National Instant Criminal Background Check and Youth Gun Carrying

We collected cross-sectional survey data from the national, school-based Youth Risk Behavior Survey (YRBS) from 1993 to 2017. The YRBS is a biennial survey conducted by the Centers for Disease Control and Prevention. With a 3-staged cluster-sampling survey design, data from the YRBS provides a nationally representative sample of US public and private school students in grades 9 to 12. A self-administered questionnaire is used to collect anonymous and voluntary responses from participating students on their demographics and health risk behaviors. In total, there were 191 391 responses over the period between 1993 and 2017. Details about YRBS sampling strategies and methodologies are reported elsewhere.¹⁸ 

Approximately 2.0% of the responses were missing state identifiers and 4.1% were missing information on the gun carrying survey item, so these responses were dropped from the analysis. Our final analysis included responses from 179 857 students from 1993 to 2017.

The YRBS in 1993–2015 was used to assess whether students carried guns by asking “During the past 30 days, on how many days did you carry a gun?” We dichotomized this question to flag those students who carried a gun on at least 1 day during the 30 days before the survey. In 2017, this question was modified to “During the past 12 months, on how many days did you carry a gun? (Do not count the days when you carried a gun only for hunting or for a sport, such as target shooting).” We used this question to flag the students if they carried a gun for at least 1 day in the past 12 months.

In November 1998, the Federal Bureau of Investigation launched the NICS. We used 1998 as the reference point and created the binary variable pre- and post-NICS period for YRBS data before and after 1998. Because the YRBS is biennial, data from years 1993–1997 were classified as pre-NICS, and data from years 1999–2017 were classified as post-NICS.

We created a data set of those states that have a requirement for a background check at the point of sale of any firearm. States, including California (1991), Colorado (2013), Connecticut (2013), Delaware (2013), the District of Columbia (1975), Maryland (2010), New York (2013), Oregon (2015), Pennsylvania (2010), Rhode Island (1990), and Washington (2014), have implemented universal background checks (U/BCs) either by requiring background checks for all gun sales conducted by licensed sellers only or by requiring licensed gun sellers, in addition to private sellers, to conduct background checks on all prospective buyers.¹⁶  Eight states, Hawaii (2013), Illinois (2013), Massachusetts (2006), New Jersey (2011), Iowa (2011), Michigan (2006), Nebraska (2010), and North Carolina (2014), implemented firearm background check requirements on private sales primarily by prohibiting private sellers to sell to buyers who did not have a requisite state license or permit and by requiring a background check before issuing the license or permit. Two of these states, Connecticut and New York, require both U/BCs and state permits to purchase firearms.

The YRBS responses were dichotomized into 2 groups: (1) respondents in states without any background check laws and (2) respondents in states with some background check laws at the point of sale or in states that require a license or permit after a background check, which was classified as U/BC. A state was identified as a U/BC state starting from the year when the law was enacted in that state.

Respondent-level variables included age, sex, race and ethnicity, state of residence, and whether the student reported any threat. The “threat” variable was obtained from the YRBS question “During the past 12 months, how many times has someone threatened or injured you with a weapon, such as a gun, knife, or club, on school property?” Studies have revealed that these variables are associated with the likelihood of gun carrying among adolescents.^19,20  We also collected state-level data on annual estimates of median income in current and 2016 Consumer Price Index adjusted dollars and data on the percentage of the total population in urban areas from the decennial US Census.^21,22 

Univariate, bivariate, and multivariable analyses were performed in Stata/SE version 14.2 (Stata Corp, College Station, TX) by using the svy command to account for the complex sample design of the national YRBS. All estimates presented in the results are weighted estimates. Weighted estimates, with appropriate accounting of the survey design, reflect the national population estimate. α was set at .05. t tests and χ² tests were used to test bivariate associations. National estimates of the percentage of high school students carrying guns is reported as well as the percentage of students carrying guns by each study variable.

The longitudinal nature of the data set allowed us to examine the effect of U/BCs before and after implementation of the NICS given the availability of responses from states that had U/BC laws in place before the NICS was implemented. To address our study objective, a longitudinal multivariable logistic regression model was used to examine the effects of the NICS and state U/BCs while controlling for age, race, ethnicity, sex, feeling threatened, and state of residence. It also controlled for time-varying effects, including implementation of the NICS, state U/BC policies, and annual state characteristics, which includes median income and the percentage of the population living in rural areas. It was used to examine the effect of the pre- and post-NICS period on the U/BC variable by using an interaction term. The interaction term between U/BC state and pre- and post-NICS period was used to examine the simultaneous effect of the NICS and state-specific U/BC laws. A postestimation command in Stata (adjrr) was used to estimate the adjusted relative risks (ARRs) for each variable in the model. Subpopulation analyses within the svy command were done to examine the effect of U/BCs on the observations before and after 1998.

A sensitivity analysis used to examine nonfirearm weapon-carrying behavior, such as knives, was conducted to determine if laws targeting firearm purchases specifically affected gun carrying. In the sensitivity analysis, we compared students who reported carrying a weapon (but did not report carrying a gun) with those who did not carry a weapon.

On average, 5.8% of high school students in the United States carried guns across the entire study period. Of those who carried guns, ∼17% were from the states that had U/BCs at the point of gun sales, whereas 83% were from states that did not have U/BCs (P < .001). Adolescents who carried guns were older (16–18 years; P < .001), male (P < .001), and white (P < .001). Students who were threatened reported carrying guns more than those who were not (28% vs 6%; P < .001) (Table 1).

On average, we found a decline in gun carrying until 1999. Since 1999, the trend begins to plateau over time (Fig 1). When comparing states with and without U/BCs, we observe that before NICS implementation, there was no difference in adolescent gun-carrying rates between U/BC and non-U/BC states. However, a difference emerges and continues after 1999, with lower proportions of students reporting gun carrying in U/BC states than in non-U/BC states throughout the remaining study period (Fig 2).

The results of the pooled and stratified weighted multivariable logistic regression model are presented as ARRs of gun carrying by high school students. During the estimation process, we controlled for age, sex, race, ethnicity, threat, state-level median income, and state percentage of the population living in rural areas. The pooled model includes the interaction between the NICS and U/BC. The pooled model reveals that the interaction term is significant, indicating that the NICS with U/BC reduced the risk of gun carrying by 25% (ARR = 0.75 [95% confidence interval [CI]: 0.566–0.995]; P = .046). There was no significant effect of U/BCs during the pre-NICS period whereas, during the post-NICS period, adolescents in states with U/BCs had a 15% reduction in the risk of carrying guns (ARR = 0.85 [95% CI: 0.738–0.984]; P = .029). Boys, African Americans, and those who received threats had a significantly higher risk of carrying guns. In the pooled analysis, ethnicity was not associated with the risk of gun carrying. However, Hispanic students, compared with non-Hispanic students, in the pre-NICS model had a higher risk of gun carrying (ARR = 1.32 [95% CI: 1.03–1.69]; P = .028), whereas they had lower risk in the post-NICS period (ARR = 0.80 [95% CI: 0.72–0.89]; P < .001). Older students in the pre- (12%; ARR = 1.12; P < .001) and post-NICS (20%; ARR = 1.20; P < .001) periods had an increased risk of carrying guns compared with younger adolescents. For each percent increase in the rural population in the post-NICS period, the risk of gun carrying increased by 1.3% (P < .001). Median income had no effect on gun carrying in the post-NICS period (P = .464) (Table 2).

From the sensitivity analysis, we observed that U/BCs had no effect (P = .090) in the control condition (nonfirearm weapon carrying), helping to rule out the possibility of spurious associations between background checks and adolescent gun carrying (Table 3).

Authors of several studies have investigated the effects of both federal and state background check laws on the adult population.^23,24  Our study of adolescents both supports and differs from studies in adults, indicating that more work needs to be done to understand the downstream effects on minors of gun-purchasing laws aimed at adults. The authors of 1 study reported that the Brady Act’s impact on firearm injury and overall homicide and suicide rates only significantly reduced suicide rates in those aged >55 years.¹¹  In 2001, a study revealed that California’s background check law reduced violent misdemeanants’ subsequent arrests for gun possession and violent crimes.²⁵  This suggests that increasing background checks for gun purchases may potentially reduce violent crimes and criminal charges because firearms may be less accessible, and carrying a deadly weapon substantially alters the severity of a criminal charge.^26,27  The authors of another study examined the association of various state background check laws with firearm homicides, finding that stricter background check regulations were associated with lower firearm homicide rates.²⁸  Recently, evidence that suicide-prone populations residing in states with less strict gun laws were at an increased risk of a completed suicide by firearm.²⁹  In light of increasing rates of suicide and suicide attempts in adolescents, it may be important to evaluate the efficacy of background checks in reducing unintended access to firearms because gunshots are the most fatal means of completing suicide.^30,31  Findings from the above studies suggest that federal and state background checks laws can reduce gun-related crimes and firearm deaths in the adult population as well as reduce adverse outcomes of firearm ownership.^29,32,33 

In our study, we did not find evidence revealing that implementation of the NICS in 1998 was independently associated with a national reduction in adolescents’ gun carrying when controlling for individual, state, and time effects. We found that adolescents living in states that required U/BCs on all prospective gun buyers were less likely to carry guns compared with adolescents living in states that only require background checks on purchases through federally licensed gun dealers. This finding is similar to that of other studies that found that state-specific gun laws reduced youth gun carrying.¹⁷  However, our results also indicate that states’ U/BC laws were not effective before NICS implementation. This suggests that the NICS may be more effective in reducing adolescent gun carrying if all gun buyers were required to have a background check. When we examined the interaction between the NICS and state U/BC laws, we found that together they significantly reduced the risk of adolescent student gun carrying by 25%. Our results may reflect several factors. First, it is possible that adolescents who purchase guns for themselves may be more likely to purchase from a private seller, particularly if they do not meet minimum age requirements. Requiring all gun sales to be made through licensed dealers, who either require a background check or a gun permit issued only after a background check, could deter gun purchases by adolescents. It is also possible that adults who would sell or allow minors access to a firearm would be less likely to be approved for a firearm purchase, thus giving them fewer options to purchase guns when residing in areas that require background checks on all buyers.

It is important to note that factors in addition to background check laws on adult gun purchases likely play a larger role in adolescent gun carrying. Adolescents often obtain firearms from their own home, purchased legally by adults who may not always secure weapons (either within a safe or using another locking mechanism).³⁴  A cross-sectional survey of adolescents attending New York City high schools on their perception of firearms revealed that 41% of students lived with an adult who possessed a firearm.³⁴  These students were more likely to be found in residences with gun-supporting families that witnessed gun violence at some point in their lives. Thus, gun-safety storage programs may be beneficial to both rural- and urban-dwelling families, particularly those with high school youth, because researchers found these programs to be helpful across populations.³⁵  Storing firearms securely reduces the likelihood that a minor residing in the home would obtain it without an adult’s knowledge as well as prevents gun theft, which is an important source of guns used in crimes or possibly sold in illegal markets to minors.³⁶ 

With our study, we add to the literature by examining how federal and state policy changes over a 24-year period, approximately a generation, interact to affect adolescent gun carrying. Specifically, we found that state laws moderate the effect of federal background check laws, which may suggest that implementing U/BCs at the national level could increase efficacy of the NICS. The NICS background check system is limited because of the potential to sidestep the system through private sales. Implementing U/BCs nationally may decrease the number of guns accessible to adolescents and, in turn, reduce their gun carrying. On the basis of studies in adults, this could also indirectly prevent firearm-related suicides, homicides, and injuries as well as reduce the likelihood of being charged with a felony that could affect employment opportunities throughout life.^33,37–40  Strengthening background check policies and making safety training available to all gun owners on proper storage of firearms may decrease the number of firearms acquired by adolescents, preventing injuries that are costly both financially and to quality of life.^41–43 

Our analyses were based on cross-sectional data of adolescents sampled in different years, and therefore conclusions about association, rather than causality, between background check laws and adolescent gun carrying can only be made. We attempted to address this using longitudinal analyses that allowed for time-varying effects, such as state U/BC laws and annual changes in state-level variables, to account for latent changes in gun carrying nationally. Another limitation is that the laws examined do not directly apply to adolescents. It is likely that the effects of these laws are mediated by adult behavior, and whether this changed the availability of firearms in the respondent’s household is unknown. Although a strength of the YRBS is that it is a long-running, nationally representative data source, information on other important outcomes is limited, and we only examined gun carrying. Studying outcomes such as gun use or gun-related injury would provide additional public health insights. The YRBS was also missing outcome data on 4.1% of the sample. However, the weighted analysis used in the study is 1 approach to produce unbiased estimates of a population. Another limitation is that data are self-reported, and adolescents not enrolled in school are excluded, likely making our estimates of adolescent gun carrying low. Finally, experiences outside of school, such as bullying (eg, in person, through text messaging, or in cyber or social media settings), anxiety around mass shootings, and other factors that might influence gun carrying, could not be accounted for. We attempted to control for this using the respondent’s self-reported variable of being threatened or injured on school property, but we do not have data on whether adolescents experienced threats outside of school or harassment online, partially because the early years of these data predate widespread Internet use.

Our findings suggest that federal systems for conducting background checks do not independently reduce adolescent gun carrying on a national level. This may be because the NICS effect was only significant in the presence of state U/BC laws. Additionally, U/BC laws did not independently affect adolescent gun carrying before implementation of the NICS. Therefore, it is possible that both federal and state background check laws work together to reduce gun carrying in high school students.

ARR

adjusted relative risk

CI

confidence interval

NICS

National Instant Criminal Background Check System

U/BC

universal background check

YRBS

Youth Risk Behavior Survey