Predictors of Weapon-Related Behaviors Among African-American, Latino, and White Youth

Abstract

Objective

To identify risk and protective factors for weapon involvement among African-American, Latino, and white adolescents.

Study design

The National Longitudinal Study of Adolescent to Adult Health is a nationally-representative survey of 7^th–12^th grade students. Predictors at Wave 1 and outcome at Wave 2 were analyzed. Data were collected in the mid-1990s, when rates of violent crime had been declining. The outcome was a dichotomous measure of weapon-involvement in the past year, created using 3 items (weapon-carrying, pulled gun/knife, shot/stabbed someone). Bivariate and multilevel logistic regression analyses examined associations of individual, peer, family, and community characteristics with weapon involvement; stratified analyses were conducted with African-American, Latino, and white subsamples.

Results

Emotional distress and substance use were risk factors for all groups. Violence exposure and peer delinquency were risk factors for whites and African Americans. Gun availability in the home was associated with weapon involvement for African Americans only. High educational aspirations were protective for African Americans and Latinos, but higher family connectedness was protective for Latinos only.

Conclusions

Interventions to prevent weapon-related behaviors among African American, Latino, and white adolescents may benefit from addressing emotional distress and substance use. Risk and protective factors vary by race/ethnicity after adjusting for individual, peer, family, and community characteristics. Addressing violence exposure, minimizing the influence of delinquent peers, promoting educational aspirations, and enhancing family connectedness could guide tailoring of violence prevention interventions.

Weapon-related violence among adolescents leads to physical and mental health problems, and results in substantial costs due to injuries and loss of productivity.¹ Almost 13% of high school students have been victimized with weapons.² Weapon-carrying is highest among whites (21%), followed by Latinos (16%), and African Americans (13%).³ Risk factors increase the odds of engaging in weapon-related behaviors,⁴ and include male sex, ^5–7 drug and alcohol use,^5–7 violent victimization,⁵ and emotional distress.⁸ Peer and community risk factors include peer delinquency,⁸ weapon availability,⁹ unsafe neighborhoods, community violence, and community economic disadvantage.^5,7 Protective factors include closeness to parents,⁶ and high educational aspirations.⁸

There may be racial/ethnic differences in risk and protective factors for weapon involvement. The influence of substance use,¹⁰ peer delinquency,¹⁰ academic achievement,⁵ neighborhood safety,⁷ and community economic disadvantage⁹ on weapon-related behaviors may vary by race/ethnicity. Family connectedness and educational aspirations also may differ by race/ethnicity,^9,11 with some studies demonstrating a greater protective influence of family connectedness for Latinos.¹² Racial/ethnic disparities may reflect socioeconomic disparities, therefore, it is particularly important to adjust for socioeconomic status in examining racial/ethnic differences.¹³ The purpose of the this study was to identify risk and protective factors for involvement in weapon-related behaviors among a nationally-representative sample of African American, Latino, and white adolescents, after adjusting for socioeconomic status. Our hypothesis was that the predictors of weapon involvement would differ by race/ethnicity.

METHODS

The National Longitudinal Study of Adolescent to Adult Health (Add Health) consists of surveys of a nationally-representative sample of 7^th–12^th grade students followed from adolescence into adulthood.¹⁴ Add Health used a school-based clustered sampling design, with over-sampling of Cubans and Puerto Ricans. Wave 1, conducted in 1994–1995, consisted of 20,745 surveys with adolescents in 7^th–12^th grades and 17,670 surveys with their parents.¹⁴ Wave 2, conducted in 1996, consisted of 14,738 surveys with adolescent participants from Wave 1. Surveys were conducted via face-to-face interviews with respondents in their homes. In order to protect adolescents’ confidentiality, data were entered into laptop computers by interviewers,; for sensitive topics, respondents listened to questions through earphones and entered their answers directly into the laptops.¹⁴ Predictors at Wave 1 and the outcome at Wave 2 were analyzed. The study was approved by the UT Southwestern Medical Center Institutional Review Board. Add Health was selected over other datasets (eg Youth Risk Behavior Survey, the National Crime Victimization Survey, and Monitoring the Future Survey) due to its longitudinal nature, and relatively high samples of racial/ethnic minorities. It was selected over other longitudinal studies (e.g. Pittsburgh Youth Study, Denver Youth Survey, and Rochester Youth Development Study) due to its national representativeness. Although survey data are from the 1990s, Add Health provides a unique opportunity to examine racial/ethnic similarities and differences in predictors of adolescent weapon involvement over time. It is important to note that rates of violent crime began declining in the early 1990s, presumed to be primarily due to policies related to policing and incarceration.¹⁵ Weapon-carrying continued to decline until 1999, but there has been no significant decline since.¹⁶

Outcome

A dichotomous variable for weapon involvement was created using three items from Wave 2. Adolescents self-reported about frequency of carrying a weapon, such as a gun, knife, or club, to school during the past 30 days, and frequency of pulling a knife or gun on someone, or shooting or stabbing someone, in the past 12 months. Respondents who reported involvement in any of these three behaviors one or more times were categorized as involved in weapon-related behaviors, whereas those who reported that they were not involved in any of these behaviors were categorized as uninvolved. These variables from Add Health have previously been used to assess engagement in severe violence.¹²

Independent variables

Wave 1 predictors (Table I) were chosen based on demonstrated association with youth violence and repeated weapon-related behaviors,^5–8,12,17 and on validated measures previously used in Add Health analyses.^4,8 Sociodemographic variables included age, sex, family structure, urban residence,^5–7 and immigrant generational status.¹² Race/ethnicity was categorized as Latino, non-Latino white, African American, Asian/Pacific Islander, American Indian/Alaska Native, or Other racial/ethnic group.⁸ Parental receipt of public assistance was used as a measure of socioeconomic status.⁸

Table 1.

Selected independent variables and measures

Variable	Measure	Variable Response Categories
Sociodemographic Factors
Family structure	Two biological parents; two parents (one or both non- biological); single mother; single father; other	Two-parents (biological/non-biological) vs. single parent or other (mother/father/other family structure)
Socioeconomic status	Mother receives public assistance Father receives public assistance	One/both parents receive assistance vs. no parents receive assistance
Immigrant generation	Were you (child) born in the US? Was your mother born in the US? Was your father born in the US?	1st gen: Foreign-born child and parents 2nd gen: US-born child and one/both foreign-born parents 3rd gen plus: US-born child and parents
Individual Risk Factors
Violence exposure	In past 12 months, how often did you see someone shoot/stab another person?	Continuous; re-coded to: One or more times vs. never
Emotional distress	11 items: How often was each of the following true in the past week? (e.g. felt blue, fearful, lonely, sad)	Never/rarely, Sometimes, A lot of the time, Most/all of the time Mean score on 11 items calculated. Higher scores indicate higher distress.
Peer, Family, and Community-Related Risk Factors
Teen social isolation the following with your friend? (e.g. go to friend’s house, talk to friend about a problem)	5 items: In past 7 days, did you do	Yes/No Questions regarding up to five male and five female friends. Higher scores indicate lower isolation.
Peer delinquency	Of your 3 best friends, how many use marijuana, alcohol, smoke?	Zero, one, two, three Mean of the three items computed
Community economic disadvantage	Census-tract level measures (e.g. unemployment rate, median household income)	Principal component score computed from 7 items
Protective Factors
Educational aspirations	On a scale of 1 to 5 how much do you want to go to college?	1–5 (1 is low, 5 is high)
Family connectedness	How much do you feel that your family understands you, pays attention to you; parents care about you; you have fun with family?	Not at all, Very little, Somewhat, Quite a bit, Very much Mean of four items calculated. Higher scores indicate higher connectedness.

Wave 1 weapon involvement, constructed identically to the dependent variable in Wave 2, was included as a control variable. Individual child risk factors included youth self-reports of prior violence exposure,¹⁷ alcohol use in the past 12 months, lifetime drug use,¹⁸ and emotional distress.⁸ Peer level risk factors included social isolation⁸ and peer delinquency.⁸ Gun availability in the home¹⁷ and community economic disadvantage¹⁷ also were examined.

Youth self-reports of educational aspirations,⁸ family connectedness,⁸ and neighborhood safety were examined as potential protective factors.

Statistical Analyses

Add Health sample weights were used to obtain national estimates and account for clustering. Missing data across independent variables were computed using five-dataset multiple imputation.^19,20 Analyses were conducted using SAS 9.2. Bivariate analyses were performed using t-tests and chi-square tests. Multilevel logistic regression analysis was used to examine associations between risk and protective factors and weapon involvement. All independent variables that were associated with the outcome in bivariate analyses were included in the multivariable model. Separate stratified analyses, using the same model, were conducted for whites, African-Americans, and Latinos to examine group similarities and differences. Analyses of other racial/ethnic groups are not presented due to small sample sizes. Survey weights were used to calculate adjusted odds ratios and 95% confidence intervals for factors associated with weapon involvement. Stratified stepwise multivariable analyses²⁰ resulted in similar findings and therefore are not presented.

RESULTS

Weapon involvement differed significantly by racial/ethnic group (Table II), with 7% of whites, 13.5% of African-Americans, and 10.4% of Latinos reporting involvement. Of those who carried weapons, 17% had also shot/stabbed someone in the past 12 months.

Table 2.

Selected characteristics for US adolescents in 7^th–12^th grades in 1994–1995^*

Characteristic	Mean (S.E.) or Proportion (%)
	Full sample (N=13503)	White (N=7447)	African American (N=2741)	Latino (N=2277)
Outcome
Weapon involvement (%)	8.5	7ab	13.5c	10.4c
Carried weapon (%)	5.6	4.7ab	8.1c	7.1c
Pulled knife/gun on someone (%)	4.6	3.3ab	8.8c	6.8c
Shot/stabbed someone (%)	1.8	1.2ab	3.3c	3.0c
Sociodemographic Factors
Mean age, years (SE)	15 (0.1)	15 (0.1)	15 (0.1)	15 (0.2)
Male sex (%)	50	50	51	51
Two-parent household (%)	75	80ab	48bc	74ac
Receive public assistance (%)	11	8ab	20c	19c
Urban residence (%)	52	44b	57b	84ac
Immigrant generation (%)
1st generation	4.7	0.4b	0.3b	20ac
2nd generation	41.1	33ab	63.3c	57.3c
3rd generation	54.2	66.6ab	36.4bc	22.7ac
Individual Risk Factors
Wave 1 weapon involvement (%)	9.3	7.5ab	14.8c	12.7c
Violence exposure (%)	11	7ab	22c	19c
Emotional distress (SE)	0.6 (0.01)	0.5 (0.01)ab	0.6 (0.01)c	0.6 (0.02)c
Alcohol use (%)	53	55.9a	45.9bc	54.5a
Drug use (%)	28	29	24	29
Peer, Family, and Community-Related Risk Factors
Teen social isolation (SE)	1.5 (0.01)	1.5 (0.02)ab	1.4 (0.02)c	1.4 (0.02)c
Peer delinquency (SE)	0.8 (0.03)	0.9 (0.04)a	0.7 (0.04)c	0.8 (0.04)
Guns available in home (%)	24	29ab	14c	11c
Community economic disadvantage (SE)	0.02 (0.08)	−0.2 (0.01)ab	0.7 (0.01)bc	0.3 (0.08)ac
Protective Factors
Educational aspirations (SE)	4.4 (0.03)	4.4 (0.03)b	4.5 (0.04)b	4.3 (0.05)ac
Family connectedness (SE)	4 (0.01)	4.0 (0.02)	4.1 (0.03)	4.1 (0.03)
Neighborhood safety (%)	89	92ab	82c	82c

^*Letters indicate significant differences among racial/ethnic groups at P<.05:

^avs African Americans;

^bvs Latinos;

^cvs whites. SE, standard error.

Risk Factors

In bivariate analyses for the full sample, African Americans had the highest unadjusted odds associated with the outcome (OR 1.9; 95% CI, 1.6–2.3), followed by Latinos (OR 1.3; 95% CI, 1.01–1.7). All variables analyzed as potential risk factors demonstrated odds ratios greater than one in bivariate analyses, and therefore were associated with higher odds of wave 2 weapon involvement. In multivariable analyses (Table III), white youth who were weapon-involved in wave 1 had more than five times the odds of weapon involvement at wave 2; African Americans had more than four times the odds and Latinos more than six times. Emotional distress was associated with higher odds of weapon involvement across racial/ethnic groups. Violence exposure, alcohol use, and peer delinquency were risk factors for whites and African Americans. Community factors were not associated with weapon involvement.

Table 3.

Factors associated with weapon involvement among full sample, white, African American, and Latino adolescents^a,^b

Characteristic	Adjusted Odds Ratios (95% Confidence Intervals) of Weapon Involvement
	Full sample (n=11207 )	White (n=6198 )	African American (n=2368 )	Latino (n=2048)
Sociodemographic Factors
Male sex	3.3 (2.6,4.2)	4.4 (3.0,6.3)	1.7 (1.2,4.4)	3.6 (2.0,6.3)
Mean age, years	0.8 (0.7,0.9)	0.8 (0.7,0.9)	0.9 (0.8,0.9)	0.8 (0.7,0.9)
Race/ethnicity
White	Reference	--	--	--
African-American	1.5 (1.2,2.1)	--	--	--
Latino	1.2 (0.9,1.8)	--	--	--
Asian/Pacific Islander	0.5 (0.1,4.2)	--	--	--
American Indian/Alaska Native	1.7 (0.7,3.7)	--	--	--
Other	1.2 (0.2,9.0)	--	--	--
Two-parent household	0.8 (0.4,1.4)	0.9 (0.7,1.4)	0.9 (0.6,1.4)	0.7 (0.6,1.3)
Public assistance	1.0 (0.6,1.8)	1.0 (0.9,1.6)	1.0 (0.7,1.6)	0.6 (0.3,1.1)
Immigrant generation
1st generation	0.5 (0.3,1.0)	0.3 (0.3,5.9)	0.5 (0.2,6.7)	0.2 (0.1,0.7)
2nd generation	1.4 (1.2,1.7)	1.2 (0.9,1.5)	1.3 (0.9,1.7)	1.5 (0.9,2.5)
3rd generation	Reference	Reference	Reference	Reference
Individual Risk Factors
Wave 1 weapon involvement	5.4 (4.2,6.9)	5.6 (4.0,7.9)	4.2 (2.7,6.6)	6.1 (3.2,11.8)
Violence exposure	1.9 (1.5,2.3)	2.0 (1.4,2.8)	2.0 (1.4,2.8)	1.5 (0.8,2.8)
Emotional distress	1.5 (1.2,1.9)	1.4 (1.02,2.0)	1.7 (1.1,2.6)	2.0 (1.1,3.8)
Alcohol use	1.4 (1.2,1.8)	1.3 (1.0,1.8)	1.4 (1.0,2.0)	1.6 (0.9,2.6)
Drug use	1.2 (0.9,1.5)	1.1 (0.8,1.5)	1.1 (0.7,1.7)	2.0 (1.2,3.3)
Peer, Family, and Community-Related Risk Factors
Peer delinquency	1.2 (1.1,1.4)	1.3 (1.2,1.6)	1.5 (1.2,1.7)	0.9 (0.7,1.2)
Guns available in home	1.2 (0.9,1.4)	1.1 (0.8,1.4)	1.8 (1.2,2.5)	0.7 (0.3,1.5)
Social isolation	0.9 (0.8,1.2)	1.1 (0.9,1.5)	1.0 (0.8,1.3)	0.7 (0.4,1.2)
Community economic disadvantage	1.1 (0.9,1.2)	1.1 (0.8,1.3)	1.1 (0.9,1.2)	1.3 (0.9,1.7)
Protective Factors
Family connectedness	0.8 (0.7,0.9)	0.9 (0.7,1.1)	0.9 (0.6,1.1)	0.7 (0.6,0.9)
Educational aspirations	0.9 (0.8,0.9)	0.9 (0.8,1.1)	0.8 (0.7,0.9)	0.8 (0.7,0.9)
Neighborhood safety	0.9 (0.7,1.2)	0.9 (0.6,1.6)	1.1 (0.7,1.7)	0.7 (0.4,1.1)

^aStatistically significant values are in highlighted in bold.

^bAnalyses adjusted for all variables shown in table.

Protective Factors

In bivariate analyses, family connectedness and educational aspirations demonstrated odds ratios less than one for all groups, and therefore were associated with lower odds of wave 2 weapon involvement, as would be expected for protective factors. In multivariable analyses, high educational aspirations were protective for both African Americans and Latinos. Family connectedness was protective only for Latinos (OR 0.7; 95% CI, 0.6–0.9), indicating that a one point increase on the family connectedness scale was associated with a 30% lower odds of weapon involvement for Latinos.

DISCUSSION

This study examined risk and protective factors for weapon-related behaviors among white, African American, and Latino adolescents. Study findings indicate that some predictors of weapon involvement differ between racial/ethnic groups even after adjusting for socioeconomic status and multiple individual, peer, family, and community characteristics. Emotional distress emerged as a risk factor for all groups, and peer and family influences differed across groups.

Several individual level risk factors were associated with weapon involvement in our study. Violence exposure was associated with higher odds of weapon involvement among whites and African Americans. Among those who have been exposed to violence, weapon-carrying may be in self-defense, due to fear of violence,^21,22 or may be associated with delinquency²³ and aggression.²⁴ The relationship between violence exposure and weapon involvement also may be mediated by mental health. Violence exposure is associated with anxiety, depression, and posttraumatic stress disorder,²⁵ and suicidal ideation is associated with weapon-carrying.⁶ Given that, in our study, emotional distress was associated with weapon involvement for all racial/ethnic groups, it is important to further study the mechanisms underlying this relationship. Interventions to address weapon-related behaviors need to address adolescent mental health for all racial/ethnic groups, in addition to reducing violence exposure.

Baseline weapon involvement and substance use were risk factors for weapon involvement for all racial/ethnic groups. These findings are consistent with Jessor’s Problem Behavior Theory which suggests that high-risk behaviors, such as alcohol and drug use, and violence co-occur among some adolescents.²⁶ Prior studies also have shown that substance use is associated with increased weapon involvement.^6,7,27 It may be beneficial to screen for and address substance use in violence prevention interventions.

Our findings suggest that the influence of peer and family factors on weapon involvement may differ by race/ethnicity. Peer delinquency increased risk for whites and African Americans, but not for Latinos. Conversely, family connectedness was protective for Latinos but not statistically significant for whites or African Americans. Strong relationships with parents can be protective against weapon violence²⁸ and some Latino families may be more family-oriented compared with whites.²⁹ There is conflicting research on the relationship between family factors and youth violence for Latinos and African Americans. Family cohesion is more strongly associated with lower aggression among Latinos compared with African Americans,²⁹ but higher family-centered beliefs are associated with more violence among Latinos and with less violence among African Americans.³⁰ In our study, the level of family connectedness did not vary across groups, but the association with weapon involvement did, suggesting there are ethnic differences in the relationship between family connectedness and weapon involvement. Promoting family connectedness may be a particularly important strategy for preventing youth violence.

Another protective factor for Latinos and African Americans is having high educational aspirations. Educational aspirations may be a component of future orientation, and are associated with decreased violence involvement and drug use.³¹ There are racial/ethnic disparities in educational aspirations and achievement.³² Educational aspirations of Latinos were found to be lower compared with whites,³² whereas those of African Americans were similar.^12,32 Promoting adolescents’ educational aspirations may protect against weapon involvement. It is necessary to also provide youth with the opportunities and resources to achieve these aspirations, while considering the unique barriers and challenges faced by youth from different cultural groups. Some variables potentially related to the outcome were not examined because they were not available in the dataset (e.g. family violence exposure) or were highly correlated with others in the model (e.g. parental education). Socioeconomic status was measured as receipt of public assistance, and family structure was categorized as two-parent vs. other, which may not capture all dimensions of these constructs. Data on students who were not in school or who were incarcerated, were unavailable, therefore, this study may underestimate the prevalence of weapon involvement. Data were collected by youth and parent self-report, which may be subject to bias, however, the use of anonymous data collection, with data for sensitive topics entered directly into laptop computers by respondents, minimizes bias.³³ Data were collected in the mid-1990s. Rates of weapon-carrying in our study (4.6%) were comparable with 2013 rates (5.2%).³ Despite these limitations, the large sample size, with over-sampling of Latino subgroups, allowed the examination of a serious, but less frequent type of violence, and of racial/ethnic differences. Although studies suggest there may be Latino subgroup differences in violence involvement,¹⁸ Latinos in this study were analyzed as one group, as sample size did not allow subgroup analysis. Future studies must examine Latino sub-group differences.

The effectiveness of current youth violence prevention interventions varies,^34,35 with one review suggesting smaller effect sizes for racial/ethnic minorities.³⁵ Findings from our study suggest that interventions to prevent weapon-related behaviors among white, African American, and Latino adolescents should address emotional distress and substance use. It is important to also mitigate the effects of violence exposure, promote educational aspirations, minimize the influence of delinquent peer groups, and focus on family connectedness to appropriately tailor programs for different racial/ethnic groups.