‘May Piece Be with You’: A Typological Examination of the Fear and Victimization Hypothesis of Adolescent Weapon Carrying

Abstract

The causes and correlates of adolescent weapon carrying have received considerable scholarly attention. One common explanation of adolescent weapon carrying, the “fear and victimization hypothesis,” identifies the fear of victimization as a motivating force behind this behavior. Empirical studies of this explanation, however, have produced mixed results. One potential source of this ambiguity is the myriad data and measurement issues that have arisen in prior studies of the topic. The current study addresses many of these limitations through the use of panel data from over 1,100 youth across the United States. Through a typological approach, results support a multiple pathways framework for explaining adolescent weapon carrying, as the perceived risk of victimization is positively associated with weapon carrying for those youth who report both victimization and offending experiences. For those youth who are “pure offenders,” fear of victimization is inversely related to weapon carrying, while the perceived risk of victimization is unrelated to carrying weapons. Consistent with prior literature, results indicate that gang members report a disproportionately high level of weapon carrying.

Acknowledgments

This research is funded by the National Institute of Justice, Office of Justice Programs, US Department of Justice, Award No. 2003‐JN‐FX‐0003.

Notes

1. A previous version of this paper was presented at the 2007 American Society of Criminology annual meeting in Atlanta, GA. The authors would like to thank Adam Watkins, Ben Voss, Richard Tewksbury, and the three anonymous reviewers for their helpful comments on previous drafts of the manuscript. Any remaining mistakes or omissions are the sole responsibility of the authors. Points of view in this document are those of the authors and do not necessarily represent the official position of the US Department of Justice.

2. The effect of age on weapon carrying in general samples appears to be positive (Lizotte, Howard, Krohn, & Thornberry, 1997; Wilcox et al., 2006). It appears this curvilinear relationship presents itself within exclusively adolescent samples (see Wilcox et al., 2006).

3. There is substantial debate as to the exact relationship between gang involvement and crime. Most of this debate concerns what have been termed the facilitation and enhancement effects of gang membership on crime (see Thornberry et al., 2003, for a review). We use the term facilitation in this context because evidence related to the effect of gang membership on gun carrying suggests facilitation rather than an enhancement effect. This may not be the case with general delinquency, as evidence suggests an enhancement effect with general delinquency.

4. This logic has also been used in current discussions about the potential benefits of allowing concealed carry permits on college campuses in the wake of the tragic campus shootings at Virginia Tech University and Northern Illinois University.

5. While the data are not nationally representative, and were collected in a purposive manner, we feel the diversity of sites included in the study strengthens the conclusions one can draw from study results.

6. Pretest data were collected prior to the administration of the evaluated curriculum during the 2004–2005 school year, and post‐tests were completed directly following the completion of the program. A 1‐year follow‐up survey was conducted in the fall of 2005. All three waves of survey data were collected using group‐administered self‐report methods, whereby subjects answered questions individually as they were read aloud by members of the research team. The approximate time needed for completion of the survey was 40–45 minutes.

7. Approximately 200 of these students were lost when they transitioned from an elementary school district to a high school district. While we sought approval for this evaluation design from the high school district, the central office failed to approve our proposal and would not allow us to administer the student surveys due to low scores on testing related to the No Child Left Behind legislation.

8. These students simply filled‐out the assent form and discontinued their participation in the survey administration.

9. Differential attrition threatens both the internal and external validity of study findings, because the final sample often fails to adequately represent the “at‐risk” portion of the general public. As Thornberry and colleagues (1993) demonstrated on a sample of school‐aged youths, those youth most likely to be unavailable for school‐administered questionnaires, what they termed “elusive” and/or “uncooperative” youth, were more likely to be delinquent. Therefore, those youth included in follow‐up school‐based surveys only represent part of this age group, leading to an underestimation of general delinquency. With this in mind, the final sample of students involved in the current study are compared with those eliminated from the sample to determine if significant differences exist between the two groups. To do so, t‐tests for independent samples and chi‐square analyses were run on key independent and dependent variables.

10. The “Other” category consists of those reporting racial/ethnic backgrounds as Native American/American Indian, Asian/Pacific Islander, Mixed/Multi‐Racial, and others that wrote in their racial heritage. When at all possible, those that indicated a racial heritage that fell under one of the preset choices, we recoded them into the proper category. For instance, some respondents indicated they were “Mexican” instead of marking “Hispanic.” These respondents were coded as “Hispanic” during the data entry process.

11. Some of Ferraro’s original questions were intentionally left out of the current study. Some questions (e.g., motor vehicle theft victimization) were deemed inappropriate for this age group while others (e.g., those pertaining to rape/sexual assault) were unable to be asked in the school‐setting.

12. Part of the debate over the proper measurement of fear and risk stems from the belief that they do not represent conceptually distinct phenomena. Evidence from the current study supports the conclusion reached by Wilcox and Land (1996) that the two should be treated as distinct. For instance, while the bivariate correlation is substantive (r = .19), the strength of the relationship is not indicative of a single shared latent construct. Further, multicolinearity diagnostics (e.g., vif scores) revealed no potential problems in all analyses.

13. This question was posed within a section devoted to the characteristics and behaviors of one’s peer‐group. For those students who had more than one distinct peer‐group, a prompt at the beginning of the section asked student to “think about the one that is most important to you” in order to alleviate any confusion about which peer‐group they were to report.

14. Due to the small number of respondents located in some categories of the typology, comparisons between groups using tests of significance are problematic, and are thus not reported.

15. Because we eliminated those respondents who did not complete all three waves of the survey, which has the potential to threaten study results because of differential attrition, we estimated all models in Table 3 with data for all respondents that had valid data at Times 1 and 2 (n = 1,354). Results of these supplementary analyses provided two notable differences; both of which further strengthened study conclusions. First, when including all valid respondents in the full model, fear of crime became significantly (p < .05) negatively associated with weapon carrying (exp(B) = .80). Second, in the Victim/Offender model gang membership became significantly (p < .05) related to weapon carrying in the expected direction (exp(B) = 2.51). Therefore, even after relaxing our sample restrictions to include all respondents that had data at Times 1 and 2 our study conclusions appear robust.