Journal of Criminal Justice 34 (2006) 195 – 207

Explaining gang homicides in Newark, New Jersey: Collective behavior or social disorganization? Jesenia M. Pizarro a , Jean Marie McGloin b,⁎ a

School of Criminal Justice, Michigan State University, 502 Baker Hall, East Lansing, MI 48824-1118, United States b Department of Criminology and Criminal Justice, University of Maryland, College Park, 2220L LeFrak Hall, College Park, MD 20742, United States

Abstract Although numerous studies examined the distinction between gang and non-gang homicides, there is nonetheless a continued need for research in this domain. Specifically, few studies investigated the etiological differences between these homicides at the multivariate level or attempted to examine the relative robustness of the primary explanations of gang homicides—social disorganization and Decker's (1996) collective behavior hypothesis of gang violence. This article therefore addresses this void by focusing on gang-related homicides in Newark, New Jersey over a sixty-six month period. The findings of this study suggested that there were significant differences between gang and non-gang homicides at the incident level. At the multivariate level, the authors found that homicides precipitated by the operationalization of Decker's (1996) escalation hypothesis were more likely to be gangrelated. Conversely, the social disorganization measure did not predict gang homicide, while poverty was a significant predictor. When measures of both potential explanations were entered into the same model only the micro-level escalation hypothesis retained its significance. © 2006 Elsevier Ltd. All rights reserved.

Introduction In recent years, scholars began to emphasize the importance of studying disaggregated homicides (Flewelling & Williams, 1999; Pridemore, 2002). According to some scholars, the disaggregation of homicides into specific types of categories is necessary because previous research suggested homicide was an outcome of various types of crimes instead of a homogeneous offense (Flewelling & Williams, 1999). Indeed, criminal justice statistics suggested that homicides occurred for many reasons (Fox & Zawitz, 2002) and, as such, ⁎ Corresponding author. Tel.: +1 301 405 3007; fax: +1 301 405 4733. E-mail address: [email protected] (J.M. McGloin). 0047-2352/$ - see front matter © 2006 Elsevier Ltd. All rights reserved. doi:10.1016/j.jcrimjus.2006.01.002

scholars could not assume that the same factors characterized and predicted these variant types. Accordingly, contemporary homicide researchers began to use different typologies in the analysis of this crime, such as the victim-offender relationship, race of the victim and/ or suspect, age of victim and/or suspect, and the circumstances that provided the impetus for the offense to occur (e.g., availability of drugs and gang involvement) (Avakame, 1998; Kubrin, 2003; Lee & Bartkowski, 2004; Parker, 2001). Of the various typologies, gang involvement subtypes (i.e., gang homicides) received increased scholarly attention in recent years for a variety of reasons (see Decker & Curry, 2002). First, the early 1990s witnessed an exponential growth of urban youth homicides, as well as street gangs. This gang growth included the

196

J.M. Pizarro, J.M. McGloin / Journal of Criminal Justice 34 (2006) 195–207

number of gangs and gang members, both within urban environments and into suburban and rural locales, and the amount of gang crime and violence (Curry & Decker, 1998; Howell, 1998; Miller, 2001; Office of Juvenile Justice and Delinquency Prevention, 1998). Across a variety of geographic locations, evidence emerged that gang homicides were disproportionately contributing to the homicide increase (Block, 1996; Braga, 2003; Maxson, 1999). In addition, gang homicides had a consistently different character than non-gang homicides. As Decker and Curry (2002) noted: “Gang homicides can be distinguished from other homicides by virtue of a number of characteristics. These characteristics include: (1) spatial concentration; (2) weapon use; (3) race of victim and perpetrator; (4) location; (5) drug involvement; (6) age; (7) sex; and (8) victim-offender relationship” (p. 345). In short, gang homicides garnered attention for both quantitative and qualitative reasons. While the literature on gang homicide tended to orient towards the aforementioned issues of proportional role in homicide rates and descriptive characteristics, some research progressed to theoretical explanations (for a review of the gang homicide literature, see Howell, 1999). In particular, two dominant explanations of gang homicide emerged in the literature—a community explanation and a collective behavior explanation. Briefly, Curry and Spergel (1988) asserted that macrolevel ecological factors, namely those inherent to the concept of social disorganization, predicted gang homicide, whereas Decker (1996) relied on microlevel variables and argued that escalating, dynamic, and reciprocal social processes explained gang homicide. These two explanations may not necessarily be mutually exclusive, given that gangs and gang homicides cluster in disadvantaged and disorganized areas, while at the same time often showing an escalation pattern related to reciprocal violence (Klein & Maxson, 1989; Maxson, 1999; Rosenfeld, Bray, & Egley, 1999). Even so the efficacy of these two explanations is not fully known. It is atypical for research to move beyond searching for significant differences with regard to descriptive characteristics of gang and non-gang homicides. A few inquiries progressed to the multivariate level (see for example Curry & Spergel, 1988; Rosenfeld et al., 1999), but it was rare, and no known study had investigated the relative robustness of these explanations. Given that these were the primary etiologies of gang homicides offered by the literature, it was important to assess the extent to which they predicted gang homicides across various data sets, as well as their comparative efficacy.

This article therefore seeks to address this void by focusing on gang related homicides in Newark, New Jersey over a sixty-six month period.1 Along this vein, the following sections further elucidate on the need for disaggregation in homicide studies and on why a gang homicide focus is appropriate and warranted, especially in the context of Newark. Next, the article progresses to a description of the research strategy, describing both the data and analytical strategy under use. After offering the results, the final section discusses the implications of these findings for theory and policy, as well as offers recommendations for future research. Disaggregating homicides Disaggregating homicides in general As previously mentioned, scholars recently began to emphasize the importance of disaggregating homicide (Flewelling & Williams, 1999; Pridemore, 2002). One reason for the emphasis was the lack of consistency among studies that examined aggregate homicide rates. Specifically, the strength and relevance of key macro-level variables, particularly those of social disorganization—economic status, family disruption, residential mobility, urbanism, and ethnic heterogeneity—in explaining homicide rates within geographic areas were still being debated (Land, McCall, & Cohen, 1990; Messner & Rosenfeld, 1999; Pridemore, 2002). To be clear, no social disorganization variable had presented a consistently positive relationship with homicide rates in geographic areas across studies (Land et al., 1990; Pridemore, 2002). For example, although the majority of studies showed that economic status and family disruption were related to homicide rates within geographic areas, a number of studies found a null relationship (Chamlin, 1989; Loftin & Parker, 1985). Furthermore, other social disorganization variables, such as residential mobility, ethnic heterogeneity, and urbanism, resulted in conflicting evidence in terms of their strength, direction, and relevance in explaining homicide rates (Chamlin, 1989; Land et al., 1990; Loftin & Parker, 1985; Peterson & Krivo, 1993; Pridemore, 2002; Shihadeh & Ousey, 1998). Some scholars asserted that aggregation bias contributed to the inconsistent findings among homicide studies (Flewelling & Williams, 1999; Pridemore, 2002). Aggregation bias refers to the grouping of individuals, situations, or crimes as if they were homogenous when in reality they were heterogeneous (Hammond, 1973). In the case of homicides, aggregation bias may have

J.M. Pizarro, J.M. McGloin / Journal of Criminal Justice 34 (2006) 195–207

existed because criminologists historically studied this crime by using aggregated rates of all the incidents that occurred in a geographic unit (Lee & Bartkowski, 2004; Pridemore, 2002). Homicide research that employed disaggregation techniques indeed revealed that different ecological factors predicted the occurrence of varying homicide subtypes (see Avakame, 1998; Krivo & Peterson, 2000; Kubrin, 2003; Kubrin & Herting, 2003; Kubrin & Wadsworth, 2003; Kubrin & Weitzer, 2003; Lee & Bartkowski, 2004; Parker, 2001; Parker & Johns, 2002; Peterson & Krivo, 1993). For example, Lee and Bartkowski (2004) revealed a difference in the social structural predictors of adult and juvenile homicide offending. They found that levels of civic engagement in communities affected adult homicide rates, but not juvenile rates. Kubrin's (2003) study also showed that certain community factors were associated with different types of homicides. She found that economic disadvantage predicted general altercation homicides, while residential instability more strongly predicted felony homicides. Moreover, Parker (2001) revealed that racial inequality affected African American but not Caucasian homicide offending. Similarly, Krivo and Peterson (2000) found that concentrated disadvantage and percentage homeownership impacted African American and Caucasian homicide rates differently. Finally, Avakame (1998) found that the correlates of homicides involving intimate partners differed from those involving people who did not know each other. He concluded that the principal predictor of stranger homicides was social disorganization, whereas gender inequality was the dominant predictor of intimate homicides. Research also revealed that the individual and situational characteristics of homicide subtypes differed. For example, Decker and Curry (2002) and Parker and Johns (2002) found that gang homicides differed from other types across such categories as offenders' past criminal history, weapon used, and the situational characteristics of the incident. Parker and Johns (2002) also showed that victims of drug-related homicides were more likely to be male and between the ages of twenty-five and thirty-four than those of other homicide subtypes. In sum, the evidence from the aforementioned studies suggested that homicide was, in fact, not a homogeneous crime, and that variant subtypes often had correspondingly variant etiologies. Therefore, future endeavors should concentrate on specific types of homicides. Accordingly, this study focused on a particular form of homicide, namely gang homicide.

197

The extent and nature of gang homicide One of the driving forces behind the research into gang homicides was the notion that the growth in urban homicides was not unrelated to or independent of the coincident growth of gangs and gang membership in the early 1990s. Indeed, many inquiries illustrated that gang homicides were disproportionately responsible for the homicide rates in multiple cities. For example, in 1991 23 percent of the homicides in Los Angeles and Chicago were gang homicides (Maxson, 1999). Moreover, Chicago witnessed a gang-motivated homicide increase of almost 500 percent from 1987 to 1994 (Block, Christakos, Jacob, & Przybylski, 1996), and nearly 45 percent of all homicides in Los Angeles County were gangmotivated by 1994–95 (Maxson, 1999). Finally, more recent analyses suggested that the increase in California homicides from 1999 to 2001 was entirely due to a rise in gang homicides in Los Angeles County (Tita & Abrahamse, 2004). Such findings were not limited to chronic gang cities. Emerging gang cities showed similar trends. For example, Braga (2003) noted of Boston: “Only about 1,300 gang members—less than 1 percent of their age groups city-wide—in 61 gangs were responsible for at least 60 percent of all the youth homicide in the city” (p. 39). Cohen and Tita (1999) also asserted that gang homicides in Pittsburgh played an important role in increasing homicide rates. Moreover, Wakeling (2003) found that 50 percent of 1997 homicides in Stockton, California were gang-related, primarily driven by conflict between Norteño and Sureño gangs. Finally, Decker and Curry (2002) showed that from 1994 to 1996, gang-related homicides accounted for approximately one-quarter of all recorded homicides in St. Louis. Interestingly, gang homicides were not confined to the aforementioned large cities. Maxson (1999) found, through surveys, that 40 percent of the cities reporting street gangs likewise reported gang homicides—2,166 incidents across 299 locations. Another reason why researchers continued to focus on gang homicides was that they appeared to be qualitatively distinct from non-gang homicides—thus, their unique character warranted disaggregation. Klein and Maxson (1989) compared gang and non-gang homicides, as recorded by both the Los Angeles Sheriff's Department and the Los Angeles Police Department, and found differences with regard to both the setting of and the participants in the homicide. As an example, they found that gang homicides, as compared to nongang homicides, were more likely to occur in public settings, to involve multiple participants, and to

198

J.M. Pizarro, J.M. McGloin / Journal of Criminal Justice 34 (2006) 195–207

involve young men who had no previous contact with each other. They also found (both here and in Maxson, Gordon, & Klein, 1985) that the victim/suspect dyad tended to be intra-racial and minority. These findings were not unique (see for example Bailey & Unnithan, 1994; Block, 1993; Brandt & Russell, 2002; Kennedy, Braga, & Piehl, 1997; Rosenfeld et al., 1999), suggesting that a similar underlying mechanism was at work across geographic locales (Decker & Curry, 2002). Explaining gang homicide Most research on gang homicides ceased at the descriptive level, differentiating the characteristics of gang and non-gang homicides. This was not to suggest that criminologists had not offered explanations for gang homicides—indeed, there were two main orientations. As previously mentioned, one explanation was rooted in macro-level factors, namely those tied to the concept of social disorganization (Curry & Spergel, 1988). The second explanation, conversely, was tied to micro-level group processes. Consequently, in order to provide a better understanding of the etiology of gang homicides, these theoretical orientations were employed and tested in this study.2 Shaw and McKay (1942) suggested that social disorganization, defined as the inability of community structures to realize the common values of residents and to maintain effective social control, was the social phenomenon underlying gang formation and maintenance. Social disorganization theory posited that factors such as economic status, urbanism, family disruption, ethnic heterogeneity, and residential instability led to the dissolution of social bonds and community-based friendships by diminishing informal social control in communities. These factors contributed to the social disorganization in neighborhoods for many reasons. Specifically, they impeded the ability of communities to sustain basic institutional structures that connect individuals to positive roles within society, while at the same time preventing the development of socials bonds and networks among community residents (Peterson, Krivo, & Harris, 2000). Furthermore, these factors negatively impacted community networks, via participation in voluntary organizations hindered surveillance mechanisms in communities, and inhibited the informal social control of youths (Sampson & Groves, 1989). Research on gangs and gang formation suggested that the “gangland represents a geographically and socially interstitial area in the city” (Thrasher, 1927, p. 6). In other words, gangs were prominent in con-

verging areas, where there were breaks in social organization and cohesion. Thus, the essential proposition emanating from social disorganization theory (see Shaw & McKay, 1942; Thrasher, 1927) was that the etiology for gang formation and continuation rested in the social cohesion and normative organization of the community at hand. Additional research that highlighted the concentration of gang homicides in poor neighborhoods characterized by social instability supported this premise (see for example Block, 1991; Kennedy et al., 1997). Curry and Spergel (1988) investigated the notion that social disorganization not only predicted street gang formation and maintenance, but also gang homicide. Upon examining gang and non-gang homicides that occurred in the city of Chicago from 1978 to 1985, they found that gang homicide rates were ecologically distinct from non-gang homicide rates, fitting nicely within a social disorganization framework. Specifically, they operationalized social disorganization by employing: (1) an economic disadvantage factor, which consisted of the percentage of people living below the poverty line, the mortgage investment per dwelling, and the unemployment rate; and, (2) a measure of the concentration of Hispanics in a community. Their research demonstrated that the concentration of Hispanics in a community and poverty were significantly related to gang homicides, adding credence to a focus on macro-level factors when investigating this particular homicide subtype. More recent research, however, complicated matters. Rosenfeld et al. (1999) found that neighborhood disadvantage was a significant predictor of both gangmotivated and gang-affiliated homicides in St. Louis. It also predicted non-gang homicides, however, suggesting that social disorganization might support homicides in general, not gang homicides per se. Further underscoring this notion, the inclusion of the percentage of the population that was African-American washed away the significance of neighborhood disadvantage in the gang homicide models, while it remained in the non-gang homicide model. At the same time, neighborhood instability only emerged as a significant predictor in the non-gang homicide model. Thus, while there were suggestions that community and neighborhood factors played a role in gang homicides, the evidence was far from clear and consistent. In contrast, Decker (1996), who was influenced by Short's (1985) emphasis on the seminal role of the group in gang violence, suggested that the etiology of gang homicide rested in micro-level, social, group processes. Drawing on the work of Short (1974) and Pfautz (1961),

J.M. Pizarro, J.M. McGloin / Journal of Criminal Justice 34 (2006) 195–207

Decker (1996) adopted the orientation that gang violence, including homicide, was an expressive result of collective behavior. He described a seven-step process whereby “threats” from symbolic enemies set an escalating dynamic in motion, culminating in violence and retaliation. Along this vein, Decker and Van Winkle (1996) found that most gang homicides were expressive, spontaneous, or retaliatory, and Klein and Maxson (1989) found that fear of retaliation was a differentiating factor between juvenile gang and non-gang homicides, with it being three times more likely to describe the former than the latter. Finally, Klein and Maxson (1989) also found that much gang violence in Los Angeles was the result of escalating actions among rival gangs. Decker's (1996) offering of the collective behavior explanation was premised on the analysis of gangmotivated violence, as described by gang members in St. Louis. It was important to note that this was not the only form of gang homicides, however. Indeed, gangaffiliated homicides also fell under this label (see for example Maxson & Klein, 1990; Rosenfeld et al., 1999). When Decker and Curry (2002) broadened their focus to include gang-affiliated homicides in St. Louis, they found that many homicides were intra-gang, as well as incidents in which only one actor was a gang member. They suggested that this pattern was due in large part to the fact that gangs in St. Louis were not well organized, and were consequently unable to exert internal social control over violence and homicide. In other words, homicide was not a product of rivalries among gang factions in this analysis. Thus, given that some researchers did not agree that all gang homicides were motivated by the gang, the natural question was whether the escalation hypothesis could frame an explanation of gang homicide in general. It was not unreasonable to assume that this process of a perceived threat to one's collective, or status within the group, which escalates into retaliatory violence, was an etiological factor for gang-related homicides, as well. In short, the extent to which this premise could explain gang-related homicides was unknown. Short and Strodtbeck (1965) suggested that an understanding of gang violence, and therefore gang homicide, must rely on both social processes and situational factors. As such, an investigation into gang homicides would be well served by a focus on both the aforementioned social disorganization factors, as well as measures of Decker's (1996) escalation hypothesis. At the moment, both explanations have some empirical support, but most of it is descriptive in nature. This is not problematic, per se, but there is a clear opportunity to contribute to the literature by investigating the ability of

199

these two premises to explain gang homicides in a multivariate model. At the same time, should these two premises significantly predict gang homicides, their relative efficacy is unknown. An investigation of this nature is especially warranted in the context of an emerging gang city, such as Newark, New Jersey. As compared to chronic gang cities, such as Los Angeles and Chicago, relatively little is known about the diverging nature of gang and non-gang homicides across emerging gang cities (Rosenfeld et al., 1999). Data and methods The City of Newark Newark covers approximately twenty-four square miles and houses nearly 273,000 people (GoNewark, 2004). It is the largest and most heterogeneous city in New Jersey (GoNewark, 2004), with a population that is comprised of African Americans, Caucasians, Africans, Hispanics, Asians, and Middle Easterners (U.S. Bureau of the Census, 2000). Approximately 28 percent of its residents live below the poverty line, and 17 percent of the families with children are female-headed (U.S. Bureau of the Census, 2000). In addition, Newark is one of the most violent municipalities in the state.3 In 2002, the overall violent crime rate in the city was approximately 1,200 per 100,000 (Federal Bureau of Investigation, 2003). Unlike Los Angeles and Chicago, Newark is an emerging gang city—official recognition of a gang presence did not occur until the late 1990s. Currently, the gang problem in Newark revolves around four primary street gangs—the Bloods, Crips, Almighty Latin King and Queen Nation, and the Ñetas. As with other locales, the Bloods and Crips in Newark are constellation gangs (see Decker, 1996), having a variety of sets within these labels. Gang territories in Newark pervade to nearly all of the neighborhoods, and consistent with previous research (see Klein, 1995), gang members are typically cafeteriastyle offenders, engaging in an array of offending behavior (McGloin, 2004). The gangs in Newark are generally disorganized, though there are pockets of cohesive subgroups within the gangs, and certain individuals are quite enmeshed in their respective gang networks (McGloin, 2004, 2005). This chaotic organization is not meant to suggest that street gangs are not problematic in Newark, however. Indeed, as the forthcoming section mentions, more than one-third of the homicides in Newark over the past five years were gang related.

200

J.M. Pizarro, J.M. McGloin / Journal of Criminal Justice 34 (2006) 195–207

The Newark homicide data set Data for this study came from the Greater Newark Safer Cities Initiative (GNSCI), which is a problem solving initiative that began in the late 1990s. GNSCI is a partnership among researchers from Rutgers University, practitioners (e.g., law enforcement, New Jersey court personnel, and members of the social service community), clergy, and community members. The goal of this initiative is to reduce violence in Newark and neighboring jurisdictions.4 Homicide investigation files from the Newark Police Department (NPD) were the primary source of data for GNSCI researchers. This research focused on the 417 homicides that occurred between January 1, 1999 and July 31, 2004. Of the 417 incidents, 69 occurred in 1999, 57 in 2000, 90 in 2001, 65 in 2002, 84 in 2003, and 52 from January 1, 2004 to July 31, 2004. The homicide files contained all incidents reported to and investigated by the NPD's Robbery and Homicide Unit,5 and included information about the incident (i.e., date, time, incident address, premise, precipitating factors, weapon used, relationship between victim and offender), the victim (i.e., demographic information, employment status, gang affiliation, past criminal history), and the suspect(s) (i.e., demographic information, employment status, gang affiliation, past criminal history). Three steps attempted to manage and control the reliability of the data. First, data collection protocols and a data collection instrument helped ensure that all of the research assistants collected the same information from the files. Second, GNSCI research assistants underwent intensive training on the coding schema and data collection protocol before examining the homicide files. Finally, reliability checks were completed after the data collection. Two research assistants coded the homicide narrative independently. When the coding did not concur, which happened in less than 10 percent of the cases, they discussed the case with a senior research associate, and when necessary, with the detective investigating the case, in order to reach consensus. It is worth highlighting that this article essentially offers a case study of a particular city's homicide problem. The decision to focus on Newark was based on three primary factors. First, Newark is an emerging gang city, and comparatively less is known about such locales when compared to chronic gang cities, such as Chicago and Los Angeles. Second, the type of data necessary for this analysis required more depth than typically found in publicly available datasets. The working relationship

between Rutgers University and the Newark Police Department provided access to data files and the relevant information. Third, much of the gang homicide literature emerged from similar city-specific studies, from Chicago (Block, 1991), to St. Louis (Decker & Curry, 2002), to Los Angeles (Klein & Maxson, 1989), to Boston (Braga, 2003), among others. Certainly, the singular focus on Newark limited external validity, but the three aforementioned factors outweighed this limitation. Measures Dependent variable The unit of analysis for this study was the homicide incident, which was disaggregated by gang involvement. Maxson and Klein (1990) highlighted the important fact that police departments and consequently researchers alike define gang homicides in two ways. First, some defined gang homicides as any homicide in which a gang member is affiliated—also known as the “Los Angeles definition.” In such homicides, the motive or reason for the homicide need not be gangrelated: only one of the actors involved must be a gang member. As such, it was a fairly broad definition and led to more generous estimates of gang homicides. In contrast, a gang-motivated homicide must be a direct result of the gang. Otherwise known as the “Chicago definition” such homicides would include incidents of gang-driven retaliation, initiation acts, and defending territories and criminal enterprises (Maxson, 1999). Thus, this was a more restrictive definition, in which gang membership was necessary but not sufficient, and led to more conservative estimates of the rates of gang homicides. These varying definitions impeded one's ability to compare the rate and proportion of gang homicides across cities. For example, Maxson and Klein (1990) revisited data on gang homicides in Los Angeles, and found that when they used the gang-motivated operationalization, the rate of gang homicides reduced by approximately one-half. This was certainly important, but not necessarily relevant to an investigation of the potential differences between gang and non-gang homicides within one city, rather than among cities with varying definitions. The more pertinent question was whether gang-related homicides and gang-motivated homicides were distinct. If so, pooling them together was problematic. After finding minimal distinction between gang-motivated and gang-affiliated homicides, Maxson and Klein (1990) concluded: “the choice of motive or member definition would make little

J.M. Pizarro, J.M. McGloin / Journal of Criminal Justice 34 (2006) 195–207

difference empirically, conceptually, or in policy relevance” (p. 83). Rosenfeld et al. (1999), however, disagreed with this assertion, finding that gang-related and gang-motivated homicides in St. Louis did have some significantly different contextual, participant, and spatial characteristics. In short, there was a potential risk of treating two distinct forms of homicide as homogenous if one failed to distinguish between gang-related and gang-motivated homicide when specifying models. Even so, studies did continue to collapse these two forms into a singular category (see for example, Cohen & Tita, 1999). The Newark Police Department employed the “Los Angeles definition” of gang homicides, such that gang homicides referred to incidents in which one of the actors involved was a gang member at the time of the homicide. Interestingly, in reviewing the homicide files, it appeared that of the 417 homicides under study, only 6 (1.4 percent) homicides could be conservatively coded as actually being gang-motivated. Unlike many other cities, therefore, gang-motivated homicides in Newark appeared to be rare.6 Thus, there simply were not enough cases to warrant further disaggregation within the gang homicide category. Seventy-five (18 percent) cases were excluded from the analysis because it was unknown whether there was any gang involvement. Of the remaining 342 cases, 137 (40 percent) were coded as gang-related (1 = yes, 0 = no). As Tables 1 and 2 illustrate, the demographic and incident characteristics of gang and non-gang homicides in Newark were consistent with previous research (Bailey & Unnithan, 1994; Block, 1993; Brandt & Russell, 2002; Kennedy et al., 1997; Klein & Maxson, 1989; Rosenfeld et al., 1999). In particular, gang homicides were more likely to occur in public settings, to involve a firearm, and involve multiple suspects. In addition, gang homicides were more likely than non-gang homicides to involve victims and suspects who were acquaintances or strangers. There were also differences with regard to the demographic characteristics of victims and suspects. Both victims and suspects of gang homicides were younger than non-gang homicides, and the victims of gang homicides were more likely to be minority males. Independent variables This study employed two primary sets of independent variables. The first variable reflected the social process of the crime, whereas the second set included macro-level explanations of gang homicides. Decker's (1996) escalation or threat hypothesis suggested that

201

Table 1 Incident characteristics of Newark homicides by homicide type Incident characteristics

Location Inside Outside in street Weapon Gun Knife/cutting object Blunt object Personal weapon Other Victim/offender relationship Domestic Friend Acquaintance Stranger Unknown Multiple suspects Yes No

Non-gang N = 205

Gang N = 137

N

%

N

%

93 112

45.4 54.6

29 108

21.2 78.8

108 42 12 31 12

52.7 20.5 5.9 15.1 5.9

123 9 2 3 –

89.8 6.6 1.5 2.2 –

52 26 86 38 3

25.4 12.7 42.0 18.5 1.5

4 5 73 41 14

2.9 3.6 53.3 29.9 10.2

45 155

22.0 75.6

54 76

39.4 55.5

Chisquare 20.95⁎⁎⁎ 53.11⁎⁎⁎

52.21⁎⁎⁎

15.25⁎⁎⁎

⁎⁎⁎ = p b .001.

gang homicides were an expressive result of collective behavior and that most gang homicides were spontaneous or retaliatory. In short, he argued that a perceived threat set a social process in motion that escalated to violent retaliation. As mentioned earlier, there was a question about whether this premise could explain gang-related homicide, as well as gangmotivated homicide. In order to allow for this potential explanation with the former, this perceived threat could not solely stem from a rival gang. In addition, when determining whether this potential explanation differentiated gang from non-gang homicides, the group of interest could not only be a street gang. To date, there was no research that had attempted to determine whether Decker's (1996) premise predicted gang homicide (with the Los Angeles definition) at the multivariate level. Thus, there was no previous operationalization upon which to rely. With this in mind, this study operationalized a variable that captured three primary elements of Decker's (1996) description: threat, escalation, and the collective. In particular, the homicide incident was coded as showing evidence of the escalation hypothesis (yes = 1, no = 0) if the homicide was sparked by a “threat” against a person's group (which could be family, social group, or gang) or the person's status within this group that then escalated into a retaliatory homicide. Of the 342 cases, 208 received a value of 0, 103 received a value of 1, and

202

J.M. Pizarro, J.M. McGloin / Journal of Criminal Justice 34 (2006) 195–207

Table 2 Demographic characteristics of the victims and suspects of Newark homicides by homicide type Victim characteristics

Gender Male Race/ethnicity African American White Hispanic Other Criminal history Yes No Unknown Suspect characteristics Gender Male Race/ethnicity African American White Hispanic Other Unknown Criminal history Yes No Unknown

Victim average age Suspect average age

Non-gang N = 205

Gang N = 137

Chisquare

N

%

N

%

162

79.0

127

92.7

11.73⁎⁎⁎

161 10 34 –

78.5 4.9 16.6 –

119 – 17 1

86.9 – 12.4 0.7

9.84⁎⁎

126 75 4

61.5 36.6 2.0

113 24 –

82.5 17.5 –

18.18⁎⁎⁎

N = 242

N = 159

226

93.8

147

92.5

0.93

196 7 35 1 3

81.0 2.9 14.5 0.4 1.2

143 1 14 – 1

89.9 0.6 8.8 – 0.6

6.90

181 53 8

74.8 21.9 3.3

139 17 3

87.4 10.7 1.9

9.53⁎⁎

Non-gang

Gang

t-statistic

31 27

28 24

−2.29⁎⁎⁎ −3.86⁎⁎⁎

⁎⁎ = p b .01. ⁎⁎⁎ = p b .001.

31 cases had missing values because there was not enough information. Prior macro-level studies of homicides and the gang homicide subtype suggested that aggregate levels of criminal violence were associated with the level of social disorganization in geographic areas, presumably because it impeded social control (Sampson & Groves, 1989; Shaw & McKay, 1942). According to this theory, a number of factors contributed to the social disorganization of a community—among them, poverty and family disruption. This research used the following measures to access these dimensions at the census tract level7: (1) the percentage of families that resided in the incident neighborhood who lived below the U.S. Social Security Administration poverty line; (2) percentage of unemployed people within the neighborhood; (3) percentage of individuals who received public assis-

tance; (4) the percentage of single-parent households with children under eighteen years of age; (5) the percentage of residents who had lived less than five years in the incident census tract; (6) population size; and (7) the number of different ethnic/racial groups that resided in the incident census tract. The first four variables conceptually tapped into the poverty dimension of social disorganization, whereas the latter three addressed other dimensions, such as residential mobility, ethnic heterogeneity, and urbanization. Given that some gang homicide researchers treated poverty as a separate dimension in their analyses (Curry & Spergel, 1988; Rosenfeld et al., 1999), and that it exhibited different prediction patterns than other measures of social disorganization, this study did the same. All the variables were standardized, resulting in two summed scores: (1) poverty, which was a sum of the z-scores for the first four variables (higher values indicated more poverty); and, (2) social disorganization, which was a sum of the z-scores for the latter three variables (higher values indicated more social disorganization). This study also controlled for the percentage of African Americans who resided in the incident census tract. Previous research demonstrated that this was a predictor of gang homicides and had the potential to strip other community variables of their significance (Rosenfeld et al., 1999). In the interest of not overstating the robustness of such predictors, therefore, this variable was included in the models. Analysis Given the dichotomous nature of the dependent variable, logistic regression was used to examine the multivariate relationship between the dependent and independent variables (Pampel, 2000). The advantage of using this technique was that it tested the goodness of fit of the entire model and provided odds ratio calculations in order to determine the relative importance of each independent variable (Pampel, 2000). Model 1 tested whether the escalation hypothesis measure predicted gang homicide, and Model 2 tested the same premise in light of the earlier specified control, percent of African-Americans in the incident census tract. Model 3 assessed whether the poverty and social disorganization measures predicted gang homicide. Then, Model 4 included the percent AfricanAmerican control variable. Finally, Model 5 included all variables in order to assess the relative robustness of these two primary modes of explaining gang homicide.

J.M. Pizarro, J.M. McGloin / Journal of Criminal Justice 34 (2006) 195–207

Results Table 3 presents the results for the five logistic regression models predicting gang homicides in Newark. Model 1 was a simple bivariate logistic regression that addressed the question of whether the escalation variable predicted gang homicides. The results revealed that it was a significant predictor and that a homicide that evidenced the characteristics of a threat, escalation, and the collective (as earlier defined) was more likely to be a gang homicide than a non-gang homicide. In short, evidence of this social process increased the log odds of a homicide being gang-related. Model 2 included the addition of the percentage population who was African American variable. Rosenfeld et al. (1999) found that this variable had the potential to strip community level explanations of their significance when predicting gang homicides. Although there was no evidence that it would operate in the same manner with this micro-level process, there was no evidence to the contrary, either. Thus, it was worthwhile to determine whether this variable also dampened the impact of another predictor of gang homicides. Model 2 showed that percentage African-American was a significant predictor, though the relationship was modest. The logit suggested that higher percentages of African-American residents in the incident census tract increased the log odds of the homicide occurring in that tract being gang-related. The logit was fairly small, however, further evidenced by essentially even odds. Additionally, it did not strip away any of the predictive ability of the escalation variable, which was the more robust predictor of the two.

203

Model 3 investigated whether the two communitylevel variables, poverty and social disorganization, significantly predicted gang homicides in Newark. The results suggested that, while the social disorganization variable was not significant, poverty predicted gang homicides. In short, as the poverty composite for the incident neighborhood increased, so did the log odds of the homicide being gang-related. This result was partly consistent with previous work by Curry and Spergel (1988), who found that poverty was related to the distribution of gang homicides. They also found, unlike here, however, that social disorganization was a predictor of gang homicides. Model 3 was more consistent with Rosenfeld et al. (1999), who found that neighborhood disadvantage (a factor that consisted of poverty rate, public assistance income, and female-headed households) was a significant predictor of both gang and non-gang homicides, whereas neighborhood instability was not. Rosenfeld et al. (1999) also discovered, however, that this significance washed away upon the inclusion of the percent population African American. Model 4 included this variable in order to determine whether that pattern repeated here. While percent African American did strip poverty of some of its significance, it continued to significantly predict ganghomicide. Model 5 placed all of the variables into the logistic regression model, investigating the relative robustness of the two theoretical explanations of gang homicide. Until Model 5, the final conclusion would be that both the escalation hypothesis and poverty explained gang homicides in Newark, but their relative strength as predictors would be unknown. Interestingly, whereas

Table 3 Logit coefficients, standard errors, and odds ratios for the logistic regression models predicting gang homicides Variable Constant Escalation

Model 1 −1.023 (.157)⁎⁎⁎ 1.396 (.255)⁎⁎⁎ 4.040

Model 2 −1.820 (.375)⁎⁎⁎ 1.348 (.258)⁎⁎⁎ 3.851

Poverty Social disorg. Percent African American −2 log likelihood Pseudo R2 N

379.558⁎⁎⁎ .131 311

.012 (.005)⁎ 1.012 373.365⁎⁎⁎ .155 311

Model 3 −.404 (.112)⁎⁎⁎ .104 (.036)⁎⁎ 1.109 −.058 (.056) .944 450.935⁎⁎ .037 342

Model 4 −1.207 (.455)⁎⁎ .077 (.039)⁎ 1.080 .045 (.078) 1.046 .011 (.006)+ 1.011 447.477⁎⁎ .051 342

Note: coefficients are presented first, with standard errors in parentheses, and odds ratios below these values. ⁎ = p b .05. ⁎⁎ = p b .01. ⁎⁎⁎ = p b .001. + = p b .10.

Model 5 −2.124 (.526)⁎⁎⁎ 1.351 (.262)⁎⁎⁎ 3.860 .060 (.043) 1.062 .113 (.086) 1.120 .016 (.007)⁎ 1.016 368.412⁎⁎⁎ .174 311

204

J.M. Pizarro, J.M. McGloin / Journal of Criminal Justice 34 (2006) 195–207

the escalation variable retained significance in Model 5 and remained a robust predictor with an odds ratio of 3.86, the poverty variable lost significance entirely. Thus, when both potential explanations were part of the model, only one emerged as a significant predictor. The pseudo-R2 further underscored this notion since the value increased from .051 in Model 4 to .174 in Model 5, simply due to the inclusion of the escalation hypothesis. Discussion Although numerous studies examined the distinction between gang and non-gang homicides, there was nonetheless a continued need for research in this domain. Specifically, few studies investigated the etiological differences between these homicides at the multivariate level. Indeed, only a handful of investigations attempted to examine the robustness of the primary explanations of gang homicides—social disorganization (Curry & Spergel, 1988) and Decker's (1996) collective behavior hypothesis. Thus, this inquiry served as a contribution in two ways. First, it tested the explanatory power of the main gang homicide explanations at the multivariate level. Second, it determined the relative efficacy of these etiological premises in order to assess if one explanation was more robust than the other. Similar to previous studies, this study found that there were significant differences between gang and non-gang homicides at the incident level. Like gang homicides in various cities (e.g., Los Angeles, St. Louis, and Chicago), incidents in Newark were more likely to occur in public settings, involve firearms, involve multiple suspects, and have minority young male victims and/or suspects than non-gang incidents. At the multivariate level, the findings suggested that gang homicides were not only different from non-gang homicides at the incident level, but that the predictors of this homicide type differed, as well. Specifically, homicides precipitated by the operationalization of Decker's (1996) escalation hypothesis were more likely to be gang-related. This was important because the definition of the escalation process was expanded to allow for collectives other than a gang. In short, this study did not ask whether gang rivalries described gang homicide, but asked whether social processes precipitated by a threat against one's group or status within a group (social, kinship, or gang) actually explained this homicide subtype. Conversely, the capacity of macro-level factors to explain gang homicides was more complicated. Whereas the social disorganization measure did not predict gang homicide (consistent with Rosenfeld et al., 1999),

poverty was a significant predictor. When both potential explanations were entered into the same model, however, only the micro-level escalation hypothesis retained its significance. Thus, the findings suggested that Decker's (1996) threat/escalation/collective behavior hypothesis explained gang homicides in Newark, New Jersey, whereas the macro-level variables of interest did not. This fits nicely with Short's (1985) suggestion to focus empirical attention on micro-level group processes. Short (1985) argued that groups (and gangs) were social systems, and, as such, interactions among individuals take on added importance. When beginning his work with Strodtbeck, the guiding theories of the time were decidedly macro-level, orienting researchers towards strain, neighborhood-based opportunity, and subcultural value systems. Instead of following this tradition, they focused on social interactions within the group, highlighting the importance of micro-level processes in research (Short & Strodtbeck, 1965). Given the potential importance of the “group” in the context of gang homicide, this research coupled such a focus with the more traditional macro-level predictors of homicide. The fact that the operationalization of Decker's (1996) social process hypothesis of gang violence emerged as the primary explanatory factor of gang homicides in Newark underscored Short's (1985) assertion that this level of explanation can lead to increased knowledge. The results of this study collectively emphasized three important points. First, this study highlighted the importance of disaggregating homicides. Previous homicide studies revealed that this crime should not be studied as if it were one offense because doing so would potentially result in aggregation bias. Specifically, contemporary homicide research emphasized the heterogeneity of this crime and suggested that homicide studies should accordingly examine this crime disaggregated by subtype. Along this vein, this study revealed that the incident level characteristics of gang and nongang homicides differed and that the strength and the relevance of homicide explanatory variables varied with respect to type of incident. Thus, this research joined others (see Avakame, 1998; Krivo & Peterson, 2000; Kubrin, 2003; Kubrin & Herting, 2003; Kubrin & Wadsworth, 2003; Kubrin & Weitzer, 2003; Lee & Bartkowski, 2004; Parker, 2001; Parker & Johns, 2002; Peterson & Krivo, 1993) in attesting to the notion that different types of homicides potentially have varying etiologies and accordingly should be studied separately. Second, this study demonstrated that micro level processes were better able to explain gang homicides than macro level social disorganization measures were.

J.M. Pizarro, J.M. McGloin / Journal of Criminal Justice 34 (2006) 195–207

One possible reason for this finding was that homicides, in general, concentrated in socially disorganized areas (Land et al., 1990). Even so, this did not mean that macro-level factors hold no value in an understanding of gang homicide. Indeed, the question of why this escalation process characterizes gang violence remains. Is there a subculture that increases the probability of this process? If so, is it a normative system defined by the gang, or might it be neighborhood-based? Future gang violence research should delve into this question because some scholars argued that micro-level dynamics were structurally conditioned (Anderson, 1999; Wilson, 1996). According to these scholars, concentrated disadvantage not only deprived neighborhoods of institutions of social control, but also increased social isolation among residents, which impeded communication and interfered with their capacity to pursue common values (Wilson, 1996). This, in turn, led the residents of these areas to adapt cultural mechanisms that enabled their survival, including aggressive behavior (Anderson, 1999; Wilson, 1996). Finally, while research that shows gang homicides stemmed from gang rivalries was seen as support for Decker's (1996) social process premise (Klein & Maxson, 1989), this study showed that Decker's hypothesis also explained gang homicides that were not necessarily precipitated by gang rivalries. Indeed, this study suggested that Decker's (1996) collective behavior theory of gang homicides was not only valid in explaining gang-motivated violence, but that it was also valid in explaining gang-related, or affiliated, incidents. This was important since some cities are characterized by disorganized intra-gang homicides. As such, these findings take on added importance (Decker & Curry, 2002). They suggested that Decker's (1996) premise held utility for both the Chicago and the Los Angeles definition of gang homicides, underscoring the robustness of the social process explanation. In conclusion, this study addressed some gaps in the homicide and gang literature, but at the same time, it also holds implications for future research. In particular, future studies should replicate the models employed here in other locales. Newark is an emerging gang city with a particular gang problem. While the incident level characteristics of gang homicides were similar to those reported elsewhere (Bailey & Unnithan, 1994; Block, 1993; Brandt & Russell, 2002; Kennedy et al., 1997; Klein & Maxson, 1989; Rosenfeld et al., 1999), and previous organizational analyses of the gangs in Newark were also consistent with other work (McGloin, 2004, 2005), external validity was nonetheless limited. It is therefore important to replicate this

205

study in other locations (both chronic and emerging gang cities) in order to determine the generalizability of these findings. Acknowledgements The authors would like to thank the Police Institute at Rutgers-Newark, the Newark Police Department Homicide Squad, and the anonymous reviewers for their helpful suggestions. Notes 1. This study examined homicides that occurred in the city of Newark, New Jersey from January 1, 1999 to July 31, 2004. 2. Again, this study employed social disorganization and the collective behavior hypothesis because they were the primary explanations of gang homicide offered by the literature. Surely, other theoretical explanations of homicide exist. This study's primary goal, however, was to investigate the explanations most prevalent in the gang homicide literature. 3. In 2002, Newark ranked as the seventh most violent city in the state with a rate of 606 violent crimes per 50,000 population. East Orange and Irvington, which are adjacent to Newark, were the most violent cities in New Jersey with rates of over 1,000 violent crimes per 50,000 residents (the rate was calculated based on a population of 50,000 because East Orange's and Irvington's respective populations are less than 100,000). 4. For more information regarding the Greater Newark Safer Cities Initiative go to http://policeinstitute.org and click on problem solving activities. 5. Technically, Newark Police Department does not investigate all homicides that occur in the city. The police department does not have jurisdiction, for example, on the rare occasion when a homicide occurs in a county park or on a state road located within the city borders. 6. This might be related to the fact that the gangs in Newark were rather disorganized (McGloin, 2004, 2005). As Decker and Curry (2002) noted, this was related to intra-gang homicide, rather than organized homicides based on gang rivalries. 7. Newark is divided into ninety census tracts (GoNewark, 2004; U.S. Bureau of Census, 2000). Census tracts were chosen as a proxy for neighborhood because the city of Newark experienced impressive change in the last eight years. The city underwent a private residential construction boom, as well as the relocation of residents, because small town houses replaced many of the old high-rise public housing buildings (GoNewark, 2004). As a result, new neighborhoods began to emerge throughout the municipality and the geographic borders of existing neighborhoods changed. Therefore, census tracts provided a more stable geographic measure for neighborhood.

References Anderson, E. (1999). Code of the street: Decency, violence, and the moral life of the inner city. New York: W.W. Norton and Company, Inc. Avakame, E. F. (1998). How different is violence in the home? An examination of some correlates of stranger and intimate homicide. Criminology, 36, 601−632.

206

J.M. Pizarro, J.M. McGloin / Journal of Criminal Justice 34 (2006) 195–207

Bailey, G. W., & Unnithan, N. P. (1994). Gang homicides in California: A discriminant analysis. Journal of Criminal Justice, 22, 267−276. Block, C. R. (1991). Gang homicide in Chicago: Patterns over time, area of city, and type of victim. Paper presented at the meeting of the Midwestern Criminal Justice Association, Chicago, IL. Block, C. R. (1993). Lethal violence in the Chicago Latino community. In A. V. Wilson (Ed.), Homicide: The victim/offender connection (pp. 267−342). Cincinnati, OH: Anderson. Block, C. R. (1996). Street gangs and crime (Research Bulletin). Chicago: Illinois Criminal Justice Information Authority. Block, C. R., Christakos, A., Jacob, A., & Przybylski, R. (1996). Street gangs and crime: Patterns and trends in Chicago. Chicago: Illinois Criminal Justice Information Authority. Braga, A. A. (2003). Serious youth gun offenders and the epidemic of youth violence in Boston. Journal of Quantitative Criminology, 19, 33−54. Brandt, G. A., & Russell, B. (2002). Differentiating factors in gang and drug related homicide. Journal of Gang Research, 9, 23−40. Chamlin, M. B. (1989). A macro social analysis of the change in robbery and homicide rates: Controlling for static and dynamic effects. Sociological Focus, 22, 275−286. Cohen, J., & Tita, G. (1999). Diffusion in homicide: Exploring a general method for detecting spatial diffusion processes. Journal of Quantitative Criminology, 15, 451−493. Curry, G. D., & Decker, S. H. (1998). Confronting gangs: Crime and community. Los Angeles: Roxbury. Curry, G. D., & Spergel, I. A. (1988). Gang homicide, delinquency, and community. Criminology, 26, 381−403. Decker, S. H. (1996). Collective and normative features of gang violence. Justice Quarterly, 13, 243−264. Decker, S. H., & Curry, G. D. (2002). Gangs, Gang homicides, and gang loyalty: Organized crimes or disorganized criminals. Journal of Criminal Justice, 30, 343−352. Decker, S. H., & Van Winkle, B. (1996). Life in the gang: Family, friends, and violence. Cambridge, UK: Cambridge University Press. Federal Bureau of Investigation. (2003). Crime in the United States: 2002 uniform crime reports. Washington, DC: U.S. Department of Justice. Flewelling, R. L., & Williams, K. R. (1999). Categorizing homicides: The use of disaggregated data in homicide research. In M. D. Smith & M.A. Zahn (Eds.), Homicide: A sourcebook of social research (pp. 96−106). Thousand Oaks, CA: Sage. Fox, J. A., & Zawitz, M. W. (2002). Homicide trends in the United States. Washington, DC: U.S. Department of Justice, Bureau of Justice Statistics. GoNewark. (2004). Neighborhoods. Retrieved May 10, 2004, from http:// www.ci.newark.newark.nj.us/Neighborhoods/Neighborhoods.htm Hammond, J. L. (1973). Two sources of error in ecological correlation. American Sociological Review, 38, 764−777. Howell, J. C. (1998). Youth gangs: An overview. Washington, DC: Office of Juvenile Justice and Delinquency Prevention. Howell, J. C. (1999). Youth gang homicide: A literature review. Crime and Delinquency, 45, 208−241. Kennedy, D. M., Braga, A. A., & Piehl, A. M. (1997). The (un)known universe: Mapping gangs and gang violence in Boston. In D. Weisburd & T. McEwen (Eds.), Crime mapping and crime prevention (pp. 219−262). Monsey, NY: Criminal Justice Press. Klein, M. W. (1995). The American street gang. New York: Oxford University Press.

Klein, M. W., & Maxson, C. L. (1989). Street gang violence. In M. E. Wolfgang & N. Weiner (Eds.), Violent crime, violent criminals (pp. 198−234). Beverly Hills, CA: Sage. Krivo, L. J., & Peterson, R. D. (2000). The structural context of homicide: Accounting for racial differences in process. American Sociological Review, 65, 547−559. Kubrin, C. E. (2003). Structural covariates of homicide rates: Does type of homicide matter? Journal of Research in Crime and Delinquency, 40, 139−179. Kubrin, C. E., & Herting, J. R. (2003). Neighborhood correlates of homicide trends: An analysis using growth curve modeling. Sociological Quarterly, 44, 329−350. Kubrin, C. E., & Wadsworth, T. (2003). Identifying the structural correlates of African American killings: What can we learn from data disaggregation? Homicide Studies, 7, 3−35. Kubrin, C. E., & Weitzer, R. (2003). New directions in social disorganization theory. Journal of Research in Crime and Delinquency, 40, 374−402. Land, K. C., McCall, P. L., & Cohen, L. E. (1990). Structural covariates of homicide rates: Are there any invariances across time and social space. American Journal of Sociology, 95, 922−963. Lee, M. R., & Bartkowski, J. P. (2004). Civic participation, regional subculture, and violence: The differential effects of secular and religious participation on adult and juvenile homicide. Homicide Studies, 8, 5−39. Loftin, C., & Parker, R. N. (1985). An errors-in-variable model of the effects of poverty and urban homicide rates. Criminology, 23, 269−285. Maxson, C. L. (1999). Gang homicide. In D. Smith & M. Zahn (Eds.), Homicide studies: A sourcebook of social research (pp. 239−254). Newbury Park, CA: Sage. Maxson, C. L., Gordon, M. A., & Klein, M. W. (1985). Differences between gang and nongang homicides. Criminology, 23, 209−222. Maxson, C. L., & Klein, M. W. (1990). Street gang violence: Twice as great, or half as great? In C. R. Huff (Ed.), Gangs in America (Vol. 1). Newbury Park, CA: Sage. McGloin, J. M. (2004). Associations among criminal gang members as a defining factor of organization and as a predictor of criminal behavior: The gang landscape of Newark, New Jersey. Ann Arbor, MI: University of Michigan. McGloin, J. M. (2005). Policy and intervention considerations of a network analysis of street gangs. Criminology and Public Policy, 4 (3), 607−636. Messner, S. F., & Rosenfeld, R. (1999). Social structure and homicide. In M. D. Smith & M.A. Zahn (Eds.), Homicide: A sourcebook of social research (pp. 27−41). Thousand Oaks, CA: Sage. Miller, W. B. (2001). The growth of youth gang problems in the United States: 1970-1998. Washington, DC: Office of Juvenile Justice and Delinquency Prevention. Office of Juvenile Justice and Delinquency Prevention. (1998). 1998 National Youth Gang Survey. Washington, DC: Author. Pampel, F. C. (2000). Logistic regression: A primer. Thousand Oaks, CA: Sage. Parker, K. F. (2001). A move toward specificity: Examining urban disadvantage and race-and relationship-specific homicide rates. Journal of Quantitative Criminology, 17, 89−110. Parker, K. F., & Johns, T. (2002). Urban disadvantage and types of race specific homicide: Assessing the diversity in family structures in the urban context. Journal of Research in Crime and Delinquency, 39, 277−303. Peterson, R. D., & Krivo, L. J. (1993). Racial segregation and urban Black homicide. Social Forces, 71, 1001−1026.

J.M. Pizarro, J.M. McGloin / Journal of Criminal Justice 34 (2006) 195–207 Peterson, R. D., Krivo, L. J., & Harris, M. A. (2000). Disadvantage and neighborhood violent crime: Do local institutions matter? Journal of Research in Crime and Delinquency, 37, 31−63. Pfautz, H. (1961). Near-group theory and collective behavior: A critical reformulation. Social Problems, 9, 167−174. Pridemore, W. A. (2002). What we know about social structure and homicide: A review of the theoretical and empirical literature. Violence and Victims, 7, 127−151. Rosenfeld, R. B., Bray, T., & Egley, H. A. (1999). Facilitating violence: A comparison of gang-motivated, gang-affiliated, and non-gang youth homicides. Journal of Quantitative Criminology, 15, 495−516. Sampson, R. J., & Groves, W. B. (1989). Community structure and crime: Testing social disorganization theory. American Journal of Sociology, 94, 774−802. Shaw, C. R., & McKay, H. D. (1942). Juvenile delinquency and urban areas. Chicago: University of Chicago Press. Shihadeh, E. S., & Ousey, G. C. (1998). Industrial restructuring and violence: The link between entry-level jobs, economic deprivation, and Black and White homicide. Social Forces, 77, 185−206.

207

Short, J. F. (1974). Collective behavior, crime, and delinquency. In D. Glaser (Ed.), Handbook of criminology (pp. 403−449). New York: Rand McNally. Short, J. F. (1985). The level of explanation of problem. In R. Meier (Ed.), Theoretical methods in criminology (pp. 51−72). Beverly Hills, CA: Sage. Short, J. F., & Strodtbeck, F. L. (1965). Group processes and gang delinquency. Chicago: University of Chicago Press. Thrasher, F. (1927). The gang: A study of 1313 gangs in Chicago. Chicago: University of Chicago Press. Tita, G., & Abrahamse, A. (2004). Gang homicide in L.A., 1981-2001 (Perspectives on Violence Prevention No. 3). Sacramento: California Attorney General's Office. U.S. Bureau of the Census. (2000). Fact sheet-Newark, New Jersey. Retrieved on May 22, 2004, from http://factfinder.census.gov Wakeling, S. (2003). Ending gang homicide: Deterrence can work (Perspectives on Violence Prevention No. 1). Sacramento: California Attorney General's Office. Wilson, W. J. (1996). When work disappears: The world of the new urban poor. New York: Random House Inc.