Sex,Race/Ethnicity,and Context in School-Associated Student Homicides

Abstract

This study assessed the importance of sex, race/ethnicity, and geographic context for incidents of school-associated student homicides between July 1, 1994 and June 30, 1999, covering 5 academic years. Using data from the Centers for Disease Control and Prevention School Associated Violent Deaths Study (n = 125 incidents), we compared percentages and medians of victim, offender, motive, and school characteristics for incidents by geographic context and race/ethnicity of the offenders. Most incidents involved urban areas (53.6%), Black and Latino offenders and victims, moderately high youth poverty, and male on male violence (77.6%) driven by disputes and gang-related motives. Suburban area incidents (31.2%) often involved offenders and victims of a different race/ethnicity (51.3%). Multiple victims and White offenders were more common in rural areas (15.2%). More than 50% of the rural incidents involved male offenders and female victims. White offender incidents more often included multiple victims and female victims while Black and Latino offenders more often included single victims of the same sex. These results emphasize the utility of an incident-based analysis of school-associated student homicides in highlighting important variations by intersections of sex, race/ethnicity, and geographic context.

Keywords

school homicides sex race/ethnicity violence geographic context and homicide

Introduction

As the second leading cause of death to 5- to 18-year-olds (Centers for Disease Control and Prevention [CDC], 2007), youth homicide is a critical concern. School-associated homicides, however, are rare and account for less than 1% of homicides to school-age children (Anderson et al., 2001; Kachur et al., 1996). Likewise, schools are generally safe and have experienced declines in school violence over the past 18 years (CDC, 2008a; DeVoe, Noonan, Snyder, & Baum, 2005). These events, however, are important because they increase fear among students and families (Addington, 2003; Brener, Simon, Anderson, Barrios, & Small, 2002; Carlson, 2001; Gallup & Gallup, 1999; Stretesky & Hogan, 2001) and may negatively affect student attendance and grades (Bowen & Bowen, 1999). Because most youth in the United States attend schools and may face this increased fear and other negative effects due to the high-profile nature of school homicides, even those occurring in distant places, these rare events are an important subject of research. In addition, any knowledge gained through such research has the potential to help prevent future violence.

Based on population surveillance research and targeted case studies of school shooters, prior research established school homicide victimization and offending risk factors (Anderson et al., 2001; Critical Incident Response Group, Federal Bureau of Investigation, 1999; Kachur et al., 1996; Meloy, Hempel, Mohandie, Shiva, & Gray, 2001; Verlinden, Hersen, & Thomas, 2000; U.S. Secret Service, 2000). Although primarily focusing on multiple victims or random victim events, a few of these case studies also provided insight into the importance of geographic context (Newman, Fox, Harding, Mehta, & Roth, 2004; National Research Council [NRC] & Institute of Medicine [IOM], 2003) and the joint influence of victim and offender characteristics that are evident by analyzing the incident (Danner & Carmody, 2001; Kimmel & Mahler, 2003; Klein & Chancer, 2000).

Our purpose in this research is to move beyond individual risk approaches and the limited focus on specific types of school homicides by examining two main questions. First, how do incidents of school homicides vary by geographic context? Second, how are joint characteristics of victims and offenders (sex, race/ethnicity) associated with geographic context? This type of approach can improve understanding of school-associated student homicide incidents more broadly and inform future research and violence-prevention efforts.

Geographic Context and School Homicides

Geographic patterns of school homicides are similar to those for general youth violence and homicide (CDC, 2007, 2008b) in that youth in urban areas face a higher risk for school homicide victimization (Anderson et al., 2001; CDC, 2008a; Kachur et al., 1996). Qualitative case studies of six school homicide incidents provide further evidence that geographic context may influence the nature of school homicides due to differences in the structure of communities in rural versus urban areas and differences in the centrality of school to community life (Newman et al., 2004; NRC & IOM, 2003). In rural areas where schools are the center of community life and marginalized youth have few alternative peer activities, incidents may be more likely to involve attacks on multiple youth and random firing (Newman et al., 2004; NRC & IOM, 2003). Such acts target the perceived unfairness of the peer status system that led to marginalization. Incidents in urban areas where youth may have additional peer groups linked to neighborhoods more typically involve specific disputes between individuals (Newman et al., 2004). Because of the limited research in this area, we first examine how characteristics of school homicide incidents vary by geographic context.

Victim and Offender Characteristics in School Homicides

There is evidence that the joint characteristics of victims and offenders (sex, race/ethnicity) are linked to context. Youth of color and males (particularly those in urban areas) are at the highest risk for school-associated homicides (Anderson et al., 2001; CDC, 2008a; Kachur et al., 1996). Population surveillance research and targeted case studies of school shooters show that offenders are more likely to have been bullied, had prior histories of aggression and problems with peers, and access to firearms (Anderson et al., 2001; Critical Incident Response Group, Federal Bureau of Investigation, 1999; Kachur et al., 1996; Meloy et al., 2001; U.S. Secret Service, 2000; Verlinden et al., 2000). Several studies using media reports of random or infamous incidents of school homicides found that many of these incidents involved targeting of girls by boys, homophobic bullying, and masculinity issues that have been overlooked by prior research and may be important for violence prevention programs (Danner & Carmody, 2001; Kimmel & Mahler, 2003; Klein, 2006; Klein & Chancer, 2000). Thus, examining the joint characteristics of victims and offenders in multiple victim and random victim school homicide incidents provided insight into topics (sex-based violence) neglected by individual risk approaches. Moreover, most of these cases that involved specific targeting of girls by boys occurred in suburban and rural areas (Danner & Carmody, 2001; Newman et al., 2004). This prior research examining the joint characteristics of victims and offenders was limited, however, by a focus on only multiple victim and random victim school homicide incidents. The vast majority of school homicide incidents involve single victims (Anderson et al., 2001; CDC, 2008a), and most incidents are driven by either disputes or gang-related motives suggesting specific targeting of victims (Anderson et al., 2001). Hence, we contribute to this literature by focusing on the joint characteristics of victims and offenders in incidents of all types of school-associated student homicides and examining the association with geographic context.

The Present Study

In this study, we build on the developing literature on lethal school violence through an incident-based analysis that examines the characteristics of school homicides by geographic context with a particular focus on the intersection (joint characteristics) of sex and race/ethnicity of victims and offenders. We use unique data collected by the Centers for Disease Control and Prevention, the School Associated Violent Deaths Study, including all incidents of school-associated student homicides during the 5 academic years from July 1994 to June 1999. We address the implications of these findings for researchers and violence-prevention programs.

Method

Data

We obtained permission to use data collected by the Centers for Disease Control and Prevention (CDC) in conjunction with the U.S. Departments of Education and Justice of the population of school-associated violent death incidents in the United States between July 1, 1994 and June 30, 1999 (see Anderson et al., 2001). Before these data were collected, the study protocol was reviewed and approved by the institutional review board of the Centers for Disease Control and Prevention in Atlanta, Georgia. The CDC identified incidents through extensive searches of newspaper and broadcast media databases (Lexis-Nexis and Dialog) in comparison with the lists of the U.S. Department of Education and the National School Safety Center (Anderson et al., 2001). A school-associated violent death was defined as a homicide, suicide, or legal intervention (justifiable shooting by law enforcement)-related death that occurred on the campus of a functioning public or private elementary or secondary school while the victim was on the way to or from regular sessions at school or while the victim was attending or traveling to/from a school-sponsored event. These researchers identified and confirmed 220 incidents (including 253 deaths) and conducted interviews with two sources that had knowledge of the incident and the individuals: law enforcement personnel such as a homicide detective or other lead police investigator and school personnel such as a principal or teacher (Anderson et al., 2001). Interviews included information about the victims, alleged offenders, school, and the incident. Our sample consists of the 130 homicide incidents that included a student as a victim or offender. There are five incidents missing data on geographic context reducing those analyses to n = 125. To further investigate our interest in the joint characteristics of victims and offenders, we also conduct analysis comparing the characteristics of the incidents by the race/ethnicity of the offenders focusing on the three largest racial/ethnic groups (White, Black, and Latino). For these analyses, the sample size is 106. The 19 excluded cases for this analysis involved either multiple offenders who were a mix of races/ethnicities or unknown offenders.

In addition to the data from the interviews, CDC researchers also matched school- and district-level data from the U.S. Department of Education’s Common Core of Data and Private School Universe Survey (U.S. Department of Education, National Center for Education Statistics, 2001a, 2001b). The U.S. Department of Education sends annual surveys to state departments of education who work with local districts and schools to provide information based on their administrative records to create the Common Core of Data. These data include information on all public schools and school districts in the United States such as demographic characteristics of the student body, fiscal data, and school district population information. The Private School Universe Survey offers a parallel data system for private schools in the United States and is based on surveys the U.S. Department of Education sends to administrators at all private schools every 2 years.

Measures

Given our two research questions, we developed measures focusing on the geographic context, demographic characteristics of the school and school district, demographic characteristics of the victims and offenders, and characteristics and motives of the homicides. Geographic context is a categorical variable coded as urban, suburban, and rural. This measure comes from the Common Core of Data Survey and the Private School Universe Surveys which are administered by the U.S. Department of Education.¹

We capture the larger socioeconomic context with a continuous measure of the percentage of children 5 to 17 years of age in poverty in the school district. The racial/ethnic makeup of the school is captured with three continuous variables: percentage of White, percentage of Black, and percentage of Latino/a youth in the school. We specifically focus on the three largest racial/ethnic groups in the sample (and the United States) and use these same groups for coding the race/ethnicity of victims and offenders.

To examine characteristics of victims and offenders, we created several categorical variables at the level of the incident. We include five dichotomous variables capturing the race/ethnicity of victims and offenders: (a) all victims and offenders are the same race/ethnicity, (b) White offender(s) only, (c) Black offender(s) only, (d) Latino/a offender(s) only, and (e) Mixed or unknown offenders. For sex of the victim(s) and offenders(s), we created four dichotomous variables: (a) male offender(s), (b) female victim in event, (c) victims and offenders of the same sex only, and (d) male offender(s) and female victims. As some incidents involved female offenders (female victims or male victims), we included the “male offender–female victim” variable to be sure to clearly identify incidents with male offenders and female victims that have been highlighted in prior research (Danner & Carmody, 2001; Klein, 2006). While the sex categories “female victim in event” and “male offender(s) and female victims” are not mutually exclusive, we are not analyzing these two variables together but using each independently to more effectively analyze how sex is significant in these incidents.

Finally, we examine characteristics of the homicides. We include two dichotomous variables denoting multiple victims and multiple offenders (0 = single, 1 = multiple). The interview sources answered yes or no to any in a long list of potential motives (disputes, gang-related, random victim, romantic dispute, robbery, rape, suicide, and others). We coded the motives with the highest frequencies in the data set (disputes, romantic disputes, gang-related, and random) into dichotomous variables.

Analytic strategy

Our primary interest is in comparing the characteristics of incidents of school associated student homicides across geographic context and considering the joint demographic characteristics of victims and offenders. We perform descriptive analyses including percentage and median comparisons by geographic context and race/ethnicity of offender(s). For categorical variables, geographic and racial/ethnic differences in proportions of victims, offenders, and motives associated with school homicides were assessed using Pearson’s chi-square analyses, and the Fisher’s Exact Test for any cells with expected counts below 5, where α < .10. We used .10 as the cutoff because of the reduced sample size with these comparisons, but most significant comparisons had α < .05. Each analysis involved an individual 2 × 2 table (e.g., Dispute Motive [1/0] × Urban [1]/Suburban[0]). As we were interested in differences between each of the three geographic categories and the three racial/ethnic categories of offenders (e.g., urban vs. suburban, urban vs. rural, Black vs. Latino offenders), we used the 2 × 2 chi-squares rather than a 2 × 3 table which would have obscured differences. For continuous variables measuring the characteristics of schools associated with homicide incidents, we first assessed whether each variable was normally distributed using the Kolmogorov-Smirnov and Shapiro-Wilk tests. As these tests were significant for our four continuous measures signifying that they are not normally distributed, we use nonparametric statistics including the median, interquartile range, and the Independent Samples Mann-Whitney U test (e.g., percentage of children in poverty tested across urban vs. suburban) where α < .05. All analyses were conducted using SPSS software.

Results

Incident Characteristics by Geographic Context

We first address how the characteristics of school-associated student homicides vary across geographic context. The majority of school homicide incidents (53.6%) occurred in urban areas, followed by suburban (31.2%) and rural (15.2%; see Table 1). More than one third of incidents involved disputes, and this was similar across geographic areas. Urban areas had a significantly higher percentage (44.8%) of gang-related incidents compared with suburban (25.6%) and rural (5.3%) areas. Less than one in five incidents involved random motives (16%) across all areas with no significant differences.

Table 1.

Chi-Square Comparisons of Victims, Offenders, and Motives in School Homicide Incidents by Geographic Location

	All Cases (N = 125)	Urban (n = 67, 53.6%)	Suburban (n = 39, 31.2%)	Rural (n = 19, 15.2%)
Characteristic	Row %	Column %	Column %	Column %
Motives
Dispute	38.4	38.8	35.9	41.2
Romantic dispute	13.6	9.0	17.9	21.1
Gang-related	32.8	44.8	25.6^a	5.3^{b c}
Random	16.0	17.9	12.8	15.8
Victims
Multiple victims	10.4	7.5	7.7	26.3^{b c}
Female victim(s)	30.4	20.9	38.5^a	47.4^b
Offenders
Multiple offenders	23.2	28.4	17.9	15.8
Male offender(s)	88.8	89.6	87.2	89.5
White offender(s)	17.6	6.0	25.6^a	42.1^b
Black offender(s)	46.4	59.7	30.8^a	31.6^b
Latino offender(s)	17.6	19.4	17.9	10.5
Victim–offender
Victim–offender same sex	69.6	77.6	64.1	52.6^b
Male offender and female victim	24.8	16.4	30.8^a	52.1^b
Victim–offender same race	61.6	65.7	48.7	73.7^c

Urban is significantly different from Suburban at p < .10.

Urban is significantly different from Rural at p < .10.

Suburban is significantly different from Rural at p < .10.

More than a quarter of the school homicide incidents in rural areas (26.3%) involved multiple victims. Suburban (7.7%) and urban (7.5%) areas had low percentages of multiple victim incidents and both were significantly less than rural areas. Both suburban areas (38.5%) and rural areas (47.4%) had a significantly higher percentage of incidents involving female victims compared with urban areas (20.9%).

Less than one fourth (23.2%) of incidents involved multiple offenders and most (88.8%) involved male offenders (see Table 1). The race and ethnicity of offenders varied significantly across geographic context. Rural area incidents involved a significantly larger percentage of White offenders (42.1%) compared with urban areas (6.0%), with suburban areas being in the middle (25.6%) but not significantly different from the rural areas. Urban area offenders involved a significantly larger proportion of Black youth (59.7%) than suburban (30.8%) and rural areas (31.6%). Less than one in five incidents (17.6%) involved Latino offenders overall, with no significant geographic context differences.

Incidents in urban areas were located in socioeconomic contexts with a significantly higher median percentage of child poverty (33.31%) compared with rural (22.88%) and suburban (14.94%) incidents (Table 2). The racial/ethnic makeup of the schools associated with these incidents varied significantly by geographic context. Rural incidents involved schools where the median of 55.14% of the youth were White compared with 44.98% in suburban areas and 9.15% in urban areas. Urban incidents involved schools where the median of 56.55% of the youth were Black compared with 17.26% in suburban areas and 9.79% in rural areas. Rural incidents involved schools with the lowest median percentage of Latino youth (0.48%) compared with suburban (14.32%) and urban (8.61%) incidents.

Table 2.

Median Comparisons of Characteristics for School Homicide Incidents by Geographic Location (Independent Samples Mann-Whitney U Test)

	All Cases (N = 125)	Urban (n = 67)	Suburban (n = 39)	Rural (n = 19)
Characteristic	Median (IQR)	Median (IQR)	Median (IQR)	Median (IQR)
% of children (5-17 years) in poverty in school district	24.83 (18.95)	33.31 (12.10)	14.94 (11.22)^a	22.88 (12.79)^{b c}
% White in school	20.96 (56.23)	9.15 (33.76)	44.98 (58.35)^a	55.14 (64.46)^b
% Black in school	33.96 (73.78)	56.55 (73.06)	17.26 (41.92)^a	9.79 (65.34)^b
% Latino/a in school	7.25 (30.23)	8.61 (29.75)	14.32 (31.10)	0.48 (1.71)^{b c}

Note: IQR = interquartile range between the 25th percentile and the 75th percentile.

Urban is significantly different from Suburban at p < .05.

Urban is significantly different from Rural at p < .05.

Suburban is significantly different from Rural at p < .05.

Joint Characteristics of Victims/Offenders by Geographic Context

In looking at our second question, sex and race/ethnicity matter in school homicides in part through the combined victim–offender characteristics shown in Table 1. Most incidents (69.6%) involved victims and offenders of the same sex (primarily male as seen previously). Incidents in rural areas (52.6%) involved a significantly lower percentage of sex matching compared with urban areas (77.6%). This contrast is most evident with a significant majority of incidents in rural areas (52.1%) involving a male offender and a female victim compared with incidents in urban areas (16.4%). Suburban areas (30.8%) fall in the middle and are significantly higher than the urban percentage. While the majority of the incidents involved victims and offenders of the same race/ethnicity (61.6%), suburban incidents (48.7%) involved a significantly lower percentage compared with rural incidents (73.7%).

To more thoroughly examine the joint characteristics of victim/offenders, we also compared the incidents by the race/ethnicity of the incident offender(s) (Table 3). More than a quarter (26.1%) of the incidents with White offenders involved multiple victims compared with 6.8% of the incidents with Black offenders and 4.2% of the incidents with Latino offenders. More than half (52.2%) of the incidents with White offenders involved female victims compared with 27.1% of the incidents with Black offenders and 16.7% of the incidents with Latino offenders. The majority of the incidents with Black offenders (79.7%) and Latino offenders (87.5%) involved sex matching with the victim compared with the significantly lower 47.8% of the incidents with White offenders. Likewise, 47.8% of the incidents with White offenders involved males killing females, compared with a much lower 18.6% of the incidents with Black offenders and 12.5% of the incidents with Latino offenders. There were two significant differences by incident motives. A significantly higher percentage of incidents with Black offenders involved disputes (47.5%) compared with incidents with Latino offenders (25.0%). More than half of the incidents with Latino offenders (54.2%) involved gang-related motives compared with 35.6% of incidents with Black offenders and the lowest 13% of incidents with White offenders.

Table 3.

Chi-Square Comparisons of Victims, Offenders, and Motives in School Homicide Incidents by Race/Ethnicity of Offender(s)

	White (n = 23)	Black (n = 59)	Latino (n = 24)
Characteristic	Column %	Column %	Column %
Victims
Multiple victims	26.1	6.8^a	4.2^b
Female victim(s)	52.2	27.1^a	16.7^b
Offender
Multiple offenders	21.7	20.3	29.2
Male offender(s)	91.3	89.8	95.8
Victim–offender
Victim–offender same sex	47.8	79.7^a	87.5^b
Male offender and female victim	47.8	18.6^a	12.5^b
Victim–offender same race	65.2	76.3	70.8
Motives
Dispute	34.8	47.5	25.0^c
Romantic dispute	17.4	15.3	12.5
Gang-related	13.0	35.6^a	54.2^b
Random	21.7	16.9	8.3

White is significantly different from Black at p < .10.

White is significantly different from Latino at p < .10.

Black is significantly different from Latino at p < .10.

From Table 4, incidents with White offenders involved a significantly lower median of youth in poverty (16.92%) compared with incidents with Black (28.45%) and Latino offenders (26.06%). The racial/ethnic characteristics of the offenders largely matched the average majority racial/ethnic group in the schools. Incidents with White offenders were in majority White schools based on the median (82.27% White). Likewise, incidents with Black offenders were in majority Black schools based on the median (68.46% Black) whereas incidents with Latino offenders were in majority Latino schools based on the median (55.09% Latino). The interquartile ranges between the 25th and the 75th percentile for the racial/ethnic makeup of the schools varied tremendously.

Table 4.

Median Comparisons of Characteristics for School Homicide Incidents by Race/Ethnicity of Offenders (Independent Samples Mann-Whitney U Test)

	White (n = 23)	Black (n = 59)	Latino (n = 24)
Characteristic	Median (IQR)	Median (IQR)	Median (IQR)
% of children (5-17 years) in poverty in school district	16.92 (13.46)	28.45 (16.76)^a	26.06 (15.95)^b
% White in school	82.27 (49.86)	10.90 (42.23)^a	23.16 (42.09)^b
% Black in school	4.39 (18.28)	68.46 (60.10)^a	7.81 (29.93)^c
% Latino/a in school	1.36 (11.26)	0.81 (14.85)	55.09 (37.53)^{b c}

Note: IQR = interquartile range between the 25th percentile and the 75th percentile.

White is significantly different from Black at p < .05.

White is significantly different from Latino at p < .05.

Black is significantly different from Latino at p < .05.

Discussion

Prior epidemiological research established that youth of color, males, and urban youth are at higher individual risk for school-associated homicides (Anderson et al., 2001; Kachur et al., 1996). More targeted studies suggested that incidents of school homicide varied by geographic context (Newman et al., 2004; NRC & IOM, 2003) and that examining the joint characteristics of victims/offenders provided insight into neglected areas (sex-based violence; Danner & Carmody, 2001; Kimmel & Mahler, 2003; Klein, 2006; Klein & Chancer, 2000). Moving beyond individual risk approaches and the limited focus of much of the prior research, we found significant variation in incident characteristics by geographic context and in ways linked to the sex and race/ethnicity of offenders and victims. These results have implications for future research and violence prevention.

The majority of the incidents occurred in urban areas, involved male victims with male offenders, Black and Latino offenders and victims living in conditions where the median of youth in poverty is more than 33%, and dispute or gang-related motives. These incidents have been neglected by most research on school homicides. For researchers and the mass media, it is important to note that youth of color in urban areas are the most critical population not only for youth violence and homicides in general but also for school-associated homicides.

The observed racial and geographic differences in school-associated homicides may be linked to economic and social reality differences in these areas and may mirror the larger patterns of homicide in each area (Cubbin, Pickle, & Fingerhut, 2000; Fingerhut, Ingram, & Feldman, 1998). The finding that most incidents in urban areas involved intraracial homicides among Blacks and Latinos while most incidents in rural areas involved intraracial homicides among Whites is consistent with research which ties race- and place-specific homicide patterns to residential patterns and racial differences in the impacts of racial residential segregation, resource deprivation, and labor market opportunity (Gallup-Black, 2004, 2005; Parker & McCall, 1999). Blacks and Latinos are more likely than Whites to reside in major urban areas, whereas the opposite is true in rural areas. It is beyond the scope of our study to specifically examine this issue. However, this finding may reflect how spatial changes and economic declines since the 1970s have led to resource deprivation and residential segregation increasing the likelihood of intraracial homicides by restricting interaction to persons of the same race and exacerbating intraracial competition for scarce resources (Gallup-Black, 2004, 2005; Parker & McCall, 1999).

Suburban incidents also appear to have some unique characteristics. For more than 30% of incidents, the race/ethnicity of offenders and their respective schools were more diverse and had lower proportions of youth in poverty compared with the urban and rural areas. Unlike urban and rural areas, the majority of the incidents in suburban areas involved offenders and victims of a different race/ethnicity. Changes in racial disposition of victim–offender patterns in school-associated homicide characteristics occurring in suburban areas may be related to shifts in economic and social opportunity structures. In comparison with urban and rural areas, suburban areas may provide greater economic opportunities or be occupied by those who have achieved greater levels of economic success regardless of their race. Suburban areas may also afford more possibilities for interactions between persons of different races which may include conflicts that may result in interracial homicide.

Shifting from urban to suburban and rural areas, the characteristics of school homicide incidents more heavily involved multiple victims, White offenders, and female victims. As noted by some key sociological researchers (Danner & Carmody, 2001; Kimmel & Mahler, 2003; Klein, 2006; Klein & Chancer, 2000), sex-based violence becomes more important with more than 30% of the suburban incidents and more than 50% of the rural incidents involving male offenders and female victims. This school-based violence against females, although noted by sociologists, has largely been absent in the epidemiological research which has focused more heavily on individual risk for school homicide. An incident-based analysis brings this important issue to the forefront. Although we cannot explain why we see an increasing proportion of incidents with male offenders and female victims shifting from urban to rural areas in our analyses, these findings mirror a larger pattern on geographic differences in intimate partner homicide. Population-based rates of intimate partner murder in rural areas from 1980 to 1999 were far greater than those observed in urban and suburban areas (Gallup-Black, 2004, 2005). Intimate partner homicide rates in nonmetropolitan areas exceeded those of urban areas, although lower than rural areas. Relevant explanations focus on rural areas’ greater presence of traditional norms encouraging more patriarchal forms of masculinity and gender roles conducive to intimate partner violence (Gallup-Black, 2004, 2005). These factors may interact with other personal factors and situational stressors to increase the likelihood of intimate partner homicide. Individuals in rural areas are more isolated and have less access to formal and informal resources that may help prevent a situation of intimate partner violence from escalating to homicide. Another possible explanation may be that there are fewer romantic options in rural areas. With fewer possible alternatives, rejection may be more likely to fuel feelings of desperation that could lead to violence.²

Future research should consider an ecological approach that accounts for broader contextual factors of communities in which school homicides occur, such as racial composition, economic factors, and other relevant factors. It is necessary to evaluate rates of school violence in conjunction with macro-level community factors to understand these differences with respect to characteristics of the different geographic locations in which they occur.

Finally, these findings have implications for violence prevention. The effectiveness of school-based violence prevention programs varies as a function of the characteristics of classrooms, schools, and the communities in which they are implemented (Ozer, 2006). Because the nature of school homicides is different in rural, suburban, and urban areas, prevention programs need to be sensitive to the demographic makeup of students and to how the community context itself may influence the nature of violence. This is critical so that all students receive quality education in safe, learning-conducive environments.

Limitations

Although informative, our analysis is limited in several areas. Our data do not provide information related to the meanings associated with these acts of violence that may address some of the more nuanced ways in which geographic context, race/ethnicity, and sex structure interactions leading to violence. Other qualitative research has provided a greater understanding of how norms related to violence and masculinities in certain contexts may produce different forms of violence (Anderson 1999; Newman et al., 2004). It would be useful for future research to use mixed methods to connect information on multiple types of school homicides with greater depth of information. Our analyses are also based on a relatively small numbers of incidents, particularly for rural areas and White and Latino offenders. While we include 5 years of data, it would be useful for future research to examine a longer time period to capture a greater number of incidents and allow for a more detailed incident-based analysis of these issues.

Given that our data cover 5 years during the 1990s, it is also possible that these findings may not generalize to other time periods. For example, Anderson et al.’s (2001) analysis of trends using this same data for the period 1992-1999 highlight that rates of single-victim student homicides decreased over the 1990s whereas rates of multiple-victim student homicides increased. CDC (2008a) researchers have continued with additional data collection covering the 1999-2006 time frame and found no significant change in rates for this more recent period. Once the CDC finishes with this more recent data collection, it will be possible for future research to examine if our results hold up over a longer time frame.

Footnotes

Acknowledgements

The authors wish to thank Thomas Simon, Bill Modzeleski, Thomas Feucht, Ryan King, Elizabeth Popp-Berman, Kecia Johnson, Kate Strully, Sapna Swaroop, and anonymous reviewers for comments on previous versions of the manuscript.

Authors’ Note

The findings and conclusions in this report are those of the authors and do not necessarily represent the official position of the Centers for Disease Control and Prevention, the Department of Education, or the Department of Justice.

Declaration of Conflicting Interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The authors received no financial support for the research, authorship, and/or publication of this article.

Notes

Bios

Joanne M. Kaufman, PhD, is an assistant professor of sociology at the University at Albany, State University of New York. Her research examines social psychological processes associated with crime and desistance, the characteristics of school related crimes, and race and gender variation in crime and deviance.

Jeffrey E. Hall, PhD, MSPH, is a behavioral scientist in the Division of Violence Prevention at the National Center for Injury Prevention and Control at the Centers for Disease Control and Prevention (CDC). He is a youth violence subject matter expert and chair of the Education and Training Subcommittee of the CDC Aging and Health Work Group. His research focuses on infant homicide, youth and young adult violence, self-neglect, and elder maltreatment.

Michelle Zagura, BA, is a doctoral student in the Department of Sociology at the University at Albany, State University of New York. Her research focuses on the intersections of age, gender, and marital satisfaction.

References

Addington

L. A.

(2003). Students’ fear after Columbine: Findings from a randomized experiment. Journal of Quantitative Criminology, 19, 367-387.

Anderson

(1999). The code of the street. New York, NY: W. W. Norton.

Anderson

Kaufman

Simon

T. R.

Barrios

Paulozzi

Ryan

, . . . the School-Associated Violent Deaths Study Group. (2001). School-associated violent deaths in the United States, 1994-1999. Journal of the American Medical Association, 286, 2695-2702.

Bowen

(1999). Effects of crime and violence in neighborhoods and schools on the school behavior and performance of adolescents. Journal of Adolescent Research, 14, 319-342.

Brener

N. D.

Simon

T. R.

Anderson

Barrios

L. C.

Small

M. L.

(2002). Effect of the incident at Columbine on students’ violence- and suicide-related behaviors. American Journal of Preventive Medicine, 22(3), 146-150.

Carlson

D. K.

(2001, March 5). Majority of parents think a school shooting could occur in their community. Retrieved from http://www.gallup.com/poll/1939/Majority-Parents-Think-School-Shooting-Could-Occur-Their-Community.aspx

Centers for Disease Control and Prevention. (2007). Web-based Injury Statistics Query and Reporting System (WISQARS), 2007 Mortality Data. Atlanta, GA: National Center for Injury Prevention and Control, Centers for Disease Control and Prevention. Retrieved from http://www.cdc.gov/ncipc/wisqars/default.htm

Centers for Disease Control and Prevention. (2008a). School-associated student homicides—United States, 1992-2006. Morbidity and Mortality Weekly Report, 57(2), 33-36.

Centers for Disease Control and Prevention. (2008b). Youth Risk Behavioral Surveillance—United States, 2007. Morbidity and Mortality Weekly Report, 57(SS-4), 1-131.

10.

Critical Incident Response Group, Federal Bureau of Investigation. (1999). The school shooter: A threat assessment perspective. Washington, DC: National Center for the Analysis of Violence Crime. Retrieved from http://www.fbi.gov/stats-services/publications/school-shooter

11.

Cubbin

Pickle

L. W.

Fingerhut

(2000). Social context and geographic patterns of homicide among US Black and White males. American Journal of Public Health, 90, 579-587.

12.

Danner

M. J. E.

Carmody

D. C.

(2001). Missing gender in cases of infamous school violence: Investigating research and media explanations. Justice Quarterly, 18, 87-114.

13.

DeVoe

J. F.

Peter

Noonan

Snyder

T. D.

Baum

(2005). Indicators of School Crime and Safety: 2005 (NCES 2006-001/NCJ 210697, U.S. Departments of Education and Justice). Washington, DC: U.S. Government Printing Office.

14.

Gallup

(1999). Special report: Violence in America’s high schools. Youthviews: The Newsletter of the Gallup Youth Survey, 7(2), 1-5.

15.

Gallup-Black

(2004). Rural and urban trends in family and intimate partner homicide: 1980-1999. Retrieved from https://www.ncjrs.gov/pdffiles1/nij/grants/208344.pdf

16.

Gallup-Black

(2005). Twenty years of rural and urban trends in family and intimate partner homicide: Does place matter? Homicide Studies, 149(9), 163-164.

17.

Fingerhut

Ingram

Feldman

(1998). Homicide rates among US teenagers and young adults. Journal of the American Medical Association, 280, 423-427.

18.

Kachur

S. P.

Stennies

G. M.

Powell

K. E.

Modzeleski

Stephens

Murphy

. . . Lowry

(1996). School-associated violent deaths in the United States, 1992-1994. Journal of the American Medical Association, 275, 1729-1733.

19.

Kimmel

M. S.

Mahler

(2003). Adolescent masculinity, homophobia, and violence: Random school shootings, 1982-2001. American Behavioral Scientist, 46, 1439-1458.

20.

Klein

Chancer

L. S.

(2000). Masculinity matters, the omission of gender from high-profile school violence cases. In Urso Spina

(Ed.), Smoke and mirrors (pp. 129-162). Lanham, MD: Rowman and Littlefield.

21.

Klein

(2006). An invisible problem, everyday violence against girls in school. Theoretical Criminology, 10, 147-177.

22.

Meloy

J. R.

Hempel

A. G.

Mohandie

Shiva

Gray

B. T.

(2001). Offender and offense characteristics of a nonrandom sample of adolescent mass murderers. Journal of the American Academy of Child and Adolescent Psychiatry, 40, 719-728.

23.

National Research Council & Institute of Medicine. (2003). Deadly lessons: Understanding lethal school violence: Case Studies of School Violence Committee ( Moore

M. H.

Petrie

C. V.

Braga

A. A.

McLaughlin

B. L.

, Eds., Division of Behavioral and Social Sciences and Education). Washington, DC: The National Academies Press.

24.

Newman

K. S.

Fox

Harding

Mehta

Roth

(2004). Rampage, the social roots of school shootings. New York, NY: Basic Books.

25.

Ozer

(2006). Contextual effects in school-based violence prevention programs: A conceptual framework and empirical review. Journal of Primary Prevention, 27, 315-340.

26.

Parker

McCall

(1999). Structural conditions and racial homicide patterns: A look at the multiple disadvantages in urban areas. Criminology, 37, 447-478.

27.

Stretesky

P. B.

Hogan

M. J.

(2001). Columbine and student perceptions of safety: A quasi-experimental study. Journal of Criminal Justice, 29, 429-443.

28.

U.S. Department of Education, National Center for Education Statistics. (2001a). Common Core of Data (CCD), 1994-1999. Retrieved from http://nces.ed.gov/ccd/index.asp

29.

U.S. Department of Education, National Center for Education Statistics. (2001b). Private School Universe Survey (PSS), 1995-1998. Retrieved from http://nces.ed.gov/surveys/pss/

30.

Verlinden

Hersen

Thomas

(2000). Risk factors in school shootings. Clinical Psychology Review, 20, 3-56.

31.

U.S. Secret Service. (2000). An interim report on the prevention of targeted violence in schools. Washington, DC: Safe School Initiative: National Threat Assessment Center, U.S. Department of Education, National Institute of Justice. Retrieved from http://www.secretservice.gov/ntac/ntac_ssi_report.pdf