You are on page 1of 24

Lexington, Kentucky

Group Violence Intervention

Problem Analysis: Full Report

May 2019

1. Executive Summary ........................................................................................................................ 3

2. GVI Problem Analysis Overview ................................................................................................. 4
2.1. Distribution of Violent Crime ............................................................................................... 5
3. Methodology .................................................................................................................................... 6
3.1. Group Audit ............................................................................................................................... 6
3.2. Violent Incident Review ......................................................................................................... 7
4. Results ................................................................................................................................................ 9
4.1 Group Audit Results ............................................................................................................... 10
4.2 Overview of Incident Review Results ............................................................................... 10
4.3 Incident Review Results: Group Member Involvement (GMI) .................................... 12
4.4 Incident Review: Group Intelligence.................................................................................. 14
5. Discussion ....................................................................................................................................... 19
6. Limitations and Implications ...................................................................................................... 20
7. Next Steps and Implementation ............................................................................................... 21

National Network for Safe Communities 2 (646) 557-4760 •

1. Executive Summary
This report presents findings from the National Network for Safe Communities’ (NNSC) problem analysis
of serious violence in Lexington, Kentucky. The purpose of this analysis was to capture frontline law
enforcement insight into local violence dynamics in order to expand opportunities for preventing violence
in Lexington.

The problem analysis comprises two exercises: a group audit and a violent incident review. During the
group audit exercise, the Lexington Police Department (LPD) personnel identified 12 active violent groups,
and provided information regarding the relationships within these groups, the areas in which they
operate, and particularly violent and/or active group members. During the violent incident review, an
additional four groups were identified. Five inactive groups were also identified as having been historically
relevant to violence in Lexington, but which are acknowledged to no longer be active in the city.

Following the group audit, NNSC conducted the violent incident review, in which law enforcement
personnel reviewed five years of homicides (from January 2014 through December 2018) and three years
of nonfatal shootings (from January 2016 through December 2018). The violent incident review covered
117 homicides and 221 nonfatal shootings. NNSC and Lexington participants found that 34% of the
homicide incidents and 30% of the nonfatal shooting incidents reviewed were confirmed to include group
member involvement.

In cities across the country where NNSC has conducted analyses, NNSC has consistently found that street
groups typically drive around half of all serious violence, and often more. The above findings indicate a
lower level of group involvement in homicides and nonfatal shootings; however, the continued review
process which is detailed in our findings does indicate there may be more group involvement in serious
violence than initially identified.

While conducting a problem analysis, NNSC is often able to identify gaps in the available intelligence in a
given city. In Lexington, insight into group dynamics was limited to the formally documented gangs in the
city, and while the assembled officers were prepared to speak to their cases and/or the incidents to which
they had been first on the scene, their information on topics such as perpetrators’ and victims’ social
associations, prior or subsequent involvement in violent crime, and knowledge of each other in particular
incidents, was limited, creating information gaps in terms of associations and group affiliations. The
problem analysis highlighted the need for a routinized process to systematically operationalize street-
level intelligence in order to track the composition of street groups and their involvement in violence.
NNSC recommends that LPD institutionalize regular incident reviews to assess homicides and nonfatal
shootings to track group involvement in those incidents.

As is the case nationally, these groups and group members are involved in a significant portion of
homicides and shootings in Lexington, KY, and there is an opportunity to reduce serious violence through
the NNSC’s Group Violence Intervention. However, as the group member involvement for both homicides
and nonfatal shootings is lower than other jurisdictions where NNSC has conducted this work, and the
current existing intelligence does not allow for a clearer picture of group-involved violence in Lexington,
the magnitude of anticipated impact on serious violence may be smaller than other jurisdictions where
groups are known to be involved in violence at higher rates.

National Network for Safe Communities 3 (646) 557-4760 •

2. GVI Problem Analysis Overview
On January 29th and 30th, National Network for Safe Communities (NNSC) staff conducted a site visit to
Lexington, Kentucky to complete a two-part problem analysis with local law enforcement partners. As
discussed above, the goal of this exercise was to capture frontline officers’ insights into local violence
dynamics and to assess opportunities for preventing violence using the NNSC’s strategies, particularly the
Group Violence Intervention (GVI). To that end, the NNSC specifically gathered intelligence on active,
violent street groups operating in Lexington. NNSC uses the term “group” instead of “gang,” defining
street groups as two or more people who engage in violence and/or criminal activity together. A group
may be considered an official “gang” according to statutory definitions but may also be a loosely knit social
network of individuals or family members that does not meet the statutory definition of a gang.

The NNSC’s Group Violence Intervention has repeatedly shown that cities can dramatically reduce
violence when community members and law enforcement join together to directly engage with active,
violent street groups and clearly communicate: (1) a credible community message against violence; (2) a
credible law enforcement message about the consequences of further violence; and (3) a genuine offer
of help for those who want it. To do this, a city must assemble a partnership of law enforcement, social
service providers, and community actors (parents, clergy, street outreach workers, neighborhood
associations, ex-offenders and others) to engage in a sustained relationship with violent groups. A key
moment in the strategy is a “call-in,” a face-to-face meeting between group members and the partnership,
repeated at intervals as necessary. The partners deliver key messages to group members: violence is
wrong and has to stop; the community needs them alive, out of prison and with their loved ones; help is
available to all who will take it; and violence will be met with clear, predictable and certain consequences.

A central operational shift is that law enforcement puts groups on prior notice that law enforcement will
meet group-involved violence with a specific and swift response directed at the group as a whole rather
than at individuals. Individual violent offenders receive the same enforcement attention as they had
previously. However, their fellow group members get new law enforcement attention for any crimes
committed, outstanding warrants, probation and parole violations, open cases and other possible legal
levers that can be pulled to apply unwanted pressure. The intervention combines this prior notice of
potential law enforcement exposure with a powerful message against violence from community members
whom the group members respect and a reorganized, streamlined social service structure tailored for
group members who need and want support. The aim is to create collective accountability to reduce
informal peer dynamics in the group that promote violence, reassert community standards against
violence, and offer genuine help for those who will take it. When considering whether GVI is appropriate
for a particular city, law enforcement personnel may begin by identifying individuals most closely
associated with violence and the groups – if any – to which they belong.

The following report details the NNSC’s data collection methodology and summarizes the key findings.
Specifically, this report documents the identification of violent street groups. This document also analyzes
the basic demographics of victims and suspects in homicides and nonfatal shootings, the nature of the
relationship between victims and suspects, as well as the level of law enforcement intelligence as it relates
to serious violence and those involved in it. Finally, the report describes the current implications of the

National Network for Safe Communities 4 (646) 557-4760 •

data analysis and immediate next steps required to move forward with implementation of violence
prevention work.

2.1. Distribution of Violent Crime

The NNSC’s problem analysis exercises are grounded in the emerging science of violence prevention.
Research has established that violent crime in America is distributed unevenly across communities, and
that this concentration – geographic, demographic, and social – has important implications for
intervention. This section, which appeared in the Initial Problem Analysis Memo delivered to the Lexington
partners, is included in this report to situate research with violence prevention efforts.

Half of America’s homicides occur in just 127 American cities; more than a quarter of homicides occur in
neighborhoods that contain just 1.5% of the country’s population, and in particular cities as little as 5% of
blocks generate over half of all complaints of crime and violence (Aufrichtig, Beckett, Diehm, & Lartey,
20171; Weisburd, 20152). Those neighborhoods where violence concentrates are often next door to
communities that experience little to no serious violence, as in Chicago, where neighborhood homicide
rates vary between under 4 per 100,000 to over 80 per 100,000 (Mason, 20173). The most serious violence
– homicides and shooting violence – are also concentrated socially and correspond to particular
demographic profiles and social connections. Victimization rates vary widely across gender, age, and race:
for example, black men 15 to 34 are seventeen times more likely to be victims of homicide than non-
Hispanic white males of the same age (Cook & Pollack, 20174).

Serious violence is also influenced by interpersonal connections. According to Tracy and collaborators:
"exposure to a victim or perpetrator of violence in one’s interpersonal relationships and social networks
increases the risk of individual victimization and perpetration" (Tracy, Braga, & Papachristos, 20165).
Papachristos et al. have analyzed social networks using arrest records and determined the consistent
concentration of a city’s gun violence within small social networks and the "transmission" of victimization
along those social connections (Green, Horel, and Papachristos, 20176). This research identifies subsets of
society that experience tremendously disproportionate amounts of the most serious community violence.

NNSC’s deep experience exploring these dynamics in cities across the country – more than two dozen in
the last six years – has confirmed that these dynamics are remarkably consistent across cities, and that
further, the social connections described by Papachristos and partners above are often recognizable on
close examination as street groups. In a review of 23 sites where NNSC has conducted a problem analysis
(including full cities and city segments or precincts), NNSC found that, on average, 0.6% of a city’s total
population was recognizable to frontline law enforcement as involved in this violent street group dynamic,
and that this tiny fragment of the overall population was, in turn, connected to on average 50% of
homicides. Group members were at least 20 times as likely as non-group members to be involved in a

1 Aufrichtig, A., Beckett, L., Diehm, J., & Lartey, J. (2017). Want to fix gun violence in America? Go local. The Guardian.
2 Weisburd, D. (2015). The law of crime concentration and the criminology of place. Criminology, 53(2), 133-157.
3 Mason, M. (2017, December). Homicide in Chicago community areas 2007–2015: Concentrated risk and stable rates. Illinois

Violent Death Reporting System.

4 Cook, P. J., & Pollack, H. A. (2017). Reducing access to guns by violent offenders. RSF.
5 Tracy, M., Braga, A. A., & Papachristos, A. V. (2016). The transmission of gun and other weapon-involved violence within social

networks. Epidemiologic reviews, 38(1), 70–86.

6 Green, B., Horel, T., & Papachristos, A. V. (2017). Modeling contagion through social networks to explain and predict gunshot

violence in Chicago, 2006 to 2014. JAMA internal medicine, 177(3), 326–333.

National Network for Safe Communities 5 (646) 557-4760 •

homicide, and in some jurisdictions that number rose to 40 times. What this tells us is that identifying
these active groups and their members allows law enforcement to accurately and effectively target the
population in their city most likely to perpetrate and be victimized by serious violence.

Legal and professional framings around such issues as “gangs”, “gang validation”, “gang statutes” and
“gang-related violence” can have a profound impact on frontline law enforcement’s perceptions of these
dynamics. American policing’s longstanding focus on formal definitions of gang membership; descriptions
of gangs as hierarchical, organized, and business-oriented; and focus on gang-involved violence as
violence conducted in furtherance of the gang’s interest often obscures what is otherwise apparent to
officer at the street level: the groups most relevant to local violence dynamics are not necessarily “gangs”
by any legal definition, nor do most of their activities have anything to do with the gang’s interest. Rather,
these groups are small, loosely connected, and fluid; the violence they are involved in is typically related
to issues of disrespect, standing personal disputes and intimate relationships, and small amounts of
money or drugs. The formal legal and professional processes of the criminal justice system that govern
intelligence management, arrests, and convictions are not responsive to this frontline reality; they are
designed to support the aforementioned longstanding formal and legal definitions of gangs and gang-
related violence. The NNSC’s Group Violence Intervention was created in partnership with frontline law
enforcement officers as a strategy that could align policing’s intelligence-gathering and law enforcement
responsibilities with the officers’ lived realities of group dynamics around violence.

3. Methodology
The problem analysis consists of two parts – the group audit and violent incident review – which will be
discussed in detail in the following sections. NNSC recommends that cities conduct a problem analysis
prior to GVI implementation to identify active groups and assess the proportion of violence that involves
group members or is otherwise connected to group dynamics.

Ideally, frontline law enforcement personnel participate as one cohort in both the group audit and violent
incident review. This allows for the collection of higher quality information; in addition, participating as
one group often leads to strengthening working relationships between the units and partners present. In
Lexington, this cohort included frontline police officers and detectives from Lexington Police Department
(LPD) including representation from different units – gang, robbery/homicide, and patrol. At the outset of
the problem analysis, the NNSC team told the participants that their contributions would be anonymous
and all the information shared during the research exercises would be confidential and protected under
a data sharing agreement in place between NNSC and Lexington Police.

3.1. Group Audit

The beginning step of the problem analysis always starts with the NNSC guiding the assembled law
enforcement personnel through a group audit, which gathers information on active violent street groups
(gangs, crews, sets, cliques, etc.). As a reminder, NNSC uses the term “group” instead of “gang,” defining
street groups as two or more people who engage in violence and/or criminal activity together who may
or may not meet the statutory definition of a “gang”. The purpose of the group audit is: (1) to identify all
groups contributing to the most serious violence in Lexington and (2) to identify those group-involved
individuals who are most closely associated with violence (as victims or offenders). The exercise is

National Network for Safe Communities 6 (646) 557-4760 •

designed to draw primarily from the street experience of knowledgeable police officers and other
frontline personnel, rather than from formal or written intelligence.

During the group audit, the law enforcement participants from Lexington identified active groups and the
following specific characteristics of the groups:

 Estimated number of members

 Estimated age range of members
 Activities of the group (e.g., assaults, homicides, narcotics sales)
 Territory, if applicable
 Impact players (by name)
 Group alliances
 Group feuds
 Current level of violence
 Current level of organization

NNSC’s goal is to examine the data along two dimensions: (1) group characteristics and (2) alliances and
conflicts between groups, when possible. Since the group audit and the incident review both generated
information on active, violent group members, the incident review, discussed later, typically includes data
from both exercises.

3.2. Violent Incident Review

Following the group audit, the research team facilitated the violent incident review. The purpose of the
violent incident review is to ascertain the nature of a city’s violent crime problem by retrospectively
reviewing each homicide and nonfatal shooting reported within the city for a pre-determined timeframe.
Prior to the violent incident review, NNSC staff explained to the law enforcement cohort that the goal was
to gather information about these incidents that is not routinely captured in the official records, such as
officers’ impressions and “word on the street.”

The NNSC team also explained the differences between “group-motivated” or “gang-motivated” and
“group member-involved” (GMI), reiterating the importance of using the group member-involved
definition to gauge the level of group violence in Lexington. Identifying GMI incidents captures all violence
in which group members participated, as victims or perpetrators, irrespective of motivation. NNSC uses
this measure because all violence connected to groups shares a specific and powerful influence: the group
itself. Any type of social group establishes expected norms and conduct for its members. This is no different
for street groups involved in violence, which have established norms and conduct such as illegal gun
carrying and the promotion of violence to settle disputes. These norms and the social pressure applied by
the group facilitate violence and encourage behaviors, which pose particular challenges to public safety.
Group culture that places a premium on status and respect, in addition to group acquisition and sharing of
weapons, means group norms and activities have significant effects on violence committed by individual
group members, even beyond violence specific to promoting the group’s interests. By assessing “group
member-involved” incidents, rather than limiting the analysis to “group-motivated” incidents, the NNSC
team captures a full measure of the violence connected to street group dynamics.

National Network for Safe Communities 7 (646) 557-4760 •

Using a PowerPoint presentation with one slide per incident, the research team systematically guided the
participants through each incident, inquiring about the following measures:

 Incident known to the room of law enforcement partners

 Narrative of the incident
 Violence was start of/part of a retaliatory cycle
 Victim(s) known to the law enforcement cohort prior to the incident
 Victim’s group affiliation
 Suspect(s) known to the law enforcement cohort prior to the incident
 Suspect’s group affiliation
 Victim/suspect relationship

A total of 338 incidents were reviewed during the exercise – 117 homicides and 221 nonfatal shootings
(NFS), spanning from January 2014 through December 2018 and January 2016 through December 2018
respectively. The purpose of this incident review was: (1) to understand the context and driving causes
of the most serious violence in Lexington and (2) to determine the extent to which the street groups
identified in the group audit contributed to the most serious violence in the city for the given time period.
The information from this session can also be used as a baseline for comparison against future violent
incident data.

The violent incident review looks particularly for insights into and patterns of group member-involved
homicides and nonfatal shootings. The process for how the NNSC defines incidents as GMI is as follows.
An incident is coded as “confirmed group member-involved” if any of the following is true:

 Victim is a known group member.

 Suspect is a known group member.
 Victim’s and/or suspect’s group involvement is unknown, but the law enforcement participants
determined to a moral certainty that one, the other, or both is group involved.

An incident is classified as “likely group member-involved” if any of the following is true:

 The victim and/or suspect are loosely associated with a group of individuals, but the associations
are not conclusively related to collaborative criminal and/or violent behavior.
 Multiple victims and/or suspects are involved in the incident, and the circumstances of the
incident suggest a group dynamic.

An incident is classified as “unknown” if there is not enough information to determine group member
involvement. Lastly, an incident is classified as “not GMI” if law enforcement knows that neither the
victim(s) nor the suspect(s) involved in the incident are associated with a group and nothing about the
incident suggests a group dynamic. To visualize how the NNSC codes incidents as GMI, Figure 1 breaks
this out into a simplified flowchart that lays out example steps to coding an incident as group member-
involved focusing on the four categories of confirmed (“Yes”) GMI, not GMI (“No”), unknown GMI (“Unk”),
and likely GMI (“Likely).

National Network for Safe Communities 8 (646) 557-4760 •

Figure 1. GMI Incident Review Coding Process Diagram

Information and current intelligence on the complex group dynamics within a city are usually scattered
among frontline personnel and not systematically collected and analyzed. Therefore, getting a current
snapshot of local group and violence patterns and concentrations is critical. The problem analysis and its
subsequent report seek to synthesize frontline intelligence across different agencies and units. Unpacking
serious violence in a city during the problem analysis allows us to better understand violence dynamics by
helping us answer certain questions such as: what proportion of the city’s serious violence involve a group
member either as a suspect or victim? To what extent are victims and suspects of serious violence
connected to groups, and which groups are driving the violence? How do non-GMI incidents compare to
those that are GMI incidents? Do victims and suspects know each other prior to an incident? What can
we learn about various characteristics, such as intimate partner violence, that overlap with the incidents
reviewed? These diagnostic questions help lay a foundational understanding of serious violence and its
intersection with group member involvement. The answers to these questions as they pertain to
Lexington are laid out in the Results section and then contextualized in the Implications and Limitations
section that follows.

4. Results
The following sections put forth NNSC’s analysis of the data collected during the problem analysis
exercises, beginning with the findings from the group audit before moving on to the findings from the
violent incident review.

National Network for Safe Communities 9 (646) 557-4760 •

4.1 Group Audit Results
The law enforcement partners in Lexington identified a total of 16 violent street groups: 12 during the
group audit and four more during the violent incident review. Law enforcement also identified five
inactive groups that had been historically relevant to violence in Lexington (including during the period
covered during the violent incident review), but which are acknowledged to no longer be active in the

4.2 Overview of Incident Review Results

This section describes basic insights about the incidents reviewed during the problem analysis. The goal
of this section is to present an overview of what serious violence in Lexington looks like from a broad lens.

Table 1 presents simple counts of the number of incidents by type (homicide and nonfatal shootings).
For the 117 homicides from January 2014 through December 2018, there were 119 victims7 and 128
identified suspects. There were 267 victims and 83 identified suspects for nonfatal shootings from January
2016 through December 2018. Table 2 displays the number of victims by incident type, indicating that
there were two homicides and 26 NFS involving more than one victim. Table 3 displays the number of
identified suspects for a given incident – 21% of homicides and 68% of NFS did not have an identified

Table 4 looks at the gender of victims and suspects. Eighty-one percent of homicide victims and 84% of
NFS victims were male. Eighty-eight percent of homicide suspects were male, whereas 83% of identified
suspects in NFS incidents were male. Figure 3 reflects the age of victims and identified suspects across
all incidents. Fifty percent of victims and 51% of suspects were between the age of 18 and 30.8 Table 5
reflects the individuals known to police broken down by incident type. Forty-eight percent of identified
suspects in NFS incidents were previously known to police whereas 44% of identified homicide suspects
were previously known to police.

Table 1. Homicide and Nonfatal Shooting Incident and Victim Counts

Counts of Incidents Homicides Nonfatal Shootings

Incidents 117 221

Victims 119 267
Identified Suspects 128 83
Note that 13 homicide incidents involved victims surviving from nonfatal gunfire; these were not added to the nonfatal shooting incident counts.

7 There were 13 homicides involving victims who survived gunfire. The victims of these incidents were included in the nonfatal
shooting victim totals if the homicide involved victims who survived gunfire.
8 Three percent of victims and 16% of suspects were missing date of birth or age data in the problem analysis slides provided by


National Network for Safe Communities 10 (646) 557-4760 •

Table 2. Number of Victims by Incident Type

Number of Victims Homicides Nonfatal Shootings

1 115 208
2 2 20
3 0 5
4 0 1

Table 3. Number of Identified Suspects by Incident Type

Number of Suspects Homicides Nonfatal Shootings

0 25 150
1 73 63
2 7 6
3 8 1
4 3 0
5 1 1

Table 4. Victim and Suspect Gender by Incident Type

Homicide Nonfatal Shooting
Gender Homicide Suspects Shooting
Victims Victims
# (%) # (%) # (%) # (%)
Male 96 (81%) 113 (88%) 225 (84%) 69 (83%)
Female 23 (19%) 9 (7%) 36 (14%) 8 (10%)
Missing 0 (0%) 6 (5%) 6 (2%) 6 (7%)
Total 119 128 267 83

National Network for Safe Communities 11 (646) 557-4760 •

Figure 3. Victims and Suspect Age – All Incidents

Table 5. Individuals Known to Police by Incident Type

Homicide Nonfatal Shooting
Known to PD Homicide Suspects Shooting
Victims Victims
# (%) # (%) # (%) # (%)
Yes 39 (33%) 56 (44%) 90 (34%) 40 (48%)
No 72 (60%) 54 (42%) 159 (59%) 31 (37%)
Unclear 8 (7%) 18 (14%) 18 (7%) 12 (15%)

4.3 Incident Review Results: Group Member Involvement (GMI)

The NNSC categorized homicide and nonfatal shooting incidents based on the involvement of individuals
associated with the city’s active, violent street groups. Incidents involving a member of a violent group
(either as a victim or suspect) are group member-involved (GMI) incidents. The explicit goal of this section
is to calculate the proportion of serious violence in Lexington that involves group members and/or group
violence contexts.

The violent incident review covered 117 homicides and 221 nonfatal shootings (Table 6).9 The NNSC and
Lexington participants found that 34% of the homicide incidents and 30% of the nonfatal shooting
incidents reviewed were confirmed to include group member involvement. These numbers increase to
45% and 42%, respectively, when including incidents coded as “likely GMI”. The tables and figures below
collectively suggest that Lexington is experiencing, at minimum, moderate levels of group member-
involved violence and that further group member-involved violence may be obscured by intelligence gaps
as mentioned throughout this report and specifically in Sections 6 and 7. These levels of group member
involvement indicate that a successful implementation of NNSC’s group violence intervention should
result in significant reductions in overall shootings and homicides.

9 Note that for this analysis, one incident was removed because it was a shots-fired incident and not a nonfatal shooting.

National Network for Safe Communities 12 (646) 557-4760 •

Table 6. GMI Coding Breakdown by Count and Percentage - All Incidents
GMI Homicides
Yes 40 (34%) 66 (30%)
Likely 13 (11%) 27 (12%)
Unknown 22 (19%) 92 (42%)
No 42 (36%) 36 (16%)
Total 117 (100%) 221 (100%)
Incident counts of homicides (2014-2018) and nonfatal shootings (2016- 2018). Note that 13 homicide incidents involved victims
surviving from nonfatal gunfire; these were not added to the nonfatal shooting incident counts.

Figure 4 below presents the results for group member involvement in homicides for the five-year period
reviewed. Overall, 45% of homicides between 2014 and 2018 were confirmed as involving a group
member or were likely to have involved a group member. Of the 117 homicide incidents covered during
the violent incident review, 40 incidents or 34% were confirmed as having group member involvement.
The other 11% of incidents were considered likely to have involved a group member because of possible
group involvement on the part of a named victim or suspect and/or a larger contextual link to group
dynamics (e.g. the incident was described as retaliation on behalf of another victim, the incident involved
multiple unidentified perpetrators acting together, etc.). Figure 4 also shows both a difference in the
percentage of confirmed GMI incidents and a marked increase in homicides considered unknown before
and after 2015.

Figure 4. Homicide GMI Percentage by Year

National Network for Safe Communities 13 (646) 557-4760 •

Figure 5 below illustrates the same concept applied to the three years of nonfatal shootings. Overall, 42%
of nonfatal shootings were either group member-involved or likely to have involved a group member.
Thirty percent of nonfatal shootings were confirmed to be GMI, slightly less than that of homicide
incidents. Additionally, 12% of nonfatal shooting incidents were coded as having a likely connection to
group involvement. Across the three years of nonfatal shootings, not only were the percentages of GMI
incidents consistent, but those of likely GMI and unknown incidents also remained stable.

Looking at the combination of homicides and nonfatal shootings together from 2016 through 2018
provides a more complete view of group-involved violence in Lexington. From 2016 through 2018, 41%
of all incidents – homicides and nonfatal shootings – were either definitively group member-involved
(29%) or likely group member-involved (12%). Based on previous experience supporting jurisdictions who
are considering GVI implementation, NNSC has found that sites typically begin implementation with a GMI
percentage between 40 and 60 percent. The findings here suggest that street-level violence dynamics in
Lexington are consistent, but on the lower range with NNSC’s national experience.

Figure 5. Nonfatal Shootings GMI Percentage by Year

4.4 Incident Review: Group Intelligence

This section explores violence dynamics specifically relevant for implementing GVI. Based on national
research, NNSC knows that group involved violence tends to involve individuals who know each other and
that groups often have longstanding vendettas or beefs with other groups. Group members also tend to
experience disproportionate victimization. To begin understanding the extent to which these elements
are present in Lexington, NNSC looked specifically at data points to illuminate these dynamics. To do this,
the NNSC team coded for the following 15 variables to further unpack the context of the violent incidents
under review:

National Network for Safe Communities 14 (646) 557-4760 •

 Robbery: did the incident involved a robbery?
 Money: was money a driver for the incident?
 Drug activity: did this incident involved a drug activity or transaction?
 Disrespect: was the main context of the incident due to disrespect?
 Retaliation: was this incident in retaliation for prior violence?
 Feud: was this incident due to an ongoing feud between groups or people familiar with each
 Unintended: was the victim an unintended victim?
 Intimate Partner Violence (IPV): did the incident involve former or current intimate partners?
 IPV spillover: did this incident involve a third party-victim connected to an IPV incident?
 Internal: was this incident internal violence within a group?
 Accidental: was the incident an accident?
 Justified: was the incident justified?
 Family violence: was the incident involving violence between family members not including IPV?
 Self-inflicted: was this incident a self-inflicted injury?
 Mental health: was there a significant mental health nexus related to this incident?

These above variables are not mutually exclusive: for example, an incident described by the cohort as “a
drug deal gone bad” may have been coded as involving money, a robbery, and drug activity, depending
on the information provided by the officers present.

Figure 6 displays the intersection of these variables with group involvement across all incidents under
review. Thirty-one percent of “confirmed GMI” and “likely GMI” incidents involved a robbery, and 29% of
the same incidents involved money in some way (e.g. dispute over money owed). Additionally, 28% of
“confirmed GMI” or “likely GMI” incidents involved a drug activity (e.g. a drug robbery, a drug deal, a
dispute among drug users at a party), 21% involving a disrespect (e.g. an argument, an online slight, etc.),
15% involved retaliation, and 14% involved an ongoing feud of which law enforcement was aware. In
short, violence in Lexington, particularly group violence, is characterized by disputes over money, drugs,
retaliation and ongoing feuds.

National Network for Safe Communities 15 (646) 557-4760 •

Figure 6. Incident Characteristics – All Incidents


Percentage (%) of All Incidents 30





No or Unknown GMI Incident GMI or Likely GMI Incident

Figures 7 and 8 below present similar information on how these additional characteristics intersect with
GMI homicides and GMI nonfatal shootings, respectively. The pie charts demonstrate the previously
discussed GMI classification percentages; the breakouts to the right show the characteristics of confirmed
GMI incidents. For homicides (Figure 9), the most frequent single characteristic was drug activity (18 GMI
homicides) followed by disrespect (14). For homicides coded as not group member-involved (left
breakout), the most frequent characteristic is intimate partner violence (14 incidents) followed by
disrespect (10), drug activity (10), and family violence (10). For GMI nonfatal shootings (Figure 10), the
top three characteristics are robbery, drug activity, and disrespect. For the 36 “not GMI incidents”, the
top characteristics were accidental, IPV spillover, and disrespect.

Figure 7. Incident Characteristics – All Homicides

National Network for Safe Communities 16 (646) 557-4760 •

Figure 8. Incident Characteristics – All Nonfatal Shootings

Extending beyond these 15 variables, the NNSC also analyzed other variables essential for understanding
group violence dynamics. NNSC learned during the problem analysis which incidents were currently open
cases as these were indicated on the slides. A total of 189 incidents (56% of the total) covered during the
problem analysis were considered to be open cases: 29 homicides (25%) and 160 (72%) nonfatal
shootings. Twenty-one out of the 29 open homicide cases are either likely or confirmed GMI incidents,
whereas 69 of 160 open NFS cases were either likely or confirmed GMI incidents. NNSC also coded
incidents based on whether a victim and/or witness was completely uncooperative and found 118 (62%
of total) such cases – of these, 109 (92%) are currently open cases, showing a clear correlation between
uncooperative victims or witnesses and open cases.

The NNSC looked at whether at least one suspect was identified in a given incident and how this interacts
with different GMI classifications and incident types. As Table 7 demonstrates, law enforcement identified
at least one suspect for 95% of homicides and 81% of NFS coded as “not GMI”. In contrast, law
enforcement identified at least one suspect in 71% of homicides but only 31% of NFS for incidents coded
as “likely GMI” or “confirmed GMI”. The law enforcement partnership is better able to identify suspects
in cases that do not involve group members, as many of the incidents involving group members are still
open cases and involve uncooperative victims and/or witnesses.

Table 7. At Least One Suspect Identified – by GMI Code

At least One Suspect Identified in Incident

Yes or Likely GMI Unknown GMI Not GMI

Homicides 71% 64% 95%
Nonfatal shootings 31% 14% 81%

NNSC codes an incident as “confirmed GMI” when the suspect and/or the victim is known by law
enforcement to be a member of an active, violent group; or when the situation/context is definitively

National Network for Safe Communities 17 (646) 557-4760 •

group-related. Figure 9 illustrates that suspects drove the categorization of GMI for homicides (52% of
confirmed GMI homicides were coded as such because of the suspect only) while NFS confirmed GMI
incidents were driven by victim (61% of confirmed GMI NFS incidents).

Figure 9. GMI Coding Source – by Incident Type

Figure 10 below assesses the extent to which victims and suspects knew each other before the incident
under review took place. For confirmed GMI and likely GMI homicides, 57% of cases involved suspects
and victims that knew each other prior to the incident; this was also true in 37% of confirmed GMI and
likely GMI nonfatal shootings. Many of the GMI or likely GMI homicides where the suspect and victim did
not know each other were cases in which the victim was an unintended target. In contrast, victims and
suspects knew each other in only 80% of non-GMI homicides and 35% of non-GMI NFS (in this case, non-
GMI refers to incidents coded as both “not GMI” and “unknown GMI”). The majority of these non-GMI
incidents where the suspect and victim knew each other involved intimate partner violence.

National Network for Safe Communities 18 (646) 557-4760 •

Figure 10. Victim and Suspect Knew Each Other Prior – All Incidents

5. Discussion
The results displayed in this report are largely in line with NNSC’s prior experience. This report
demonstrates that Lexington is in fact experiencing group dynamics driving violence similar to those found
in cities that have successfully implemented GVI. Lexington has active violent groups that were discussed
during the group audit and the incident review. The contours of violence in Lexington certainly points to
many of the hallmarks of group violence occurring in cities across the US. A large proportion of the
violence is connected to group members, both in terms of victimization and perpetration. As of January
2019, when these data were gathered, this report finds that a significant proportion of the city’s shootings
– at minimum 30% to upwards of 42%, including likely GMI incidents – and a significant and increasing
proportion of the city’s homicides – from 34% to 45%, including likely GMI incidents – are associated with
groups and group members. In NNSC’s experience, there is reason to believe that “likely” incidents are in
fact GMI; however, NNSC does not code incidents as “confirmed GMI” without conclusive statements
from law enforcement that the victim, suspect, or both is in fact a group member. An innocent bystander
shot during a drive-by shooting at a street shrine for a previously killed group member, on the anniversary
of that group member’s death, in which no suspect information was gathered, is an example of an incident
that would be coded as “likely GMI”, but is almost certainly driven by group dynamics.

The data show that a very small number of group-involved individuals, who largely are known to law
enforcement, are associated with the aforementioned significant proportion of the city’s most serious
violent crime. These individuals involved tend to be males between the ages of 18 and 30. The findings
also show indications that group-involved violence in Lexington is cyclical, retaliatory, and exacerbated by
standing beefs. The most common contexts for GMI homicides and nonfatal shootings in Lexington were

National Network for Safe Communities 19 (646) 557-4760 •

disputes over money or drugs, robberies gone awry, and perceived disrespect. GMI incidents also tended
to involve people who knew each other (57% of the GMI homicides and 37% of nonfatal shootings, noting
that the NFS percentage is low because of lacking suspect information) and who are known to law
enforcement. These data are consistent with national research underscoring the extent to which violence
moves through and between groups and other types of social networks. Furthermore, the NNSC has
reason to believe that retaliation and ongoing feuds are likely higher than what is stated in this report due
to group intelligence limitations. Investment in bolstering group intelligence (e.g. periodic audits of group
membership and dynamic, as well as regular reviews of violent incidents) will uncover these cycles of
retaliation exhibited in cities with serious group violence dynamics. Overall, these findings are consistent
with what NNSC has witnessed in other cities where this type of analysis has been completed and suggest
that a successful implementation of GVI will produce a significant reduction in the city’s homicides and

6. Limitations and Implications

The problem analysis was designed to capture frontline law enforcement insight into local violence
dynamics in order to expand opportunities for preventing violence in Lexington. This process is intended
to unpack serious violence within the city, starting with a blank slate and making no assumptions about
the context of the violence. The group audit and violent incident review are structured both as data
collection exercises to inform this problem analysis report, and as examples of processes critical to the
successful implementation of the NNSC’s Group Violence Intervention and related strategies.

While conducting a problem analysis, NNSC is often able to identify gaps in the available intelligence in a
given city. In Lexington, insight into group dynamics was limited to the formally documented gangs in the
city, and while the assembled officers were prepared to speak to their cases and/or the incidents to which
they had been first on the scene, their information on topics such as perpetrators’ and victims’ social
associations, prior or subsequent involvement in violent crime, and knowledge of each other in particular
incidents, was limited.

As highlighted in the Results section, the law enforcement representatives assembled identified 17 groups
during the group audit, all of them formal gangs according to the Lexington Police Department’s gang
definition.10 While these findings are somewhat consistent with NNSC’s prior experience conducting
group audits in similar sized cities, there are a few limitations to the findings in Lexington. Of the 12 active
groups identified, the partners assembled had limited insight into their most active members, any
alliances or conflicts with other groups, or which, if any, recent violent incidents were connected with
their members. Therefore, NNSC did not provide in this report certain supplemental information we
typically would, including a detailed group audit table outlining all the findings discussed during the group
audit, a sociogram displaying current group alliances and conflicts, or an estimated total group population
which allows us to determine the proportion of the overall city’s population connected to serious violence.

NNSC also found dynamics suggestive of some number of unnamed groups — not gangs, but informal,
loosely knit, crews, and cliques — active in Lexington and contributing to homicides and nonfatal
shootings, often with connections to drug business and robberies. However, information on these

10 Four additional active groups were identified during the violent incident review.

National Network for Safe Communities 20 (646) 557-4760 •

unnamed groups may not be currently captured by law enforcement intelligence and were not known by
the law enforcement cohort present for the analysis. A tool like social network analysis uses co-arrest
records, field stop data, incident data, and etc. to illustrate associations between individuals who
commonly commit crimes or are arrested together or as part of the same incident. This can help build on
existing group intelligence especially for lesser known, more loosely affiliated groups by looking at co-
offending networks to help identify group members. All of this is an effort to narrow law enforcement’s
focus on the small number of people actively engaging in violence. For the purposes of this report,
however, NNSC did not have access to arrest records that linked multiple individuals arrested as part of
the same incident, and was unable to conduct the social network analysis to show such associations.

Moreover, the NNSC is intentionally conservative in coding an incident as “confirmed group member-
involved;” incidents that may be GMI are coded as “likely GMI” or as “unknown GMI” unless it is certain
that the incident was in fact group involved. As discussed in Section 4.4, Lexington presented intelligence
gaps, albeit predictable and normal, especially with regard to nonfatal shootings, which likely obscure a
fuller picture of group dynamics. An important difference found by analyzing nonfatal shootings (relative
to homicides) is that the percentage of nonfatal shootings in which officers were hesitant to determine
group member involvement – incidents coded as “unknown” – was more than double that of homicides
at 42% (92 nonfatal shooting incidents). In the case of homicides, there are several factors that may
contribute to a fuller understanding of shooting dynamics in Lexington. In general, it is likely that the
results here underestimate the proportion of nonfatal shooting incidents that are GMI. It is normal for
frontline personnel to be able to discuss group member involvement in homicides with a greater degree
of certainty and consistency than in nonfatal shootings. This national trend was reflected in Lexington –
in nearly 20% of homicides and 42% of NFS incidents reviewed, personnel in the room were unable to
confirm whether an incident was group involved or not. While some incidents’ group involvement will be
forever unknown due to a lack of intelligence, information sharing, and/or cooperation from witnesses,
and victims, NNSC’s experience has been that, as further intelligence emerges, a certain proportion of
these “unknown” incidents will eventually be confirmed as group member-involved.

These limitations around group intelligence are not unique to Lexington and the following section outlines
operational recommendations to directly address these limitations which NNSC can help support. These
recommended next steps are intended to continue building intelligence on the small, highest-risk
population and to work toward the potential implementation of GVI in Lexington.

7. Next Steps and Implementation

NNSC recommends the following next steps to build on the intelligence gathering practices established
during the problem analysis and to support the potential implementation of GVI. The following areas could
be developed to ensure a higher level of accuracy and, therefore, analysis moving forward. Each task
requires time and effort; NNSC recommends LPD consider prioritizing those that would complement their
existing efforts

1. Compare NNSC findings with current Lexington Police Department intelligence.

LPD should vet NNSC’s findings regarding the total number of active groups in Lexington and build out the
existing intesocialligence on all but especially lesser-known groups. LPD should work to improve the

National Network for Safe Communities 21 (646) 557-4760 •

information on all groups identified during the group audit, keeping in mind that groups can vary in size
and level of organization. Additionally, LPD should continue to work on identifying the smaller subsets of
larger groups. GVI works in part by leveraging the informal social control that individuals have over one
another when they have social relationships; it is critical to identify the smallest meaningful subsets of
larger umbrella groups in order to harness those peer dynamics.

2. Routinize group audits.

The group information presented in this report is current as of January 2019, when the data was collected.
In NNSC’s experience, groups are fluid and naturally transition over time, with changes in membership,
activities, conflicts and alliances. It is therefore essential to develop a system to regularly review the
nature of active groups in Lexington at any given time. This process should focus not only on the groups
mentioned in this report, but also include any groups of two or more people who engage in crime and/or
violence together.

Law enforcement partners should meet for routine intelligence updates on at least a semiannual basis
and discuss all group characteristics reviewed during the group audit process as well as the emergence of
any new groups or the desistance of any groups and group members. This will ensure valid identification
of active groups, key individuals, and their social networks for the most effective use of various GVI
operational activities and communications mechanisms, including call-ins, custom notifications, and
group enforcement actions.

3. Begin planning and conducting regular shooting reviews to monitor violence and track group activity.

NNSC recommends that the law enforcement partners in Lexington work to formalize weekly violent
incident reviews, often referred to operationally as shooting reviews, focused on nonfatal shootings and
homicides that occur in the city. Shooting reviews are operational, frontline meetings where recent
shootings and homicides are discussed with law enforcement partners. They serve many purposes: to
gather the best intelligence on group involvement in serious violence, to identify the most violent groups,
to track the changing dynamics of groups, to share information among all operational partners, to devise
operational responses to violence, and to hold partners accountable to commitments that are made.
NNSC has found that instituting regular shooting reviews is the timeliest and most practical method for
continually assessing group and gang dynamics and their impact on violence. Weekly shooting reviews
will ensure this information is widely known across units and agencies, so it is also important to ensure
that all relevant law enforcement partners attend these regular reviews, to harness and share as much
information across agencies and units as possible. As previously mentioned, this process can enhance
partnerships among different units and agencies and provide regular opportunities for information
sharing. A couple steps should be taken before instituting shooting reviews to maximize productivity and
potential outcomes:

 NNSC has a network of sites that conduct regular shooting reviews and could facilitate a peer
exchange between Lexington partners and another site to observe this process firsthand and
discuss the process with law enforcement peers.
 Determine the best cohort to participate in shooting reviews in order to have the richest, most
up-to-date information about these incidents, those involved, and connections to active, violent

National Network for Safe Communities 22 (646) 557-4760 •

groups in the city. The participant list will understandably change and grow over time, but
Lexington should consider who best to include at the start which could be informed by discussion
with NNSC and other partners in our network.
 Lexington should begin by reviewing and coding any nonfatal shootings that occurred during the
week prior to the problem analysis and up through the present.

Weekly shooting reviews are also critical to keeping the promises that law enforcement typically makes
to group members and community stakeholders as part of GVI: that violence will be met with a swift and
certain response directed at the group as a whole rather than at individuals. This process helps ensure
that GMI incidents and the groups involved in these incidents are accurately identified in a timely manner.
This information also provides operational personnel with the intelligence necessary to determine the
strategic application of direct communication mechanisms, like custom notifications, to prevent
retaliatory shootings. Information shared during shooting reviews can be used to track the most violent
groups in the city based on their involvement in homicides and shootings which can help direct law
enforcement’s resources to those driving the violence.

4. Enhance existing intelligence gathering processes with social network analysis.

In addition to routinizing group audits and instituting shooting reviews, Lexington partners should work
with NNSC to determine how social network analysis (SNA) can augment understanding and responses to
group violence. For example, SNA can be deployed to better understand changing group dynamics and to
illuminate how gun violence “cascades” within social networks, particularly those involving group
members. SNA can also be used to supplement intelligence for especially lesser known groups by looking
at co-offending networks. NNSC could assist Lexington in discussing and even working with co-arrest data,
which can serve as the basis for these analyses.

Importantly, data analytics should never replace human intelligence or the expertise of those working in
our communities. Rather, SNA will help map particular social problems and provide a tool to reorganize
available data and expertise. The results must be read and vetted by those using them to inform strategic
decisions, but the networks cannot make decisions on their own.

5. Engage other partners.

NNSC believes that including other law enforcement agency partners in future group audits and incident
reviews may prove to be beneficial. These processes benefit from the inclusion of all frontline law
enforcement personnel who can share intelligence on the individuals and dynamics involved in particular
incidents. At every stage of the strategy – regular shooting reviews, routine group audits, strategic
enforcement action, and call-ins – NNSC has found that a comprehensive and committed partnership
ensures access to accurate intelligence and effective enforcement.

NNSC recommends expanding frontline law enforcement engagement to include participants from
narcotics, homicide, major crimes, and a wide representation of patrol officers, particularly those who
work nights in higher-violence areas. Additionally, the GVI strategy benefits from committed engagement
and collaboration from other law enforcement partners and agencies (e.g., probation, parole,
prosecution, task forces, etc.).

National Network for Safe Communities 23 (646) 557-4760 •

6. Take the following operational next steps to launch GVI.

A series of initial steps provides the foundation for a city to get to its first call-in, typically the event that
launches the strategy. The following is a checklist which outlines actions that the Lexington partnership
should complete in advance of holding the first call-in.

Identify a local, dedicated project manager. During the past 20 years of helping cities to launch GVI, it has
become clear that a full-time, local project manager is essential to sound, sustainable implementation.
The project manager oversees the effort, ensures fidelity to the model, and holds the strategy’s partners
accountable. The role and responsibilities of the project manager position include identifying both short-
term operational gaps, long-term sustainability issues, and solutions to those problems in real time.

Establish the governing board. The project manager helps identify principals from the law enforcement
agencies, social service providers, and community organizations committed to the strategy. This group
should meet regularly to take high-level responsibility for the GVI strategy; provide organizational
support; establish measures of accountability and progress; and use its influence to work through any

Identify local capacity for meaningful support and outreach. Before direct communication with groups can
begin, the GVI partnership must include a real, credible offer of support and resources for group members
who want to access help. The project manager and GVI governance board must identify this capacity and
coordinate logistics such that group members are able to access support and outreach resources that can
begin to conduct affirmative outreach to group members, help protect group members from harm, help
group members address trauma, help fulfill group members’ often-small but significant needs
(transportation, groceries, fines cleared off licenses, etc.) and help connect group members with
traditional services when they are ready.

Identify group members to attend the call-in. Law enforcement works together with the project manager
to select and notify group members to attend the call-in, typically through the channels of probation or
parole. The group members chosen are not necessarily those at highest risk but simply representatives of
each active, violent group in the city who can be compelled to attend the call-in by the conditions of their
probation or parole. Their role at the call-in is to act as messengers to carry the call-in message back to
other members of their groups. The goal is to represent every single active, violent group in the city at the

Plan and conduct the call-in. The project manager organizes the call-in by ensuring that the partnership
finds space, coordinates the selection and notification of group members, chooses speakers, holds
rehearsals, invites community members, and creates an effective security plan. All participants must be
clear about their roles and messaging, and all logistical details must be worked out in advance, in order to
ensure the smooth and sound execution of the call-in.

National Network for Safe Communities 24 (646) 557-4760 •

You might also like