Adolescent Bullying and Weapon Carrying: A Longitudinal Investigation

Abstract

Using four data waves from an ongoing longitudinal study, we examined the longitudinal association between weapon carrying and bullying (i.e., bully, victim, bully-victim). Participants were 1,042 adolescents (55.9% female) recruited from Texas high schools, including Hispanic (31.7%), White (30.3%), African American (26.6%), and “other” (11.4%) with a mean age of 15.1 years at baseline. Multivariate logistic regression revealed that only baseline bully-victim was linked to weapon carrying and only one year later (Adjusted OR=1.76, 95% CI: 1.08, 2.85), after controlling for baseline weapon carrying, age, gender, race, and parent education. No significant longitudinal association was found with bully or victim. Findings emphasize the importance of interventions targeting both bullying perpetrators and victims, while also highlight the necessity of longitudinal research.

Weapon carrying, especially on school property, is a significant problem among youth. The 2016 Indicators of School Crime and Safety suggests that 6% of high school students report being physically hurt or threatened with a weapon, such as a gun or knife, while on school premises (Musu-Gillette, Zhang, Wang, Zhang, & Oudekerk, 2017). Moreover, adolescent weapon carrying is associated with other violent behaviors such as involvement in fights (Forrest, Zychowski, Stuhldreher, & Ryan, 2000) and gang membership (Bradshaw, Waasdorp, Goldweber, & Johnson, 2013), and is linked to substance abuse (McKeganey & Norrie, 2000) and emotional distress (Walsh et al., 2013). It is therefore important to investigate factors that contribute to adolescent weapon carrying behaviors.

Bullying, defined as repeated aggressive behavior intended to cause harm or distress to victims based on a power imbalance (Olweus, 2013), is identified as one possible explaining factor (Baly, Cornell, & Lovegrove, 2014). In a nationally representative sample of youth ages 12–18, those who experienced bullying in school were more likely to have access to a firearm at home and were more likely to obtain a weapon without permission (Simckes, Simonetti, Moreno, Rivara, Oudekerk, & Rowhani-Rahbar, 2017). Weapons may be used as a tool for bullying, potentially to intimidate others (Dijkstra, Gest, Lindenberg, Veenstra, & Cillessen, 2012). Conversely, victims of bullying may carry weapons to feel safe or for self-protection. Bully-victims (those who bully and are victims of bullying) are potentially at even higher risk of weapon carrying because they share the characteristics of the bully and the victim (Van Geel, Vedder, & Tanilon, 2014). Indeed, a recent meta-analysis synthesizing 35 studies (Valdebenito, Ttofi, Eisner, & Gaffney, 2017) revealed that, cross-sectionally, adolescents involved in bullying – whether as bully, victim, or bully-victim – had higher odds of carrying a weapon compared to their non-involved counterparts.

Despite the well-established cross-sectional link between bullying and weapon carrying, little is known about the longitudinal associations. Among one of the few studies using longitudinal data, Baly and colleagues (2014) found that cumulative bullying victimization in grades 6, 7, and 8 was associated with weapon carrying at grade 8. Turner et al. (2015) also found that repeated victimization (i.e., victimized during childhood and adolescence) was associated with gun carrying in adulthood. In another study, Hemphill et al. (2011) found that bullying perpetration and victimization were both associated with weapon carrying longitudinally, however, after adjusting for covariates such as family and peer factors, these associations became non-significant. Overall, although a limited number of longitudinal studies have examined the associations between weapon carrying and bullying victimization, research on bullying perpetration is scarce and the longitudinal association with bully-victimization remains completely unknown. Furthermore, as Valdebenito and colleagues (2017) suggest, given an overall lack of longitudinal research, conclusions cannot be drawn and further investigations are needed to examine the causal link between bullying and weapon carrying as well as to establish the time sequence of the relationship (i.e., if bullying leads to weapon carrying or vice versa). To address this knowledge gap, we utilize four years of data to examine the longitudinal link between bullying (i.e., bully, victim, bully-victim) and weapon carrying.

Methods

Participants and Procedure

Data are from an ongoing longitudinal study of adolescent health (Temple, Shorey, Fite, Stuart, & Le, 2013), in which 1,042 adolescents were recruited from seven public high schools in southeast Texas to participate in annual surveys. Participants who returned signed parental consent forms and gave assent completed paper-and-pencil surveys in class. At later assessments, participants who no longer attended the recruitment schools completed the survey online.

At baseline, participants self-identified as Hispanic (31.7%), White (30.3%), African American (26.6%), Asian/Pacific Islander (3.4%), and other (8%) with a mean age of 15.1 years (SD = .79). We used data from baseline (W1, spring 2010, 55.9% female), Wave 2 (W2, spring 2011, n = 964, 55.9% female), Wave 3 (W3, spring 2012, n = 894, 55.8% female), and Wave 4 (W4, spring 2013, n = 776, 57.6% female). Study procedures were approved by the last author’s Institutional Review Board.

Measures

For weapon carrying, participants indicated yes/no to the question “In the past year, did you carry a weapon such as a gun, knife, or club?”. For bullying, after reading the definition and an example of bullying, participants responded on a scale of 1 (never) to 4 (many times) to two questions, “how often have you been bullied in the past 12 months?” and “how often have you bullied other teens in the past 12 months”. Two dichotomous items (yes/no) were created based on these two questions to represent if a participant had bullied others (“bully”) or been bullied (“victim”). Participants who endorsed both of the above items were classified as “bully-victims”. Demographic information including age, gender, race, and parent education were collected at W1.

Data Analysis

Preliminary analyses were performed in SPSS 24 to examine variable frequencies and Pearson correlations among variables. Next, a series of logistic regressions were conducted in Mplus 8 (Muthén & Muthén, 1998–2017) to test the longitudinal associations with bully (W1), victim (W1), and bully-victim (W1) predicting weapon carrying one year later (W2), two years later (W3), and three years later (W4), controlling for age, gender, race, parent education, and W1 weapon carrying.¹ Standard errors of school clusters were adjusted by including six dummy-coded school variables. Robust maximum likelihood estimation method was used. Missing data were handled with full information maximum likelihood (Graham, 2012).

Results

Frequencies and correlations among variables are shown in Table 1. Multivariate logistic regressions indicated that bully perpetration and bully victimization were not significantly associated with weapon carrying longitudinally when age, gender, race, parent education, and W1 weapon carrying were controlled for (see Tables 2 and 3). Only being a bully-victim significantly predicted weapon carrying and only at one year later (Adjusted OR = 1.76, 95% CI: 1.08, 2.85), after controlling for age, gender, race, parent education, and W1 weapon carrying (see Table 4). ²

Table 1.

Frequencies and Correlations among Variables

	Frequencies (%)	1	2	3	4	5	6
1. Weapon W1	176 (17.0%)	-
2. Weapon W2	143 (14.9%)	.54**	-
3. Weapon W3	141 (15.8%)	.45**	.56**	-
4. Weapon W4	103 (17.4%)	.36**	.57**	.59**	-
5. Bully W1	374 (36.1%)	.15**	.12**	.12**	.07	-
6. Victim W1	369 (35.6%)	.07*	.10**	.07*	.08	.44**	-
7. Bully-victim W1	237 (22.9%)	.08**	.12**	.10**	.10*	.73**	.73**

Notes.

^*p < .05

^**p < .01.

Table 2.

Logistic Regression Results of Bully Perpetration Predicting Weapon Carrying at Waves 2–4

	Weapon W2 AOR [95% CI]	Weapon W3 AOR [95% CI]	Weapon W4 AOR [95% CI]
Bully perpetration W1	1.17 [.73, 1.89]	1.40 [.88, 2.21]	1.11 [.66, 1.86]
Gender (female as reference)	3.04*** [1.91, 4.85]	2.95*** [1.84, 4.72]	2.47** [1.46, 4.16]
Race (White as reference)
Hispanic	.52 [.25, 1.06]	.33** [.17, .65]	.37* [.16, .82]
Black	.62 [.27, 1.41]	.48* [.23, .99]	.65 [.27, 1.54]
Other	1.07 [.46, 2.50]	.54 [.25, 1.18]	1.53 [.69, 3.38]
Age	.76 [.55, 1.06]	.93 [.68, 1.28]	.83 [.57, 1.22]
Parent Education	.73** [.57, .93]	.64*** [.50, .81]	.68** [.52, .87]
Weapon W1	17.08*** [10.52, 27.71]	11.11*** [6.85, 18.02]	7.57*** [4.19, 13.67]

Notes.

^*p < .05

^***p < .001.

AOR = adjusted odds ratio, CI = confidence interval.

Table 3.

Logistic Regression Results of Bully Victimization Predicting Weapon Carrying at Waves 2–4

	Weapon W2 AOR [95% CI]	Weapon W3 AOR [95% CI]	Weapon W4 AOR [95% CI]
Bully Victimization W1	1.52 [.94, 2.47]	1.26 [.80, 1.98]	1.30 [.78, 2.18]
Gender (male as reference)	3.13*** [1.95, 5.02]	3.00*** [1.88, 4.78]	2.52** [1.49, 4.26]
Race (white as reference)
Hispanic	.53 [.26, 1.08]	.33** [.17, .64]	.37* [.16, .84]
Black	.64 [.28, 1.48]	.48* [.24, .99]	.64 [.27, 1.53]
Other	1.06 [.45, 2.48]	.54 [.25, 1.16]	1.54 [.70, 3.41]
Age	.78 [.56, 1.09]	.94 [.69, 1.28]	.85 [.58, 1.25]
Parent Education	.74* [.58, .94]	.64*** [.50, .81]	.69** [.53, .89]
Weapon W1	17.10*** [10.63, 27.49]	11.38*** [7.08, 18.30]	7.56*** [4.21, 13.58]

Notes.

^*p < .05

^***p < .001.

AOR = adjusted odds ratio, CI = confidence interval.

Table 4.

Logistic Regression Results of Bully-victimization Predicting Weapon Carrying at Waves 2–4

	Weapon W2 AOR [95% CI]	Weapon W3 AOR [95% CI]	Weapon W4 AOR [95% CI]
Bully-victimization W1	1.79* [1.06, 3.00]	1.49 [.90, 2.45]	1.64 [.93, 2.90]
Gender (male as reference)	3.15*** [1.96, 5.05]	3.01*** [1.89, 4.81]	2.56** [1.51, 4.35]
Race (white as reference)
Hispanic	.54 [.27, 1.11]	.34** [.17, .66]	.38* [.17, .85]
Black	.67 [.29, 1.57]	.50 [.24, 1.05]	.68 [.28, 1.65]
Other	1.09 [.46, 2.56]	.55 [.25, 1.18]	1.58 [.72, 3.46]
Age	.77 [.56, 1.08]	.94 [.69, 1.28]	.85 [.58, 1.25]
Parent Education	.73* [.58, .93]	.64*** [.50, .81]	.69** [.53, .89]
Weapon W1	17.08*** [10.61, 27.48]	11.30*** [7.01, 18.21]	7.56*** [4.20, 13.62]

Notes.

^*p < .05

^***p < .001.

AOR = adjusted odds ratio, CI = confidence interval.

Discussion

Using four waves of data on ethnically diverse adolescents, we examined the longitudinal associations between bullying and weapon carrying. Consistent with limited previous research (Hemphill et al., 2011), neither bully perpetration nor bully victimization were found to be significantly associated with weapon carrying after controlling for a series of covariates and prior weapon carrying. Only bully-victims had a higher likelihood of carrying weapons one year later. These findings extend research showing that bully-victims are more likely than others, including victims, to be at heightened risk for negative outcomes – in this case, weapon carriage (Liang, Flisher, & Lombard, 2007; Nansel, Overpeck, Haynie, Ruan, & Scheidt, 2003). Coupled with prior research showing that bully-victims are more likely to report actually using a weapon than victims (Srabstein & Piazza, 2008), our study highlights the potential consequences of bully-victimization and warrants further investigation to explore possible explanations and intervention development.

Notably, our study only identified longitudinal associations between weapon carriage and being a bully-victim and only at one year. One possible explanation is that other risky behaviors play more important roles in explaining both bullying and weapon carrying. For example, a recent study (Pham, Schapiro, John, & Adesman, 2017) found that the association between victimization and weapon carrying became non-significant after controlling for fighting at school, being threatened or injured at school, and skipping school because of fear for one’s safety. It is thus important for future research to consider the temporal relationship between weapon carrying and bullying, and to identify and include mediators and moderators to further verify this relationship (Holt & Gini, 2017).

It should be noted that our study did not distinguish weapon types, which limits exploration of possible different implications of carrying a gun versus a knife or club. Another limitation resides in the overall lack of consistency of bullying measures in this area of research (Volk, Veenstra, Espelage, 2017). Although our study followed one of the suggested methods by providing a bullying definition before the question (Volk et al., 2017), the lack of behavior specific measures for bullying should be noted. Furthermore, school related variables, such as school climate, that have been shown to relate to bullying and weapon carrying behaviors among adolescents (Peres et al., 2018; Shukla, Konold, & Cornell, 2016) were not included in this study due to limitation of the survey. Future research should further verify the associations controlling for school factors.

Despite these limitations, this is among the first study to investigate longitudinal associations between bullying and weapon carrying, and the first to examine all three types of bullying in a longitudinal investigation. Our finding that being both a victim and a perpetrator of bullying predicted weapon carriage one year later emphasizes the importance of prevention and intervention programs that target bullying perpetration and victimization. In addition, programs could include conflict resolution skills for bullying as an alternative to weapon carriage.