prevalence of youth gun carrying. We found evidence of mediation for the fully adjusted model (Sobel test statistic = −2.18; P = .03).
For the mediation analysis at the individual level, the gun law score was inversely associated with adult gun ownership (β = −3.49; P < .01), and in the absence of adult gun ownership, the state gun law score was inversely associated with the odds of youth gun carrying (β = −0.09; P < .01; AOR, 0.91 [95% CI, 0.86-0.96]) (Table 3 and Table 4). Absent of gun law score, there was significant association between adult gun ownership and the odds of youth gun carrying (β = 0.01; P = .01; AOR, 1.01 [95% CI, 1.003-1.02]). However, the association between gun laws and odds of youth gun carrying was no longer significant after adjusting for adult gun ownership (β = −0.07 [95% CI, −0.15-0.01]; AOR, 0.94 [95% CI, 0.86-1.01], based on 500 bootstrap resampling replications). Finally, there was no statistical interaction between the gun law score and adult gun ownership (β = −0.001 for the interaction term; P = .57). We observed approximately 29% attenuation of the β coefficient (ie, from a coefficient of −0.09 to −0.07) and a 3% attenuation of the AOR (ie, from an AOR of 0.91 to 0.94), respectively, in the adjusted association between state gun law score and prevalence of youth gun carrying. We found evidence of full mediation for the fully adjusted model (Sobel test statistic = −1.14; P = .03).
Separate sensitivity analysis with gun ownership measured as the weighted average of FS/S and the hunting license rate yielded generally consistent results.
To our knowledge, this is the first study to ascertain the association between the state gun law environment and gun carrying among high school youth in the United States. We found an inverse association between gun policies and both state-level prevalence and individual-level odds of youth gun carrying. Specifically, a 10-point increase in gun law score (indicating a stronger gun control environment) was associated with a 9% reduction in the odds of youth gun carrying in the fully adjusted individual-level model.
The associations between state gun law score and youth gun carrying were significant for both girls and boys, and for the group of 11th and 12th graders. The associations were significant for non-Hispanic white individuals and non-Hispanic other races. The associations were not statistically significant for the younger group of ninth and 10th graders, nor for non-Hispanic black or Hispanic individuals, although the direction of the protective effect on gun laws was consistent with expectation. Notably, non-Hispanic black and Hispanic youth have a high prevalence of gun carrying compared with non-Hispanic white youth,13 and these 2 groups may be subject to other individual- or neighborhood-level factors unaccounted for in this study.
We also found support of full mediation of adult gun ownership for the association between state gun laws and youth gun carrying. Our findings are consistent with the evidence base from survey literature showing that about a quarter of US adolescents reported easy access to a gun in the home,36 and the number of suicide, homicide, and unintentional firearm deaths among youth is higher in states and regions with higher levels of household gun ownership.29,37,38 Evidence from the 1995 National Survey of Adolescent Males39 showed that the likelihood of gun carrying increases notably with gun ownership prevalence in the community. Our study extends the evidence base by demonstrating the explanatory role of adult gun ownership in the association between state gun laws and youth gun carrying, using multiple years of YRBS data.
Our study is subject to several limitations. The findings are largely cross-sectional in nature, despite controlling for unrestricted time effects. Neighborhood-level unmeasured factors, including culture and attitudes toward guns, and individual-level factors, including socioeconomic status, may confound the association between gun laws and youth gun carrying. Although we accounted for a wide range of relevant state- and individual-level covariates, including urbanization, selection bias remains a potential threat to internal validity. While we used a weighted score of state gun laws based on laws’ respective importance, these scores provided by Brady Center represent only one way of characterizing the state gun law environment. Other approaches that rely on a slightly different set of state gun laws with different weighting algorithms may characterize the state gun law environment differently. Validation of the scores assigned by the Brady Center is warranted in future study. Furthermore, policies that are promulgated at the local level were not included in the weighted gun law scores. In addition, enforcement efforts likely contribute to the effectiveness of gun policies, but there is no publicly available information about levels of gun policy enforcement. The YRBS samples include high school youth who attend school and are in school during the day of the survey and are therefore not fully representative of all youth. Finally, the gun carrying data are based on self-report, and there are no objective data on whether the individual respondent actually has carried a gun.
As a final caveat, we must emphasize that our largely cross-sectional analysis examines only associations and cannot determine causation. For example, we cannot rule out the possibility that levels of youth gun carrying could somehow affect the gun law environment or the level of household gun ownership.
Gun violence poses a substantial public health threat to adolescents in the United States. Existing evidence points to the need for policies to reduce gun carrying among youth. We find that the strength of gun policies including both adult-focused and youth-focused policies is inversely associated with youth gun carrying. These findings are relevant to gun policy debates about the critical importance of comprehensive state-level gun law environment to prevent youth gun carrying.
Corresponding Author: Ziming Xuan, ScD, SM, MA, Department of Community Health Sciences, Boston University School of Public Health, 801 Massachusetts Ave, Boston, MA 02118 (email@example.com).
Accepted for Publication: June 19, 2015.
Published Online: September 21, 2015. doi:10.1001/jamapediatrics.2015.2116.
Author Contributions: Dr Xuan had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Xuan.
Acquisition, analysis, or interpretation of data: Both authors.
Drafting of the manuscript: Xuan.
Critical revision of the manuscript for important intellectual content: Both authors.
Statistical analysis: Xuan.
Conflict of Interest Disclosures: None reported.