Geographic Networks of Gun Violence: Exploring Hot Spot Connectivity Through Ballistic Evidence and Social Network Analysis

Abstract

Objectives

While crime-and-place research consistently demonstrates that a minority of places account for most of the crime in a given jurisdiction, much less is known about the extent to which these hot spots are interconnected. Informed by the social network analysis literature, we explore whether gun violence networks occur in spatial forms, with hot spots connected by distinct firearms used in the commission of crime.

Methods

We construct a network where ballistic evidence from crime guns ties together street segments in Kansas City, Missouri. Data from the National Integrated Ballistic Information Network is used to construct connections between street segments that contain ballistic evidence associated with the same crime gun.

Results

Findings indicate that street segments classified as gun violence hot spots were more likely to be connected to one another than would be expected by chance (i.e., hot spot homophily). This suggests hot spots do not operate as self-contained entities but are components of a larger geographic network of gun violence.

Conclusion

The current study suggests that crime gun linkages can enhance our understanding of how gun violence hot spots not only share certain structural features but also are connected in ways suggestive of spatial and social interdependence.

Keywords

gun violence hot spots national integrated ballistic information network (NIBIN)social network analysis exponential random graph models (ERGMs)

Introduction

The criminology of place has emerged as a key perspective over recent decades. This line of research consistently finds that a minority of geographic hot spots account for the majority of crime in a given jurisdiction (Weisburd 2015; Weisburd et al. 2024). Much less is known about the interconnected nature of hot spots with only a few studies exploring the spatial overlap of hot spots of different crime types (Connealy 2023; Connealy and Piza 2019; Haberman 2017) as well as the social proximity between high-crime places (H. Wang et al. 2016; Kelling et al. 2021; Papachristos et al. 2024).

We believe a key overlooked issue in crime-and-place research is how hot spots are interconnected within the spatial networks that shape gun violence. Social network analysis demonstrates that membership within co-offending networks is highly associated with an increased risk of both criminality and victimization (Di Méo 2023; Fox et al. 2021; Papachristos et al. 2015a,b; Van Deuren et al. 2025), with network and geographic factors jointly influencing patterns of violence (Brazil et al. 2025; Papachristos et al. 2013; Schaefer 2012; Schaefer et al. 2014). We argue that gun violence networks may also occur in spatial forms, with hot spots connected by distinct firearms used in the commission of crime.

The current study empirically explores the concept of geographic networks of gun violence in Kansas City, Missouri. We leverage data from the National Integrated Ballistic Information Network (NIBIN) to construct connections between street segments that contain ballistic evidence associated with the same crime gun. Using this information, we construct a network where the spatial movement of crime guns is represented by edges that tie together microgeographic units, or nodes. Our analyses find that the spatial network exhibits higher levels of homophily on gun violence concentration than what would be expected by random chance (i.e., hot spots are more likely to be connected to other hot spots). This hot spot homophily suggests high-crime places do not operate as self-contained entities but are elements of a larger geographic network. As such, crime hot spots not only share structural similarities, as prior literature posits, but they may also be connected more directly in ways that suggest spatial and social interdependence. We argue theoretical and policy insights can be gleaned by focusing on the interconnections between hot spots.

Review of Relevant Literature

Criminology of Place and Crime Concentration

Criminology of place research reports that crime is not evenly distributed throughout jurisdictions but is highly concentrated within distinct places (Pierce et al. 1988; Sherman et al. 1989). A recent systematic review of 47 studies reported a median of 50% and 25% of crimes occurring within 4.5% and 1.25% of street segments, respectively (Weisburd et al. 2024). Typically conceptualized as hot spots, areas with high crime concentrations are observed across a large range of settings and crime types (Lee et al. 2017; Weisburd 2015; Weisburd et al. 2024).

For gun violence, in particular, there is consistent evidence that these incidents concentrate in geographic areas, though some nuance should be acknowledged. For instance, between 1980 and 2008, less than 3% of street segments and intersections in Boston generated more than half of the city's overall shooting incidents (Braga et al. 2010). Gun violence becomes increasingly concentrated during crime spikes. At the start of the COVID-19 pandemic, as many as 55% of the increased number of shootings occurred in the top decile of census block groups in several metropolitan areas (MacDonald et al. 2022). However, when gun homicides are isolated from overall shooting incidents, there is less evidence that gun homicides significantly concentrate in hot spots—a finding that replicates in multiple cities (Chalfin et al. 2021; Hatten 2022).

Hot Spot Connectivity

Much of the extant empirical research on crime and place operates under an implicit assumption that hot spots exist in isolation, overlooking how these micro-geographic units are nested within interconnected geographies. This mirrors research focused on the neighborhood level, with spatial contiguity and proximity the most commonly explored sources of crime diffusion (Aleinzi et al. 2025). Although there are notable exceptions (Davies and Johnson 2015; Haberman 2017; Ratcliffe and Rengert 2008), this oversight remains a key limitation given the tenets of environmental criminology. For example, crime pattern theory conceptualizes a “pattern” as both the explicit geography of an area and the social interconnectedness of features within the environmental backcloth (Brantingham and Brantingham 1993). The interconnectedness of places influences how crime opportunities are presented to motivated offenders throughout the urban landscape (Brantingham and Brantingham 2008).

Promising avenues of crime-and-place research are beginning to explore the interlinkages between high-crime places. One line of this work considers the extent to which hot spots of different crime types overlap, generally finding that hot spots are unique to specific types of crime. Haberman (2017) finds minimal cooccurrence across the hot spots of 11 crime types in Philadelphia. Connealy and Piza (2019) uncover a similar lack of overlap between types of robbery hot spots in Denver, with less than 3% of high-risk places being a hot spot for more than one of four robbery categories (street, commercial, carjacking, and residential). Only 5% of the identified street segments in Indianapolis were multi-crime hot spots, though these areas accounted for 27.5% of all crime incidents (Connealy 2023). This research showcases that there is limited potential for different types of geographic hot spots to overlap and influence one another, demonstrating complex patterns of spatial interconnection among these locations. Somewhat mixed results were found by Andresen and Linning (2012) when analyzing hot spots of several individual crime types across three spatial units (census tracts, dissemination areas, and street segments) in Ottawa and Vancouver. In five of six tests, none of the individual crime types were significantly correlated, indicating little overlap in where crimes concentrate. Given these findings, Andresen and Linning (2012) argue that hot spot research based on aggregate crime types may not provide meaningful information in regards to either theory or policy, which reflects the highly crime-specific nature of criminal opportunities (Clarke 1995, 2012).

Near repeat analysis applies the concept of contagion to explain how crime concentrates across both place and time. Near repeat analysis assumes that offenders apply knowledge gained from the successful completion of a criminal act (e.g., levels of natural surveillance, entry and escape routes, routine activities of capable guardians) to other targets close in spatial proximity (Johnson 2014). While near repeat analysis originally applied to property crime, near repeat patterns have more recently been observed for gun violence in various study settings (Ratcliffe and Rengert 2008; Sturup et al. 2018, 2020; Wells et al. 2012). A recent study incorporated NIBIN data to contextualize near repeat shooting patterns in Detroit, MI, finding that over half of shooting incidents in near repeat patterns involved a firearm with a prior criminal history (i.e., NIBIN evidence from the firearms was collected from at least two different crime scenes) (De Biasi et al. 2026).

Finally, recent research demonstrates that the layout of street networks can determine how crime opportunities are arranged in physical space. As explained by Davies (2023), while places themselves provide settings for crime, whether individuals encounter crime opportunities depends on how places are configured and visited during travel. In this sense, research on street networks reflects core constructs of routine activities (Cohen and Felson 1979) and crime pattern theory (Brantingham and Brantingham 1993, 2008). Given that street networks simultaneously encourage and restrict travel in urban areas, they can determine how crime events distribute across the landscape. Research in this area finds that a street segment's number of connections to other street segments is positively associated with reported crime levels (Birks and Davies 2017; Davies and Johnson 2015; Haberman and Kelsay 2021; Johnson and Bowers 2010).

A Social Network Approach to Crime and Place

As research continues to grapple with how to conceptualize and measure the social phenomena that link physical spaces defined by crime hot spots, we argue a social network approach can help further develop the crime-and-place perspective. Applications of social network theory and methods have become increasingly common in criminology (Faust and Tita 2019). Recent work applies techniques from social network analysis to consider co-offending (Papachristos et al. 2015a,b), illicit criminal networks (Calderoni 2012; Morselli 2009), and gang rivalries (Tita and Greenbaum 2009; Tita and Radil 2010).

Many applications of networked criminology also include spatial factors to unpack how geographies can both shape and be shaped by the social connectivity of residents. Schaefer (2012) found that youth are more likely to co-offend with peers from census tracts with similar sociodemographic profiles, net of the physical distance between their communities. In their study of Dutch outlaw motorcycle gangs, Van Deuren et al. (2025) found that similarity in age, rank within the motorcycle club, and shared club membership predict frequency of co-offending, while distance between motorcycle clubhouses demonstrates no significant effects. Co-offending behavior itself is also a key driver of crime in urban environments. Papachristos et al. (2024) uncovered that neighborhoods in New York City with greater densities of local ties between offenders have higher shooting rates and that neighborhoods sharing social connections between offenders report similar levels of violence.

A recent scoping review by Brazil et al. (2025) demonstrates how network analysis principles enable social scientists to measure the interconnectedness of geographies and test how network factors influence a range of health outcomes, including crime victimization. Incorporating social network perspectives in spatial analysis acknowledges that neighborhood boundaries are a consequence of both spatial and social proximity (Brazil et al. 2025; Hipp et al. 2012; Sampson 2012). When neighborhoods are defined solely by geographic contiguity, they are less likely to capture residents who report similar perceptions of neighborhood crime and disorder than when population mobility patterns are also used to inform these boundaries (Boessen et al. 2014; Boessen and Hipp 2022; Hipp et al. 2012). Brazil et al. (2025) identified 32 articles published between 2014 and 2023 that applied social network approaches to the analysis of geographic areas, with half published between 2022 and 2023 and over a third analyzing patterns of crime. While many of these studies rely on contemporary data sources to measure ties between neighborhoods, such as cell phone or social media data, Brazil et al. (2025) recommend that “future work should also investigate other types of ties beyond those created via daily mobility” (p. 17).

Scholarship not included in Brazil et al. (2025) also applies novel data sources to model interconnectedness between neighborhoods to analyze local crime rates. H. Wang et al. (2016) find that social proximity implied by taxi flows (i.e., taxi drop-offs originating from another neighborhood) increases precision in neighborhood crime estimates in Chicago. Johnson and Roman (2025) use cellphone-based mobility data to measure the impact of drug market activity on gun violence across three US cities. Results indicate that network exposure to drug markets is associated with local gun violence in each city, demonstrating how drug markets can produce spillover effects of violence beyond immediate spatial surroundings. Aleinzi et al. (2025) analyzed inter-neighborhood correlations of shootings in Chicago, focusing on three network processes: spatial contagion, mobility patterns (measured through cell phone GPS data), and neighborhood homophily, which they define as the interconnectedness of neighborhoods through similarity of sociodemographic characteristics. Aleinzi et al. (2025) found that while spatial proximity of neighborhoods plays a role in explaining shooting correlations, both mobility flows and neighborhood homophily significantly drive shooting correlations between spatially distant pairs of neighborhoods. Forati et al. (2023) applied social network analysis to analyze flows of overdose patients from communities of origin to destination in Milwaukee, WI. They found distinct sociodemographic characteristics between origin and destination communities and that non-discordant overdose deaths (i.e., overdoses for which origin and destination neighborhoods are the same) were more likely to involve opioids other than fentanyl or heroin and result from suicide.

Although this prior work emphasizes that scholars should account for the social connectivity between neighborhoods, research has yet to consider the social connections between geographic areas designated as hot spots. It remains unknown whether patterns of social connectivity link together geographic concentrations of violence at disproportionate rates. This body of research further does not follow the recommendation of Brazil et al. (2025) to use a wider range of data to measure ties between geographic areas, with spatial network studies predominantly using data on mobility patterns. For these reasons, we consider NIBIN data as a novel data source that can help fill these gaps in the literature, as described subsequently.

Crime Gun Linkages Within a Spatial Network

NIBIN, managed by the Bureau of Alcohol, Tobacco, and Firearms (ATF), collects data on the ballistic imaging of spent projectiles and cartridges collected at crime scenes by police or test-fired from recovered firearms.¹ Such evidence is submitted to an integrated ballistic identification system (IBIS), which conducts scans on the unique tool marks (e.g., ejector marks, firing pin, breech face marks) found on the item. Tool marks are then compared to previously scanned evidence to identify ballistic items fired from the same firearm (King et al. 2013). The introduction of 3D imaging greatly improved the efficiency and accuracy of “hit” identification in NIBIN since the early 2000s (Braga and Pierce 2011). During the collection process, data is also gathered on the event (e.g., the specific crime or type of incident, the location, time of day), which allows investigators to further contextualize the gun evidence and pinpoint the guns most commonly discharged within a jurisdiction (Joyce Foundation 2024; King et al.2013).

The movement of individuals or groups of people is a necessary condition for ballistics associated with the same gun to be recovered across multiple geographies. Prior research considers the recovery of ballistics from the same gun at different places as strongly indicative of repeat offenders inflicting violence across different areas and/or crime guns being shared within individual criminal networks (De Biasi et al. 2026), which reflects prior research on prolific offenders and co-offending networks (Ciomek et al. 2020; Hureau and Braga 2018; Roberto et al. 2018). Prior work discusses firearms as “tools of the trade” used for protection and to resolve disputes, which facilitates their movement between areas (Blumstein and Cork 1996, p. 10).

Conceptualizing the spatial movement of ballistic evidence as a gun violence network offers an informative complement to prior work that constructs geographic networks of co-offending. Co-arrest data relies on incidents known to and documented by law enforcement, which can restrict the number of ties present in co-arrest networks (Bright et al. 2021; Gill and Fox 2022). Underreporting is less likely to bias data on the criminal trajectories of firearms since the collection of ballistic evidence requires fewer investigative efforts than the identification and arrest of a suspect and can also incorporate incidents that did not involve victims (e.g., calls of sounds of shots fired; King et al. 2013). Furthermore, since most co-offending research relies on arrestees’ home addresses to conceptualize ties between places (e.g., Papachristos et al. 2024; Papachristos and Bastomski 2018; Schaefer 2012), it cannot determine whether social phenomena link together geographies where crime actually occurs. Recovered ballistic evidence provides a more ambient perspective on how individuals and groups of offenders traverse physical space by identifying explicit linkages between the places where crime guns were used.

Despite its potential to capture understudied social processes, NIBIN data remains an underutilized resource for researchers studying the movement and spread of firearm evidence (King et al. 2013). A recent study by Gill and Fox (2022) represents a noteworthy exception that incorporates NIBIN data and incident reports from a large urban county in the Pacific Northwest of the United States. By adopting a network approach, they found that nearly 5% of weapons were responsible for 80% of gun events, with each incident connected to an average of 7.7 others via shared crime guns. However, to our knowledge, empirical literature has yet to consider whether the criminal trajectories of firearms produce an interconnected network of geographic gun violence hot spots.

Scope of the Current Study

The potential for theoretical contributions to the crime and place literature, the scarcity of comprehensive studies utilizing novel NIBIN data, and the unique advantages of this dataset motivate our exploration of how gun violence hot spots are situated within geographic networks. We explore whether hot spots are more likely to be linked to other hot spots in spatial networks of gun violence, a pattern which would be characterized as evidence for homophily in the social network literature (Khanam et al. 2023). Social network research traditionally describes homophily as the human tendency to associate with like people; for example, people of similar social and demographic backgrounds are more likely to form friendship ties (McPherson et al. 2001). Prior research identifies consistent evidence that homophily defines a variety of social networks, particularly when nodes are linked by strong ties (Kretschmer et al. 2025). In the current study, we apply the traditional concept of homophily to geography. This follows the approach of recent scholarship that identifies homophily among neighborhoods, with interconnected neighborhoods significantly more likely to exhibit similar characteristics than disconnected neighborhoods (Aleinzi et al. 2025; Graif et al. 2017, 2021). The current study extends the concept of geographic homophily to street segments, a micro-level unit of analysis that is privileged within the crime-and-place literature (Linning and Eck 2021; Schnell et al. 2017; Steenbeek and Weisburd 2016; Weisburd et al. 2024).

We are interested in whether gun violence hot spots are more likely to connect to other hot spots than to street segments with lower levels of gun violence. The presence of such hot spot homophily would indicate that hot spots do not operate as self-contained entities but are parts of a larger geographic network of gun violence, a finding that could have significant implications. If gun violence hot spots are connected in space, this could lead to better targeted intervention strategies, with potential preventive spillover effects spreading throughout the gun violence network. The theoretical insights gleaned from a social network perspective may also uncover new social and spatial factors that help develop and sustain gun violence hot spots.

Study Setting

Kansas City is the largest city in Missouri, with a residential population of approximately 508,000 according to the 2020 decennial census. Approximately 28% and 11% of residents identify as Black or Latino, respectively. The poverty rate is 15% with a median household income of approximately $65,000. The regional Midwest Crime Gun Intelligence Center (CGIC) was established in 2014 in response to elevating gun violence rates (Novak and King 2020). The CGIC model involves interagency collaboration to facilitate the immediate collection, management, and analysis of crime gun evidence (Police Executive Research Forum 2017). In Kansas City, the CGIC significantly improved the collection, processing, and analysis of ballistic evidence (Novak and King 2020).

The Midwest CGIC built upon existing crime gun investigative infrastructure, including NIBIN and the eTrace firearms tracing and analysis system (Novak and King 2020). KCPD implemented a range of strategic changes to its CGIC operations between 2017 and 2018. KCPD formalized partnerships with various local and federal law enforcement agencies, adjusted NIBIN processes to facilitate faster turnaround of ballistic evidence, and empowered CGIC detectives to establish and pursue investigative cases (Novak and King 2020). Consequently, the CGIC substantially increased KCPD's capacity to collect and analyze ballistic evidence, making 2014–2019 an appropriate study period for this analysis.

Methodology

Network Data

This study emerged from a long-standing research partnership between the lead author and the KCPD. KCPD originally provided gun violence, police enforcement, and NIBIN data as part of a separate project funded by the National Institute of Justice. All KCPD data includes the incident location at the precise address level. We use data from NIBIN to capture how patterns of gun violence connect geographic space. We construct a geographic network of gun violence by considering the repeated use of the same crime guns across different incidents. Nodes, or the individual entities in the network, represent street segments. Edges, or relational connections, link street segments if a NIBIN hit associated with the same crime gun was obtained in each location. Nearly 78% of the NIBIN hits were generated from ballistic evidence (i.e., cartridge casings and projectile fragments) collected from crime scenes, while about 22% were generated via test fires of guns seized by law enforcement. Any guns seized by law enforcement operate as crime guns since NIBIN does not generate a hit unless a gun is associated with ballistic evidence used in another criminal event (King et al. 2013).

Due to Kansas City's high rate of processing cartridges and projectiles, collected ballistic evidence accounts for more NIBIN leads in Kansas City than the national average, where about 60% of NIBIN leads are generated from test fires of seized guns (Joyce Foundation 2024, p. 57). We believe the enhanced processing of cartridges and projectiles in Kansas City leads to a higher proportion of the city's crime guns (as reflected in NIBIN) being represented in our network than would be expected in most U.S. cities.²

We consider street segments—the two block faces on both sides of a street between two intersections—as the nodes of our network because criminology of place scholarship documents many benefits to analyzing spatial phenomena at this unit of geography. Street segments are simultaneously small enough to avoid aggregation errors such as the ecological fallacy and large enough to avoid coding errors associated with small units such as street addresses (Braga et al. 2010; Weisburd et al. 2012). This unit of analysis is ideal for measuring hot spots as it accurately models real-world geography and captures crime-generating processes that unfold within micro-level behavior settings (Groff et al. 2023; Linning and Eck 2021). Micro-level units like street segments reflect what Taylor (2015) (also see Taylor et al. 1984) articulates as “behavior settings;” well-observed, well-documented units of social activity that can typically be seen in their entirety at the street level. Street segments further explain higher levels of variance as compared to meso- and macro-level geographies (Schnell et al. 2017; Steenbeek and Weisburd 2016), and reflect the preference for smaller geographic units of analysis in social network analysis (Faust and Tita 2019).

The edges of our gun violence network connect pairs of street segments when ballistic evidence from the same crime gun appeared at least once in both geographic areas between 2014 and 2019.³ All edges are undirected, meaning that street segments are linked by symmetric edges regardless of the temporal ordering of police reports about recovered ballistic evidence.⁴ We interpret the links between street segments as proxies for the social processes that facilitate the movement of crime guns throughout space. For example, edges may indicate the movement of a single offender across geographic space or cooperation among co-offenders who share crime guns through underground markets (Ciomek et al. 2020; Hureau and Braga 2018; Roberto et al. 2018).

Although Kansas City comprises 33,848 street segments, we limit our analytic sample to include only those where crime gun evidence was identified. Including street segments that did not appear in the NIBIN database would potentially bias the estimates of our analytic models since it was not possible for these areas to be linked to other geographies via connections of common ballistics evidence. As a result, including these street segments may artificially inflate the coefficients of terms that are also associated with ballistics evidence being present. To ensure that our estimates of hot spot homophily are not upwardly biased, our network includes 1,701 street segments that are connected by 2,248 edges indicating the co-occurrence of evidence from 1,349 crime guns reported via NIBIN hits (i.e., the ballistic data submitted to IBIS that is connected to at least one other crime gun).⁵ The gun violence network incorporates data from 300 NIBIN leads that were associated with only one crime gun and, as a result, includes 189 street segments with no connections (i.e., isolates).

Street Segment Characteristics

The analysis is primarily concerned with how gun violence concentration impacts patterns of clustering in the geographic network of gun violence. Street segments are classified as hot spot, low gun crime, or no gun crime areas. The analysis uses 5-years of gun violence data (2014–2019), covering the period since the establishment of the Midwest CGIC. Gun violence incidents include shootings (both fatal and non-fatal), as well as gun-related aggravated assaults and robberies.⁶^,⁷

Counts of shootings, gun assaults, and gun robberies are summed within each street segment. To prevent the more frequently occurring crime types (i.e., gun assaults and gun robberies) from exerting more influence on the street segment classification, standardized values are calculated for each category using the robust standardization tool in ArcGIS Pro.⁸ The three standardized values are then summed to calculate a gun violence index for all street segments in Kansas City. We first classify the street segments where no gun violence occurred as no gun crime units. For the remaining street segments, we consider their gun violence index relative to the observed mean and standard deviation across all 33,848 street segments in Kansas City. Street segments with gun violence index values greater than 1 standard deviation above the mean (2.86 + 4.72 = 7.58) are classified as hot spots. The street segments with gun violence index values between 0 and 7.58 are classified as low gun crime units (see Figure 1). Across all of the street segments in Kansas City, 1.48% (n = 500) were categorized as hot spots, which mirrors the proportion observed in many prior crime-and-place studies (Connealy and Hart 2024; Haberman 2017).⁹^,¹⁰

Figure 1.

Gun violence at street segments in Kansas city, Missouri: 2014–2019.

Our analyses also include a range of control variables, organized according to underlying theoretical constructs relating to gun violence. Prior research has found each of the control variables to be positively correlated with gun violence hot spots to various extents. Their inclusion in the models provides a robust test of hot spot homophily by ensuring that any connectivity of hot spots is not simply a byproduct of the ecological characteristics of high-crime places. All data for the control variables are provided by the Kansas City Police Department or collected from third-party sources (the U.S. Census Bureau's American Community Survey or the Data Axle business license database).

Three binary variables measure aspects of the built environment, reflecting the crime-generating or attracting processes posited in environmental criminology (Brantingham and Brantingham 1993, 2008). A binary variable measured whether one or more consumer-facing businesses (retail stores, restaurants, food/drug stores, and general services) are located on the street segment in recognition of the increased foot traffic and potential crime-generation caused by such businesses (Kim and Hipp 2021)¹¹. Two additional variables measure whether street segments were within a quarter-mile of a sports stadium (Bagwell 2026; Campedelli et al. 2024; Newton 2018) or within 172.5 feet (half the median block size) of a public park (Boessen and Hipp 2018; Groff and McCord 2012).¹²

Three variables measure factors associated with social disorganization. In ArcGIS Pro, each street segment was assigned the value of the encompassing census tract.¹³ Concentrated disadvantage is measured through summed standardized percentages of households receiving public assistance, households below the poverty line, persons unemployed, households with a single female head and child under the age of 18, persons without a high-school diploma or equivalent, male residents aged 15–29, vacant properties, and renter-occupied properties (Sampson et al. 2002). Racial minority population is measured through the standardized percentage of non-white residents (Morenoff et al. 2001). A binary variable measures whether the count of 311 calls for physical disorder was greater than one standard deviation above the mean for all street segments (O’Brien and Sampson 2015).¹⁴

Four binary variables measure police enforcement or surveillance associated with the targeting or responding to high-crime places. These measures indicate whether the street segment falls within the ShotSpotter gunshot detection coverage area (Piza et al. 2024),¹⁵ contains one or more video surveillance cameras (Piza et al. 2019), received a count of tickets and citations that is greater than one standard deviation above the mean, and experienced a count of arrests that is greater than one standard deviation above the mean (Nix et al. 2024; Wu and Lum 2020).¹⁶

Because ballistic evidence from the same crime gun may tend to be recovered from geographically proximate street segments, we control for the distances between each pair of street segments in miles and whether two street segments were embedded in the same census tract. We also construct a measure of each street segment's distance in miles to the nearest hot spot, fixing this value to zero miles for all street segments characterized as hot spots themselves. This control ensures that our findings are not biased by any tendency for ballistic evidence to be recovered more frequently in areas that are physically near hot spots. Additionally, we control for the length of each street segment (measured in miles) since larger geographic areas may provide more opportunities for ballistic evidence to be recovered.

In line with previous research suggesting that hot spot analyses should account for the population at risk of victimization (Telep and Hibdon 2017), we include a measure of ambient population—an estimate of the average daily on-street population in the surrounding square-kilometer area. This measure is derived from the Oak Ridge Laboratory's LandScan database (https://landscan.ornl.gov/), which measures ambient population by triangulating sources, including census data, transportation networks, landmarks, land use, and nighttime lights. Finally, a binary variable measures whether the street segment experienced a count of citizen calls for shots fired that was greater than one standard deviation above the mean.

Analytical Approach

To understand connectivity among hot spots, we estimate exponential random graph models (ERGMs). ERGMs are a multivariate statistical method that compares the dyadic, or pairwise, patterns in an observed network to a set of randomly simulated networks (Cranmer et al. 2020; Robins et al. 2007). By treating the empirical network as the dependent variable, ERGMs can determine whether the patterns of edges observed in the network are statistically significantly different from what would be expected by random chance. Importantly, the multivariate nature of the ERGM allows us to compare how our observed network varies from otherwise similar, randomly generated networks that control for all individual, dyadic, and structural parameters included in the model. Following Faust and Tita's (2019) recommendations, all ERGMs presented here are fit using Markov Chain Monte Carlo MLE. We present technical details about the ERGM, goodness of fit statistics, and assessments of multicollinearity in the Supplemental Materials (Part A).

ERGM Parameters

We include various parameters in our ERGMs to assess whether node-level, dyad-level, and structural processes inform the relational patterns of the gun violence network. Additional details about all parameters are included in the Supplemental Materials (Part B). To determine whether street segments are more likely to be connected if they are assigned to the same category of gun violence concentration (hot spot, low gun crime, or no gun crime), we begin by including homophily parameters. Homophily parameters evaluate whether street segments defined by the same gun violence concentration are more likely to be connected in the observed gun violence network than would be expected by random chance. We begin by including a gun violence concentration homophily parameter that assumes this clustering is consistent across the three categories of gun violence concentration (i.e., uniform homophily). In a sense, this parameter averages the level of clustering among hot spots, low gun crime, and no gun crime street segments and weights its estimate according to the distribution of these gun violence concentration categories. A positive and significant coefficient would indicate that when street segments are defined by the same gun violence concentration, they are more likely to be connected in the network. Given our interest in evaluating whether hot spot street segments are more likely to be connected to other hot spots, we also estimate an ERGM with separate parameters for each category of gun violence concentration: hot spot homophily, low gun crime homophily, and no gun crime homophily (i.e., differential homophily). In particular, the hot spot homophily parameter tests whether pairs of hot spots are linked together in the gun violence network at higher rates than expected.

We further evaluate how gun violence concentration shapes patterns of connectivity by estimating an ERGM that includes a set of mixing terms. The mixing terms consider whether links of gun violence are more likely to connect certain pairs of street segments than would be expected by chance. By setting pairs of street segments containing two hot spots as our reference category, we can evaluate whether ties are more likely to connect hot spots to other hot spots than all other pairings of street segments (e.g., hot spots and low gun crime areas, two no gun crime areas). A negative and significant coefficient of an included mixing term would indicate that ties between the focal pairing of street segments are less likely to occur in the observed network than ties between pairs of hot spots. This final modeling strategy allows us to ensure that any observed homophily is not an artifact of varying levels of connectivity across street segments with different concentrations of gun violence.

We incorporate additional ERGM terms to account for how our control variables shape the structural patterns of the gun violence network. First, we include connectivity parameters that consider whether street segments defined by a characteristic of interest tend to be more embedded in the gun violence network. Positive and significant coefficients indicate that higher values of a segment-level characteristic are associated with being more connected to other street segments in the gun violence network. We include several connectivity parameters to evaluate whether street segments accumulate more linkages if they have: (1) consumer-facing businesses present, (2) parks present, (3) stadiums present, (4) higher levels of concentrated disadvantage, (5) higher levels of non-white residents, (6) ShotSpotter gunshot detector coverage, (7) CCTV coverage, (8) higher rates of tickets/citations, (9) higher arrest rates, (10) higher levels of shots fired calls, (11) higher rates of 311 calls for physical disorder, (12) closer geographic distance to hot spots, (13) larger geographic sizes, and (14) higher ambient populations.

Second, we include an edgewise parameter to evaluate whether edges are more likely to link street segments that are physically proximate. A negative value of the coefficient would suggest that street segments are less likely to be connected when they are more physically distant. Third, we include a homophily measure to test whether street segments are more likely to be linked if they fall within the same census tract, with a positive coefficient providing evidence for this phenomenon. Finally, an absolute difference term is incorporated to account for any tendencies for street segments with similar racial/ethnic distributions to be connected. Negative coefficients indicate that the odds of two street segments being linked are higher as the absolute difference between their proportion of racial/ethnic minority residents increases.

All ERGMs include three structural parameters to control for well-established, endogenous processes that can impact network topology (following Kreager et al. 2021; McMillan et al. 2022; Papachristos et al. 2013). We account for patterns of triadic closure, or the tendency for nodes with mutual connections to be linked to one another, by including the geometrically weighted edgewise shared partner (GWESP) parameter. The geometrically weighted degree (GWD) parameter controls for the observed network's degree distribution, including the number of isolated street segments. Finally, the edges parameter accounts for the base log-odds that two street segments will be connected in the gun violence network.

Results

Descriptives

As expected, hot spot and low gun crime street segments are overrepresented in the geographic network of gun violence (see Table 1).¹⁷ In the analytical sample of 1,701 street segments, there are 298 (17.52%) hot spot street segments, 1,053 (61.90%) low gun crime street segments, and 350 (20.58%) no gun crime street segments in the gun violence network. No gun crime street segments can be associated with NIBIN because ballistic evidence can be collected through events like gun arrests and shots fired, where no one is struck by gunfire. In such cases, ballistic evidence is collected, but no gun crimes are recorded. On average, each street segment is connected to 2.46 other street segments in Kansas City via ballistics from the same crime gun. Patterns of connectivity vary, however, according to street segments’ levels of gun violence concentration. For instance, hot spots are connected to an average of 3.96 street segments, while low gun crime and no gun crime segments are less embedded in the network of gun violence. Slightly fewer than half of the edges in the network link street segments defined by the same level of gun violence concentration. Of particular interest, 25.11% of the hot spot street segments’ connections are with other hot spots, even though hot spots only represent 17.52% of the street segments in the sample.

Table 1.

Descriptive Statistics for Street Segments in the Gun Violence Network.

	Total		Hot Spot		Low Gun Crime		No Gun Crime
	n = 1701		n = 298		n = 1053		n = 350
Number of connections	2.643	(3.357)	3.956	(4.786)	2.525	(3.087)	1.880	(3.367)
% of connections between segments of the same gun violence concentration	44.13%		25.11%		59.42%		16.41%
Distance to the nearest hot spot	0.293	(0.829)	0	(0)	0.236	(0.583)	0.712	(1.435)
Length of street segment	0.112	(0.082)	0.127	(0.073)	0.112	(0.083)	0.098	(0.082)
High rate of shots fired calls	38.86%		75.17%		38.27%		9.71%
Ambient population	1.535	(1.920)	1.404	(1.069)	1.516	(1.903)	1.705	(2.454)
Businesses present	19.81%		33.56%		16.43%		18.29%
Stadiums present	0.82%		0.67%		1.04%		0.29%
Public parks present	14.64%		14.77%		15.48%		12.00%
Concentrated disadvantage	3.560	(3.398)	4.583	(2.944)	3.781	(3.221)	2.025	(3.765)
High rate of 311 disorder calls	31.57%		42.62%		33.33%		16.86%
ShotSpotter coverage	15.40%		21.81%		15.29%		10.29%
CCTV coverage	6.70%		15.10%		4.46%		6.63%
High rate of tickets/citations	7.94%		15.10%		6.93%		4.86%
High rate of arrests	12.76%		33.89%		8.74%		6.86%
Prop. of racial minority residents	0.653	(0.265)	0.714	(0.221)	0.682	(0.250)	0.515	(0.295)
Distance between street segment pairs	5.867	(4.274)

We visualize patterns of connectivity in the gun violence network in Figure 2. For visualization, we focus on those street segments in the largest connected component (LCC), or the largest cluster of street segments that are all reachable by direct or indirect connections in the network. The LCC includes 911 (or 53.56%) of the total street segments in the spatial network of gun violence. The remaining 790 street segments are embedded in pairs or small groups of connected street segments (n = 601) or are not linked to any other street segments via ballistic evidence (n = 189). From Figure 2, it is apparent that hot spots are highly prevalent in the central core of the violence network, while low gun crime and no gun crime segments are more likely to be situated on the peripheries. Furthermore, hot spots tend to cluster together in small pockets throughout the network's center, suggesting that micro-geographic units defined by high concentrations of gun violence are interlinked with one another.

Figure 2.

Largest connected component of the gun violence network.

Figure 3 displays all street segments in the gun violence network (i.e., street segments where NIBIN evidence was collected) symbolized by their degree centrality (i.e., number of connections to other street segments). These street segments in the gun violence network are the units included in the ERGM models discussed subsequently. Similar to gun violence hot spots (see Figure 1), high-degree street segments generally cluster within the central and southeastern portions of Kansas City. However, visual inspection of maps suggests the presence of multiple hot spot street segments with low measures of degree centrality. In other words, overall gun violence incident counts do not seem to fully explain a street segment's connectivity.

Figure 3.

Street segments in the gun violence network by degree centrality: 2014–2019.

Multivariate ERGM Results

We turn to multivariate ERGM analyses to evaluate whether hot spot street segments continue to cluster together in the gun violence network after accounting for other factors that shape patterns of connectivity (see Table 2). Model 1 begins by introducing controls related to the geographic structure of Kansas City, the tendency for street segments to receive shots fired calls, the ambient population, and structural network processes. When we consider all street segments, regardless of their gun violence category, we find no evidence for homophily by gun violence concentration (b = −.061, SE = .039, p = .112). These results suggest that there is no uniform tendency for hot spot, low gun crime, and no gun crime street segments to be linked to others that share their same category of gun violence concentration.

Table 2.

ERGMs Estimating Patterns of Connectivity in the Gun Violence Network.

	Model 1			Model 2			Model 3			Model 4
	b	SE		b	SE		b	SE		b	SE
Independent variables of interest:
Gun violence concentration homophily	−.061	(.039)
Hot spot homophily				.384	(.078)	***	.361	(.081)	***
Low gun crime homophily				−.082	(.040)	*	−.082	(.041)	*
No gun crime homophily				−.359	(.149)	*	−.372	(.140)	**
Street segment mixing (ref: both hot spots)
Both no gun crime										−.753	(.163)	***
No gun crime and low gun crime										−.667	(.098)	***
Both low gun crime										−.449	(.081)	***
Hot spot and no gun crime										−.300	(.109)	**
Hot spot and low gun crime										−.177	(.088)	*
Street segment-level connectivity controls:
Distance to the nearest hot spot	.037	(.022)		.049	(.021)	*	.057	(.022)	**	.076	(.022)	***
Length of street segment	.446	(.147)	**	.404	(.159)	**	.198	(.163)		.147	(.165)
High rate of shots fired calls	.198	(.026)	***	.159	(.027)	***	.130	(.028)	***	.078	(.028)	**
Average ambient population	−.010	(.007)		−.009	(.008)		−.008	(.008)		−.004	(.008)
Businesses present							−.025	(.033)		−.045	(.034)
Stadiums present							.013	(.163)		−.024	(.158)
Public parks present							.035	(.035)		.033	(.036)
Concentrated disadvantage							−.011	(.004)	**	−.014	(.004)	***
High rate of 311 disorder calls							.074	(.028)	**	.066	(.028)	*
ShotSpotter coverage							−.050	(.035)		−.062	(.036)
CCTV coverage							.006	(.048)		−.008	(.051)
High rate of tickets/citations							.074	(.042)		.066	(.043)
High rate of arrests							.089	(.038)	*	.062	(.039)
Dyad-level controls:
Edgewise geographic distance	−.043	(.006)	***	−.043	(.006)	***	−.044	(.006)	***	−.045	(.006)	***
Census tract homophily	.822	(.111)	***	.826	.115	***	.701	.116	***	.695	(.117)	***
Diff. in prop. of racial minority residents							−.511	(.099)	***	−.491	(.099)	***
Network structure controls:
GWESP (α = .01)	3.059	(.047)	***	3.063	(.045)	***	3.061	(.046)	***	3.065	(.046)	***
GWD (α = .1)	.840	(.085)	***	.859	(.081)	***	.872	(.085)	***	.942	(.086)	***
Edges	−8.105	(.079)	***	−8.084	(.080)	***	−7.855	(.100)	***	−7.419	(.129)	***
AIC	27,258			27,229			27,196			27,136
BIC	27,380			27,375			27,464			27,428

Notes: *p < .05, **p < .01, ***p < .001.

However, Model 2 demonstrates that this finding is due to varying patterns of connectivity across street segments defined by different categories of gun violence concentration. We find that hot spots are more likely to be linked to other hot spots in the gun violence network than would be expected to occur by random chance (b = .384, SE = .078, p < .001). If two street segments are hot spots, this increases the odds that ballistic evidence from the same crime gun appeared in each unit by 46.8%, after accounting for all controls in the model. Interestingly, this tendency toward homophily by gun violence concentration does not define the connectivity patterns of low gun crime or no gun crime street segments. Low gun crime street segments are linked to other low gun crime street segments at significantly lower rates than if edges were randomly assigned in the network of street segments (b = −.082, SE = .040, p < .05). Similarly, there are significantly fewer links between pairs of no gun crime street segments than would be expected by chance (b = −.359, SE = .149, p < .05).

Model 3 introduces additional controls to account for how features related to the built environment, social disorganization, and police enforcement impact patterns of connectivity. The coefficient for hot spot homophily remains positive and significant (b = .361, SE = .081, p < .001). Even after accounting for various features that are associated with patterns of gun violence, pairs of hot spots have 43.5% greater odds of being connected than would be expected to occur by chance. Alternatively, low gun crime and no gun crime street segments remain significantly less likely to be linked to street segments of their same category of gun violence (Low Gun Crime: b = −.082, SE = .041, p < .05; No Gun Crime: b = −.372, SE = .1420, p < .01).

To ensure that our homophily findings are not driven by any tendencies for hot spots to be more connected in the gun violence network, we estimate a fourth ERGM that includes a set of mixing terms. Results indicate that crime gun linkages are significantly more likely to connect pairs of hot spots than all other possible pairings of street segments. For instance, a pair of two no gun crime street segments are 52.9% less likely to be connected in the gun violence network than a pair of two hot spot street segments (b = −.753, SE = .163, p < .001). Furthermore, edges connecting hot spots to low gun crime areas (b = −.177, SE = .088, p < .05) and hot spots to no gun crime areas (b = −.300, SE = .109, p < .01) are significantly less likely to occur in the observed network than edges linking pairs of hot spots.

Control variables also highlight intriguing patterns in the gun violence network. As each street segment's physical distance to the nearest hot spots increases, it is more likely to be linked via shared ballistic evidence to all other street segments in the gun violence network (b = .076, SE = .022, p < .001). Experiencing a high number of 311 disorder calls and shots fired calls further increases a street segment's odds of being connected to other areas (311 disorder: b = .066, SE = .028, p < .05; Shots fired: b = .078, SE = .028, p < .01). Every one unit increase in a street segment's level of concentrated disadvantage decreases its odds of being connected to another street segment in the gun violence network by 1.4% (b = −.014, SE = .004, p < .001). We suspect that these patterns reflect how resource deprivation limits the capacity for people engaged in gun violence to move throughout the city.

At the dyadic level, we find that street segments are more likely to be linked by ballistic evidence from the same crime gun if they are in close physical proximity (b = −.045, SE = .006, p < .001) or located within the same census tract (b = .695, SE = .117, p < .001). Furthermore, street segments are more likely to be connected if their residents are characterized by similar racial/ethnic backgrounds (b = −.491, SE = .099, p < .001). For example, if all residents in a street segment identify as white, the odds of being connected to an area where all residents identify as racial minorities are 38.8% lower.

Structural network controls provide evidence that the gun violence network is also defined by processes that are endogenous to the system itself, which supports our decision to adopt a network approach. The positive and significant coefficient for the GWESP parameter indicates that there are tendencies toward triadic closure in the gun violence network (b = 3.065, SE = .046, p < .001). In other words, if ballistic evidence from the same crime gun appeared in street segments A and B and street segments B and C, then street segments A and C are more likely to be connected than would be expected by random chance. The positive and significant GWD coefficient indicates that the degree distribution of the gun violence network is relatively uniform (b = .942, SE = .086, p < .001). Finally, the edges coefficient is negative and significant, reflecting the general sparsity of connections (b = -7.419, SE = .129, p < .001).

Discussion and Conclusion

Prior scholarship demonstrates that crime is highly concentrated in micro-places (Weisburd 2015; Weisburd et al. 2024). Findings of the current study indicate that, rather than being self-contained, the high concentration of gun violence inherent to hot spots appears to interconnect geographic units, linking certain hot spots together through an intricate network of gun violence. Gun violence hot spots in Kansas City are more likely to be connected to each other through the social ties implied from crime gun evidence than to low gun crime and no gun crime street segments. Clustering among hot spot street segments—or what we define as hot spot homophily—continues to be present after controlling for a range of factors related to geographic proximity, the built environment, social disorganization, ambient population, and police enforcement, as well as various network processes of connectivity and triadic closure. The social connections implied by the movement of crime guns transcend geographic boundaries in ways that compound exposure to violence within and between hot spots.

Our finding that hot spots are more likely to be connected to other hot spots in spatial networks of gun violence provides important nuance to the study of crime and place and reflects the core principle in the spatial network literature that nodes can be highly connected while geographically distant (Brazil et al. 2025). We consider this to be an important finding, given that the spatial social network literature has predominately measured homophily at the neighborhood level (Aleinzi et al. 2025; Brazil et al. 2025; Johnson and Roman 2025). The observation of homophily at the street-segment level demonstrates that network analysis principles operate within the spatial unit of analysis most aligned with hot spots research. This observation both advances our theoretical understanding of hot spots and offers practical solutions that could be applied to prevent gun violence in such areas.

The current study indicates that being connected to extant hot spots through the co-occurrence of crime guns may be a risk factor that can enhance our understanding of how and why violent crime concentrates in certain geographic spaces. Similar to what has been observed with co-offending networks (Papachristos et al. 2013, 2024; Schaefer 2012; Schaefer et al. 2014), hot spot homophily may interact with features of the surrounding environment in a manner that provides increased opportunities for gun violence.

Related to these theoretical insights, we also believe this study has implications for place-based crime prevention. Our findings suggest that considering the spatial networks of gun violence may offer opportunities to increase the efficiency of place-based policing interventions (Braga et al. 2019; Braga and Weisburd 2022). For example, given the unique positionality of high gun crime street segments, law enforcement agencies could prioritize intervening in hot spots with large numbers of NIBIN connections, particularly if these links connect them to other hot spots. Successfully addressing crime in such places may create residual prevention effects by reducing the risk of gun crime diffusing to connected places. Given that over 50% of the street segments in our study were connected in a large gun violence network through NIBIN incidents, such an approach may potentially impact a large proportion of high-risk locations within a city. This mirrors arguments for identifying and responding to crime events with a high likelihood of contagion in order to prevent the formation of near-repeat patterns (Santos and Santos 2015).

Despite the implications, this study, like all research, suffers from some limitations. We used NIBIN data to operationalize ties between street segments given our focus on gun violence hot spots. Alternative data points, such as arrestees, could have been used to create connections between hot spot units. However, prior research has acknowledged that enforcement data are often incomplete—given police do not identify and apprehend all offenders—which can present challenges to studying social network related research questions (Bright et al. 2021; Faust and Tita 2019). Enforcement data may further reflect artificial ties between actors, such as when large numbers of people are “linked” through mass arrests during targeted police operations (e.g., shuttering of encampments) or large public gatherings (e.g., protests) (Faust and Tita 2019).

Although our analysis of NIBIN data overcomes many limitations of using arrest records, NIBIN records have their own limitations. For instance, the NIBIN database does not include data on ballistic evidence and crime guns that were not recovered by law enforcement. As a result, our gun violence network may be missing connections between certain geographic areas. Since it is not possible to determine instances where this missing data occurred, we cannot apply multiple imputation methods for missing social network data (C. Wang et al. 2016; Krause et al. 2020). However, we suspect that these patterns of missingness are minimal and, thus, unlikely to bias our findings. Additionally, while NIBIN hits reflect the movement of crime guns throughout space, we are unable to determine the precise social processes responsible for this phenomenon. While we suggest some possibilities (e.g., repeat gun offenders traveling through space or crime guns being shared within offender networks), we lack the data necessary to isolate an exact cause. Moving forward, researchers should closely partner with police agencies to collect the data necessary to derive more meaning from the NIBIN hits that link street segments, such as detective case files. Yet regardless of whether the ties we identify represent the movement of individual offenders or interactions among those involved in illegal gun markets, these connections represent understudied social dynamics that link places with high concentrations of gun violence.

The control variables used to measure the built environment speak to the limited range of environmental criminology constructs available in our data. While we are able to control for many built environment features mentioned in criminology of place studies, factors such as offender anchor points represent other key environmental criminology constructs (Bernasco and Block 2011). Unfortunately, such data were not accessible to us. Future research should aim to build upon the methods advanced in the current study.

While acknowledging these limitations, the current study makes an important contribution to the crime-and-place literature. Results indicate that NIBIN evidence links geographic hot spots, forming a complex network of gun violence. Beyond sharing structural traits, as noted in prior research, gun violence hot spots may also be linked through spatial interdependencies. As such, these findings suggest that analyzing the spatial networks of gun violence could enhance the effectiveness of targeted policing strategies by promoting positive spillover effects, mitigating the societal costs of violence, and enhancing community well-being.

Supplemental Material

sj-docx-1-jrc-10.1177_00224278261451742 - Supplemental material for Geographic Networks of Gun Violence: Exploring Hot Spot Connectivity Through Ballistic Evidence and Social Network Analysis

Supplemental material, sj-docx-1-jrc-10.1177_00224278261451742 for Geographic Networks of Gun Violence: Exploring Hot Spot Connectivity Through Ballistic Evidence and Social Network Analysis by Eric L. Piza, Cassie McMillan and Daniel Trovato in Journal of Research in Crime and Delinquency

Footnotes

Acknowledgements

Crime and police data for this study were originally collected as part of a project funded by the National Institute of Justice, grant number 2019-R2-CX-0004. We thank Captain Jonas Baughman of the Kansas City Police Department for ensuring access to this data and for facilitating discussions between the research team and agency leadership.

Funding

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

Supplemental Material

Supplemental material for this article is available online.

ORCID iDs

Eric L. Piza

Cassie McMillan

Daniel Trovato

Notes

Author biographies

Eric L. Piza is the Lipman Family Professor of Criminology & Criminal Justice and Director of Crime Analysis Initiatives at Northeastern University. His research centers on the spatial analysis of crime patterns, evidence-based policing, crime control technology, and the integration of academic research and police practice.

Cassie McMillan is an Assistant Professor of sociology and criminology at Northeastern University. Her research focuses on social networks, youth crime, substance use, and social inequality. In particular, she is interested in developing and applying network methodologies to better understand disparities in crime and delinquency.

Daniel Trovato is an Assistant Professor in Criminology and Criminal Justice at Florida International University. His research focuses on life course criminology, the sociology of punishment, and violence prevention.

References

Aleinzi

Sharkey

Pryce

Hughes

. 2025. “Spatial Networks of Neighborhood Violence.” Journal of Quantitative Criminology. 10.1007/s10940-025-09637-3

Andresen

M. A.

Linning

S. J.

. 2012. “The (In)appropriateness of Aggregating Across Crime Types.” Applied Geography 35(1–2):275–82. 10.1016/j.apgeog.2012.07.007

Bagwell

2026. “The Geography of Game Day: Stadiums, Football, and the Spatial Clustering of Crime.” Crime & Delinquency. 10.1177/00111287251411789

Bernasco

Block

. 2011. “Robberies in Chicago: A Block-Level Analysis of the Influence of Crime Generators, Crime Attractors, and Offender Anchor Points.” Journal of Research in Crime and Delinquency 48(1):33–57. 10.1177/0022427810384135

Birks

Davies

. 2017. “Street Network Structure and Crime Risk: An Agent-Based Investigation of the Encounter and Enclosure Hypotheses.” Criminology; An Interdisciplinary Journal 55(4):900–37. 10.1111/1745-9125.12163

Blumstein

Cork

. 1996. “Linking Gun Availability to Youth Gun Violence.” Law and Contemporary Problems 59(1):5. 10.2307/1192207

Boessen

Hipp

J. R.

. 2018. “Parks as Crime Inhibitors or Generators: Examining Parks and the Role of Their Nearby Context.” Social Science Research 76:186–201. 10.1016/j.ssresearch.2018.08.008

Boessen

Hipp

J. R.

. 2022. “The Network of Neighborhoods and Geographic Space: Implications for Joblessness While on Parole.” Journal of Quantitative Criminology 38(3):597–636. 10.1007/s10940-021-09510-z

Boessen

Hipp

J. R.

Smith

E. J.

Butts

C. T.

Nagle

N. N.

Almquist

. 2014. “Networks, Space, and Residents’ Perception of Cohesion.” American Journal of Community Psychology 53(3-4):447–61. 10.1007/s10464-014-9639-1

10.

Braga

A. A.

Papachristos

A. V.

Hureau

D. M.

. 2010. “The Concentration and Stability of Gun Violence at Micro Places in Boston, 1980–2008.” Journal of Quantitative Criminology 26(1):33–53. 10.1007/s10940-009-9082-x

11.

Braga

A. A.

Pierce

G. L.

. 2011. “Reconsidering the Ballistic Imaging of Crime Bullets in Gun Law Enforcement Operations.” Forensic Science Policy & Management: An International Journal 2(3):105–17. 10.1080/19409044.2011.613444

12.

Braga

A. A.

Turchan

B. S.

Papachristos

A. V.

Hureau

D. M.

. 2019. “Hot Spots Policing and Crime Reduction: An Update of an Ongoing Systematic Review and Meta-Analysis.” Journal of Experimental Criminology 15(3):289–311. 10.1007/s11292-019-09372-3

13.

Braga

A. A.

Weisburd

D. L.

. 2022. “Does Hot Spots Policing Have Meaningful Impacts on Crime? Findings from an Alternative Approach to Estimating Effect Sizes from Place-Based Program Evaluations.” Journal of Quantitative Criminology 38(1):1–22. 10.1007/s10940-020-09481-7

14.

Brantingham

. 2008 “Crime Pattern Theory.” Pp. 78–93 in Environmental Criminology and Crime Analysis1st ed. New York, NY: Willan Publishing.

15.

Brantingham

P. L.

Brantingham

P. J.

. 1993. “Nodes, Paths and Edges: Considerations on the Complexity of Crime and the Physical Environment.” Journal of Environmental Psychology 13(1):3–28. 10.1016/S0272-4944(05)80212-9

16.

Brazil

Candipan

Levy

Tom

. 2025. “Beyond the Residential Neighborhood: A Scoping Review of Research on Urban Neighborhood Networks.” Social Science & Medicine 372:117945. 10.1016/j.socscimed.2025.117945

17.

Bright

Brewer

Morselli

. 2021. “Using Social Network Analysis to Study Crime: Navigating the Challenges of Criminal Justice Records.” Social Networks 66:50–64. 10.1016/j.socnet.2021.01.006

18.

Calderoni

2012. “The Structure of Drug Trafficking Mafias: The ‘Ndrangheta and Cocaine.” Crime, Law and Social Change 58(3):321–49. 10.1007/s10611-012-9387-9

19.

Campedelli

G. M.

Piza

E. L.

Piquero

A. R.

Kurland

. 2024. “Evidence on the Impact of the Prudential Center on Crime in Downtown Newark.” Journal of Experimental Criminology 20(4):1225–51. 10.1007/s11292-023-09576-8

20.

Chalfin

Kaplan

Cuellar

. 2021. “Measuring Marginal Crime Concentration: A New Solution to an old Problem.” Journal of Research in Crime and Delinquency 58(4):467–504. 10.1177/0022427820984213

21.

Ciomek

A. M.

Braga

A. A.

Papachristos

A. V.

. 2020. “The Influence of Firearms Trafficking on Gunshot Injuries in a Co-Offending Network.” Social Science & Medicine 259:113114. 10.1016/j.socscimed.2020.113114

22.

Clarke

R. V.

1995. “Situational Crime Prevention.” Crime and Justice: A Review of Research 19:91–150. 10.1086/449230

23.

Clarke

R. V.

2012. “Opportunity Makes the Thief. Really? And so What?” Crime Science 1(1):3. 10.1186/2193-7680-1-3

24.

Cohen

L. E.

Felson

. 1979. “Social Change and Crime Rate Trends: A Routine Activity Approach.” American Sociological Review 44(4):588. 10.2307/2094589

25.

Connealy

N. T.

2023. “The Influence, Saliency, and Consistency of Environmental Crime Predictors: A Probability Score Matching Approach to Test What Makes a Hot Spot Hot.” Justice Quarterly 40(5):670–93. 10.1080/07418825.2022.2106292

26.

Connealy

N. T.

Hart

T. C.

. 2024. “‘Top 10’ Policing as an Alternative Place-Based Strategy: Responding to the Overcomplication and Underestimation of the Law of Crime Concentration.” Police Practice and Research 25(1):71–89. 10.1080/15614263.2023.2183202

27.

Connealy

N. T.

Piza

E. L.

. 2019. “Risk Factor and High-Risk Place Variations Across Different Robbery Targets in Denver, Colorado.” Journal of Criminal Justice 60:47–56. 10.1016/j.jcrimjus.2018.11.003

28.

Cranmer

S. J.

Desmarais

B. A.

Morgan

J. W.

. 2020. Inferential Network Analysis. Cambridge: Cambridge University Press.

29.

Davies

2023 “Street Networks.” Pp. 75-81 in Understanding Crime and Place: A Methods Handbook edited by Groff

Haberman

C. P.

. Philadelphia, PA: Temple University Press.

30.

Davies

Johnson

S. D.

. 2015. “Examining the Relationship Between Road Structure and Burglary Risk via Quantitative Network Analysis.” Journal of Quantitative Criminology 31(3):481–507. 10.1007/s10940-014-9235-4

31.

De Biasi

Rojek

McGarrell

. 2026. “From Cartridge Cases to Spatial Patterns: Leveraging NIBIN to Identify Near-Repeat Shootings in Detroit.” Journal of Criminal Justice 102:102588. 10.1016/j.jcrimjus.2025.102588

32.

Di Méo

2023. “Historical Co-Offending Networks: A Social Network Analysis Approach.” The British Journal of Criminology 63(6):1591–611. 10.1093/bjc/azad005

33.

Faust

Tita

G. E.

. 2019. “Social Networks and Crime: Pitfalls and Promises for Advancing the Field.” Annual Review of Criminology 2(1):99–122. 10.1146/annurev-criminol-011518-024701

34.

Forati

Ghose

Mohebbi

Mantsch

J. R.

. 2023. “The Journey to Overdose: Using Spatial Social Network Analysis as a Novel Framework to Study Geographic Discordance in Overdose Deaths.” Drug and Alcohol Dependence 245:109827. 10.1016/j.drugalcdep.2023.109827

35.

Fox

A. M.

Novak

K. J.

Van Camp

James

. 2021. “Predicting Violent Victimization Using Social Network Analysis from Police Data.” Violence and Victims 36(3):436–54. 10.1891/VV-D-19-00037

36.

Gill

L. M.

Fox

A. M.

. 2022. “Using Social Network Analysis to Examine Gun Violence.” Journal of Forensic Sciences 67(6):2230–41. 10.1111/1556-4029.15132

37.

Graif

Freelin

B. N.

Kuo

Y.-H.

Wang

Kifer

. 2021. “Network Spillovers and Neighborhood Crime: A Computational Statistics Analysis of Employment-Based Networks of Neighborhoods.” Justice Quarterly 38(2):344–74. 10.1080/07418825.2019.1602160

38.

Graif

Lungeanu

Yetter

A. M.

. 2017. “Neighborhood Isolation in Chicago: Violent Crime Effects on Structural Isolation and Homophily in Inter-Neighborhood Commuting Networks.” Social Networks 51:40–59. 10.1016/j.socnet.2017.01.007

39.

Groff

McCord

E. S.

. 2012. “The Role of Neighborhood Parks as Crime Generators.” Security Journal 25(1):1–24. 10.1057/sj.2011.1

40.

Groff

E. R.

Brenneman

Haberman

C. P.

. 2023 “Micro Units.” Pp. 51-6 in Understanding Crime and Place: A Methods Handbook edited by Groff

E. R.

Haberman

C. P.

. Philadelphia, PA: Temple University Press.

41.

Haberman

C. P.

2017. “Overlapping Hot Spots?: Examination of the Spatial Heterogeneity of Hot Spots of Different Crime Types.” Criminology & Public Policy 16(2):633–60. 10.1111/1745-9133.12303

42.

Haberman

C. P.

Kelsay

J. D.

. 2021. “The Topography of Robbery: Does Slope Matter?” Journal of Quantitative Criminology 37(3):625–45. 10.1007/s10940-020-09451-z

43.

Hatten

D. N.

. 2022. Where Gunshots Turn Fatal: A Geographic Examination of the Spatial Patterning of gun Violence. New York, NY: City University of New York.

44.

Hipp

J. R.

Faris

R. W.

Boessen

. 2012. “Measuring ‘Neighborhood’: Constructing Network Neighborhoods.” Social Networks 34(1):128–40. 10.1016/j.socnet.2011.05.002

45.

Hureau

D. M.

Braga

A. A.

. 2018. “The Trade in Tools: The Market for Illicit Guns in High-Risk Networks.” Criminology; An Interdisciplinary Journal 56(3):510–45. 10.1111/1745-9125.12187

46.

Johnson

N. J.

Roman

C. G.

. 2025. “Spillover Effects of Drug Markets on Gun Violence Across a Network of Neighborhoods in Three Cities.” Journal of Quantitative Criminology. 10.1007/s10940-025-09621-x

47.

Johnson

S. D.

2014. “How Do Offenders Choose Where to Offend? Perspectives from Animal Foraging.” Legal and Criminological Psychology 19(2):193–210. 10.1111/lcrp.12061

48.

Johnson

S. D.

Bowers

K. J.

. 2010. “Permeability and Burglary Risk: Are Cul-de-Sacs Safer?” Journal of Quantitative Criminology 26(1):89–111. 10.1007/s10940-009-9084-8

49.

Joyce Foundation

The

. 2024. Optimizing Gun Crime Intelligence: An Analysis of Current Usage and Recommendations for Improved Impact. Chicago, IL.

50.

Kelling

Graif

Korkmaz

Haran

. 2021. “Modeling the Social and Spatial Proximity of Crime: Domestic and Sexual Violence Across Neighborhoods.” Journal of Quantitative Criminology 37(2):481–516. 10.1007/s10940-020-09454-w

51.

Khanam

K. Z.

Srivastava

Mago

. 2023. “The Homophily Principle in Social Network Analysis: A Survey.” Multimedia Tools and Applications 82(6):8811–54. 10.1007/s11042-021-11857-1

52.

Kim

Y.-A.

Hipp

J. R.

. 2021. “Density, Diversity, and Design: Three Measures of the Built Environment and the Spatial Patterns of Crime in Street Segments.” Journal of Criminal Justice 77:101864. 10.1016/j.jcrimjus.2021.101864

53.

King

Wells

Katz

Maguire

Frank

. 2013. “Opening the black box of NIBIN: A descriptive process and outcome evaluation of the use of NIBIN and Its effects on criminal investigations.” U.S. Department of Justice, Office of Justice Programs, National Institute of Justice.

54.

Krause

R. W.

Huisman

Steglich

Snijders

. 2020. “Missing Data in Cross-Sectional Networks – An Extensive Comparison of Missing Data Treatment Methods.” Social Networks 62:99–112. 10.1016/j.socnet.2020.02.004

55.

Kreager

D. A.

Young

J. T.

Haynie

D. L.

Schaefer

D. R.

Bouchard

Davidson

K. M.

. 2021. “In the Eye of the Beholder: Meaning and Structure of Informal Status in Women’s and Men’s Prisons.” Criminology: An Interdisciplinary Journal 59(1):42–72. 10.1111/1745-9125.12260

56.

Kretschmer

Leszczensky

McMillan

. 2025. “Strong Ties, Strong Homophily? Variation in Homophily on Sociodemographic Characteristics by Relationship Strength.” Social Forces (1):1–20.

57.

Lee

Eck

J. E.

SooHyun

Martinez

N. N.

. 2017. “How Concentrated is Crime at Places? A Systematic Review from 1970 to 2015.” Crime Science 6(1):6. 10.1186/s40163-017-0069-x

58.

Linning

S. J.

Eck

J. E.

. 2021. Whose “Eyes on the Street” Control Crime?: Expanding Place Management into Neighborhoods. Cambridge: Cambridge University Press.

59.

MacDonald

Mohler

Brantingham

P. J.

. 2022. “Association Between Race, Shooting Hot Spots, and the Surge in gun Violence During the COVID-19 Pandemic in Philadelphia, New York and Los Angeles.” Preventive Medicine 165:107241. 10.1016/j.ypmed.2022.107241

60.

McMillan

Kreager

D. A.

Veenstra

. 2022. “Keeping to the Code: How Local Norms of Friendship and Dating Inform Macro-Structures of Adolescents’ Romantic Networks.” Social Networks 70:126–37. 10.1016/j.socnet.2021.11.012

61.

McPherson

Smith-Lovin

Cook

J. M.

. 2001. “Birds of a Feather: Homophily in Social Networks.” Annual Review of Sociology 27(1):415–44. 10.1146/annurev.soc.27.1.415

62.

Morenoff

J. D.

Sampson

R. J.

Raudenbush

S. W.

. 2001. “Neighborhood Inequality, Collective Efficacy, and the Spatial Dynamics of Urban Violence.” Criminology; An Interdisciplinary Journal 39(3):517–58. 10.1111/j.1745-9125.2001.tb00932.x

63.

Morselli

. 2009. Inside Criminal Networks, Vol. 8. Springer New York. 10.1007/978-0-387-09526-4

64.

Newton

. (2018). Macro-Level Generators of Crime, Including Parks, Stadiums, and Transit Stations (G. J. N. Bruinsma & S. D. Johnson, Eds.; Vol. 1). Oxford University Press. 10.1093/oxfordhb/9780190279707.013.16

65.

Nix

Huff

Wolfe

S. E.

Pyrooz

D. C.

Mourtgos

S. M.

. 2024. “When Police Pull Back: Neighborhood-Level Effects of de-Policing on Violent and Property Crime, a Research Note.” Criminology; An Interdisciplinary Journal 62(1):156–71. 10.1111/1745-9125.12363

66.

Novak

K. J.

King

W. R.

. 2020. “Evaluation of the Kansas City Crime Gun Intelligence Center.” Bureau of Justice Assistance. https://bja.ojp.gov/library/publications/evaluation-kansas-city-crime-gun-intelligence-center

67.

O’Brien

D. T.

Sampson

R. J.

. 2015. “Public and Private Spheres of Neighborhood Disorder: Assessing Pathways to Violence Using Large-Scale Digital Records.” Journal of Research in Crime and Delinquency 52(4):486–510. 10.1177/0022427815577835

68.

Papachristos

A. V.

Bastomski

. 2018. “Connected in Crime: The Enduring Effect of Neighborhood Networks on the Spatial Patterning of Violence.” American Journal of Sociology 124(2):517–68. 10.1086/699217

69.

Papachristos

A. V.

Braga

A. A.

Piza

Grossman

L. S.

. 2015a. “The Company You Keep? The Spillover Effects of Gang Membership on Individual Gunshot Victimization in a Co-Offending Network.” Criminology: An Interdisciplinary Journal 53(4):624–49. 10.1111/1745-9125.12091

70.

Papachristos

A. V.

Hureau

D. M.

Braga

A. A.

. 2013. “The Corner and the Crew: The Influence of Geography and Social Networks on Gang Violence.” American Sociological Review 78(3):417–47. 10.1177/0003122413486800

71.

Papachristos

A. V.

Murphy

J. P.

Braga

Turchan

. 2024. “The Importance of Neighborhood Offending Networks for Gun Violence and Firearm Availability.” Social Forces 103(2):780–801. 10.1093/sf/soae099

72.

Papachristos

A. V.

Wildeman

Roberto

. 2015b. “Tragic, but not Random: The Social Contagion of Nonfatal Gunshot Injuries.” Social Science & Medicine 125:139–50. 10.1016/j.socscimed.2014.01.056

73.

Pierce

G. L.

Spaar

Briggs

L. R.

. 1988. “The character of police work: Strategic and tactical implications.” Center for Applied Social Research, Northeastern University.

74.

Piza

E. L.

Hatten

D. N.

Mohler

G. O.

Carter

J. G.

Cho

. 2024. “Gunshot Detection Technology Effect on gun Violence in Kansas City, Missouri: A Microsynthetic Control Evaluation.” Criminology & Public Policy 23(4):863–92. 10.1111/1745-9133.12648

75.

Piza

E. L.

Welsh

B. C.

Farrington

D. P.

Thomas

A. L.

. 2019. “CCTV Surveillance for Crime Prevention: A 40-Year Systematic Review with Meta-Analysis.” Criminology & Public Policy 18(1):135–59. 10.1111/1745-9133.12419

76.

Police Executive Research Forum. 2017. “The “crime gun intelligence center” model: Case studies of the Denver, Milwaukee, and Chicago approaches to investigating gun crime.” https://www.policeforum.org/assets/crimegunintelligencecenter.pdf

77.

Ratcliffe

J. H.

Rengert

G. F.

. 2008. “Near-Repeat Patterns in Philadelphia Shootings.” Security Journal 21(1-2):58–76. 10.1057/palgrave.sj.8350068

78.

Ratcliffe

J. H.

Taniguchi

Groff

E. R.

Wood

J. D.

. 2011. “The Philadelphia Foot Patrol Experiment: A Randomized Controlled Trial of Police Patrol Effectiveness in Violence Crime Hotspots.” Criminology: An Interdisciplinary Journal 49(3):795–831. 10.1111/j.1745-9125.2011.00240.x

79.

Roberto

Braga

A. A.

Papachristos

A. V.

. 2018. “Closer to Guns: The Role of Street Gangs in Facilitating Access to Illegal Firearms.” Journal of Urban Health 95(3):372–82. 10.1007/s11524-018-0259-1

80.

Robins

Pattison

Kalish

Lusher

. 2007. “An Introduction to Exponential Random Graph (p*) Models for Social Networks.” Social Networks 29(2):173–91. 10.1016/j.socnet.2006.08.002

81.

Sampson

R. J.

2012. Great American City: Chicago and the Enduring Neighborhood Effect. Chicago, IL: The University of Chicago Press.

82.

Sampson

R. J.

Morenoff

J. D.

Gannon-Rowley

. 2002. “Assessing “Neighborhood Effects”: Social Processes and New Directions in Research.” Annual Review of Sociology 28(1):443–78. 10.1146/annurev.soc.28.110601.141114

83.

Santos

R. G.

Santos

R. B.

. 2015. “Practice-Based Research: Ex Post Facto Evaluation of Evidence-Based Police Practices Implemented in Residential Burglary Micro-Time Hot Spots.” Evaluation Review 39(5):451–79. 10.1177/0193841X15602818

84.

Schaefer

D. R.

2012. “Youth Co-Offending Networks: An Investigation of Social and Spatial Effects.” Social Networks 34(1):141–9.

85.

Schaefer

D. R.

Rodriguez

Decker

S. H.

. 2014. “The Role of Neighborhood Context in Youth Co-Offending.” Criminology; An Interdisciplinary Journal 52(1):117–39. 10.1111/1745-9125.12032

86.

Schnell

Braga

A. A.

Piza

E. L.

. 2017. “The Influence of Community Areas, Neighborhood Clusters, and Street Segments on the Spatial Variability of Violent Crime in Chicago.” Journal of Quantitative Criminology 33(3):469–96. 10.1007/s10940-016-9313-x

87.

Sherman

L. W.

Gartin

P. R.

Buerger

M. F.

. 1989. “Hot Spots of Predatory Crime: Routine Activities and the Criminology of Place.” Criminology; An Interdisciplinary Journal 27(1):27–56. 10.1111/j.1745-9125.1989.tb00862.x

88.

Steenbeek

Weisburd

. 2016. “Where the Action Is in Crime? An Examination of Variability of Crime Across Different Spatial Units in The Hague, 2001–2009.” Journal of Quantitative Criminology 32(3):449–69. 10.1007/s10940-015-9276-3

89.

Sturup

Gerell

Rostami

. 2020. “Explosive Violence: A Near-Repeat Study of Hand Grenade Detonations and Shootings in Urban Sweden.” European Journal of Criminology 17(5):661–77. 10.1177/1477370818820656

90.

Sturup

Rostami

Gerell

Sandholm

. 2018. “Near-Repeat Shootings in Contemporary Sweden 2011 to 2015.” Security Journal 31(1):73–92. 10.1057/s41284-017-0089-y

91.

Taylor

R. B.

2015. Community Criminology: Fundamentals of Spatial and Temporal Scaling, Ecological Indicators, and Selectivity Bias. New York, NY: New York University Press.

92.

Taylor

R. B.

Gottfredson

S. D.

Brower

. 1984. “Block Crime and Fear: Defensible Space, Local Social Ties, and Territorial Functioning.” Journal of Research in Crime and Delinquency 21(4):303–31. 10.1177/0022427884021004003

93.

Telep

C. W.

Hibdon

. 2017. “Identifying and Responding to Hot Spots: Are Crime Counts Enough?” Criminology & Public Policy 16(2):661–71. https://doi.org/10.1111/1745-9133.12304

94.

Tita

G. E.

Greenbaum

R. T.

. 2009 “Crime, Neighborhoods, and Units of Analysis: Putting Space in its Place.” Pp. 145–70 in Putting Crime in its Place, edited by Weisburd

Bernasco

Bruinsma

G. J. N.

. Springer New York. 10.1007/978-0-387-09688-9_7

95.

Tita

G. E.

Radil

S. M.

. 2010. “Making Space for Theory: The Challenges of Theorizing Space and Place for Spatial Analysis in Criminology.” Journal of Quantitative Criminology 26(4):467–79. 10.1007/s10940-010-9115-5

96.

Van Deuren

Diviák

Blokland

. 2025. “Co-offending among Outlaw Motorcycle Gang Members: The Role of Social and Geographical Proximity.” Journal of Criminal Justice 96:102340. 10.1016/j.jcrimjus.2024.102340

97.

Wang

Butts

C. T.

Hipp

J. R.

Jose

Lakon

C. M.

. 2016. “Multiple Imputation for Missing Edge Data: A Predictive Evaluation Method with Application to Add Health.” Social Networks 45:89–98. 10.1016/j.socnet.2015.12.003

98.

Wang

Kifer

Graif

. 2016. “Crime Rate Inference with Big Data.” Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, pp. 635–44. 10.1145/2939672.2939736

99.

Weisburd

2015. “The Law of Crime Concentration and the Criminology of Place.” Criminology: An Interdisciplinary Journal 53(2):133–57. 10.1111/1745-9125.12070

100.

Weisburd

Groff

Yang

S.-M.

. 2012. The Criminology of Place: Street Segments and Our Understanding of the Crime Problem. Oxford: Oxford University Press.

101.

Weisburd

Zastrow

Kuen

Andresen

. 2024. “Crime Concentrations at Micro Places: A Review of the Evidence.” Aggression and Violent Behavior 78:101979. 10.1016/j.avb.2024.101979

102.

Wells

. 2012. “Patterns of Near-Repeat gun Assaults in Houston.” Journal of Research in Crime and Delinquency 49(2):186–212. 10.1177/0022427810397946

103.

Lum

. 2020. “The Practice of Proactive Traffic Stops.” Policing: An International Journal 43(2):229–46. 10.1108/PIJPSM-06-2019-0089

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.23 MB