Factors Influencing Victim and Witness Cooperation

Cooperation and Clearance: Victim Cooperation in Shooting Crimes
Jeffrey Ho
Public Policy Studies Honors Thesis
Duke University
Durham, North Carolina
1
Table of Contents
Abstract................................................................................................................................ 3
Introduction ......................................................................................................................... 4
Theoretical Framework........................................................................................................ 6
Crime Trends in the United States ..................................................................................................6
Factors Influencing Clearance Rates .............................................................................................7
Factors Influencing Victim and Witness Cooperation .....................................................................9
Methods.............................................................................................................................. 11
Data .................................................................................................................................... 13
Sources of Data ........................................................................................................................... 13
Clearance Model Variables .......................................................................................................... 14
Cooperation Model Variables ....................................................................................................... 18
Data Analysis .............................................................................................................................. 18
Results and Discussion ....................................................................................................... 19
Crime Patterns in Durham, NC .................................................................................................... 19
Cooperation Patterns in Investigations ......................................................................................... 22
Factors Influencing Clearance Rates ........................................................................................... 23
Factors Influencing Victim and Witness Cooperation ................................................................... 27
Reasons for Victim and Witness Noncooperation .......................................................................... 29
Investigator Strategies for Eliciting Cooperation........................................................................... 32
Conclusion and Recommendations..................................................................................... 34
References .......................................................................................................................... 42
Appendix A ........................................................................................................................ 44
Appendix B ........................................................................................................................ 45
2
Abstract
This thesis examined the factors affecting police clearance rates and citizen cooperation in
investigations, particularly as they related to shooting crimes in Durham, North Carolina. Existing
literature suggests that victim demographics and crime circumstances are limiting factors in police
investigations and may influence the noncooperation of victims and witnesses. The study further
explored these relationships through the use of statistical regression analyses of predictive factors
for police clearance and victim cooperation in an examination of investigative case data and a
qualitative analysis of police interview data. A secondary goal of the analysis was to provide
insight and recommendations for reducing noncooperation in police investigations. Based on an
examination of administrative data from the Durham Police Department records of shooting crimes
in 2015, a multivariate logistic regression revealed that crime circumstances were highly
associated with crime clearance. In addition, respondents who were male, racial and ethnic
minorities, or between the ages of 20 and 30 were more likely to be uncooperative in police
investigations. An analysis of semi-structured interviews with the Durham police department
investigators assigned to the cases helped provide potential explanations for the noncooperation,
including the presence of a pervasive culture of fear and the general mistrust in the police. I
consequently proposed potential recommendations for improving citizen cooperation, such as
through the use of community policing, providing credible relocation services for victims, and
expanding staffing and resource availability.
3
Introduction
In most jurisdictions across the United States, gun-related homicides are far more likely to
result in an arrest than gun-related aggravated assaults (Federal Bureau of Investigation 2014). In
many respects, homicides and aggravated assaults share similar properties – the only difference is
whether or not the victim survived the encounter. Thus, the disparity in criminal arrest rates
between the two violent crimes is one that appears to defy conventional wisdom about police
investigations and the criminal justice system. Presumably, the presence of a living victim, one
who can identify the offender and point the police in the direction of additional evidence, should
enhance, rather than hinder, investigative efforts.
One potential explanation lies in the victims themselves and their willingness to cooperate
with the police. The relationship between a law enforcement agency and its surrounding
community is integral to its success in identifying and arresting offenders, particularly in violent
crimes. Consequently, if a victim or potential witness chooses not to cooperate, police investigators
may struggle to gather the necessary evidence to arrest a suspect, thus reducing the efficacy of the
criminal justice system. Indeed, the resolution of criminal cases is an essential component of
American criminal justice. When law enforcement agencies fail to solve crimes, the justice system
weakens in its ability to incapacitate violators and deter future crimes. Thus, it is in the best interest
of police investigators to pursue the avenues necessary to identify and arrest criminal offenders,
particularly in violent crimes.
In this thesis, I examine the relationship that victim and witness cooperation have with
police arrest rates in Durham, North Carolina, particularly as they pertain to shooting crimes.
Through my analysis of police records and interviews, I elucidate whether cooperation, as well as
other factors, such as victim demographics and crime circumstances (Paré, Felson, and Ouimet
4
2007, Puckett and Lundman 2004), was predictive of success (the arrest of a suspect) in over 150
shooting police investigation in Durham during the 2015 investigative year. In addition, the thesis
examines victim cooperation with the police and explores methods of improving victim and
witness cooperation, such as through police tactics and initiatives.
Thus, the purpose of the thesis is to answer three primary questions. First, under what
circumstances are police investigations of shooting crimes likely to be successful? Secondly, what
factors are predictive of effective victim cooperation with police investigations? Lastly, given the
importance of cooperation, what measures can be taken by law enforcement to encourage greater
citizen cooperation with police investigations? Answering these questions could provide new
information on the factors most predictive of successful police investigations and offer insight for
policy recommendations to reform and improve citizen cooperation in the future.
The thesis begins by analyzing the existing literature on the factors affecting successful
crime investigations in the United States and the extent of citizen cooperation in investigations.
Subsequently, the thesis introduces data collected from the Durham Police Department, the
research methodology, and the interview process with the police investigators. The following
sections outline the regression models and important covariates examined in the models.
Accordingly, the thesis interprets the results of the models and the interviews to draw conclusions
about cooperation and arrest rates in Durham, North Carolina. My findings suggest that
cooperation, along with several victim demographic and crime circumstance variables, was a
strong predictor of arrest. Moreover, victim noncooperation levels appeared to vary with victim
demographics, and two potential explanations for the relationship are the pervasiveness of fear and
skepticism about the police at the community level. The conclusion discusses the implications of
5
the thesis findings and provides policy recommendations to improve citizen cooperation and
overall police investigations.
Theoretical Framework
Crime Trends in the United States
In the past twenty years, violent crime rates have declined by nearly half in the United
States, from a rate of 713.6 per 100,000 in 1994 to 365.5 per 100,000 in 2014 (Federal Bureau of
Investigation 2014). However, gun-related offenses continue to make up a consistent proportion
of violent crime occurrences, constituting approximately 68% of all homicides and 23% of
aggravated assaults. Similarly, while crime rates have fallen in the United States in the past two
decades, the rate at which crimes are solved has remained fairly constant at 45% (Federal Bureau
of Investigation 2014).
Crime clearance is traditionally regarded as the best measure of success for a criminal
investigation. Criminal offenses can be cleared in one of two manners. First, a crime can be cleared
by arrest if a law enforcement agency arrests, charges, and turns over an individual to the court for
the prosecution of an offense (Federal Bureau of Investigation 2014). Secondly, a crime can be
cleared by exceptional means if certain factors inhibit the ability of law enforcement to arrest and
charge a suspect. These circumstances include the death of the offender, if the offender has fled to
another jurisdiction and cannot be extradited, or if the victim refuses to cooperate with the police
investigation (Federal Bureau of Investigation 2014).
Despite extensive advances in investigative strategies and technology, the national
clearance rate of violent crime has remained largely constant over the past twenty years at
approximately 45% (Federal Bureau of Investigation 2014). Amongst homicides, the clearance
rate has persisted at approximately 65%, and amongst aggravated assaults, the clearance rate has
6
stayed at 56% (Federal Bureau of Investigation 2014). Such stability suggests that either clearance
rates are independent of developments in investigative techniques and forensics technology or
other counteracting factors may exist in contemporary criminal investigations. Indeed, it is of
particular interest to identify the factors most associated with clearance rates and propose potential
policy recommendations that remediate the limitations to police investigations.
Factors Influencing Clearance Rates
Currently, several conflicting viewpoints provide explanations for the factors affecting
crime clearance. In a seminal study, Black (1970) proposed that victims are the prime movers of
the American legal system. In essence, police investigations are driven primarily by the behavior
and demographics of the victim; the victims most valued by society are those who draw the police’s
attention. Thus, Black’s theory of law (1976) suggests that the criminal justice system exists to
protect privileged citizens, and police investigators exercise discretion in their investigations to
favor victims in positions of power. Consequently, Black’s theory would imply that low crime
clearance rates are associated with victims in weaker societal positions: the young, women, and
racial and ethnic minorities.
Many contemporary scholars reject Black’s explanation in favor of others. Several scholars
found no association between victim characteristics and clearance (Litwin 2004, Puckett and
Lundman 2004). Other scholars, in fact, found evidence contrary to Black’s hypothesis, that crimes
with female victims are more likely to be cleared than crimes with male victims (Roberts 2008).
Instead, the circumstances of a crime may be the most indicative of its clearance. More serious
offenses are more likely to be cleared, such as homicides and nonlethal offenses with severe
injuries, though the explanations for such patterns are not well established (Paré, Felson, and
Ouimet 2007, Roberts 2008).
7
Other crime details may also be predictive of eventual crime clearance, namely time and
location. Crime patterns vary by time of day (Pittman and Handy 1964), and thus, the time of
occurrence for a crime, particularly when contrasting daylight and night hours, may offer
predictive value to clearance because police investigators may have access to more readily
available resources during daylight hours. Some scholars, though, have suggested that time of day
is not significantly related to homicide clearance (Roberts 2008). Similarly, the location of a crime
may relate to clearance. Whether a crime occurred in a public or private area could affect the
accessibility of evidence and witnesses in police investigations (Litwin 2004).
Hence, a more complex view of police clearance, one which incorporates victim
demographics as well as crime circumstances, may provide the best explanation for clearance.
However, more recent literature suggests that exogenous, predetermined factors may not be the
sole predictors of crime clearance. In Boston, increased police activity and intervention was found
to be related to improved homicide clearance rates (Braga and Dusseault 2016). Police reforms,
particularly if they target factors associated with clearance, may help improve clearance rates in a
jurisdiction.
Indeed, recently, Braga and Dusseault (2016) found significant differences in Boston
clearance rates between pre-intervention and post-intervention periods. Though they were unable
to specifically identify the intervention measures that contributed to the increase in clearance, the
results of their study suggest that some combination of increased police resources, more officers,
easier access to evidence analysis, and more contact with witnesses may contribute to higher
clearance rates. Thus, identifying the specific factors may prove useful in prescribing
recommendations to the criminal justice system.
8
One such potential area of explanation is the impact of victim and witness cooperation in
police investigations. For instance, the disparities in clearance rates across victim demographics
may be better explained by police devaluation rather than victim devaluation (Keel, Jarvis and
Muirhead 2009). Certain groups, such as those that are less advantaged, may have a greater distrust
of the criminal justice system. Consequently, a general loss of faith in the police may subsequently
hinder police investigative capacities (Keel, Jarvis, and Muirhead 2009). More specifically,
Puckett and Lundman (2004) propose that homicides in predominantly black communities are
solved at lower rates than those in predominantly white communities because of a general
skepticism of police, and subsequently, investigators may struggle in garnering the necessary
citizen cooperation to clear a case.
Factors Influencing Victim and Witness Cooperation
Citizen cooperation plays a crucial role in the functioning of the criminal justice system.
From the reporting of crimes to the discovery of evidence to testifying in the courtroom, victims
and witnesses are essential to the eventual clearance of a crime (Cook and Fischer 1976). Indeed,
the relationship between cooperation and clearance is well documented. According to Cook and
Fischer (1976), witness cooperation plays a major role in the vast majority of robbery clearances,
and in homicides, witness cooperation is essential to compiling the necessary evidence to arrest a
suspect (Armstrong, Plecas, and Cohen 2013).
Whereas the factors influencing clearance are well-documented, less research has been
conducted on the predictive factors of effective victim and witness cooperation. Cooperation may
vary based on the demographics of the victim or witness, and those in a more disadvantaged
position in society may be less likely to assist in a police investigation due to an inherent distrust
in the criminal justice system (Litwin 2004, Riedel 1999). Similarly, the level of witness
9
cooperation is significantly less in gang-related homicides, potentially due to the fear of retaliation
(Armstrong, Plecas, and Cohen 2013).
Indeed, perhaps the greatest barrier to victim and witness cooperation is victim
intimidation. Fear of retaliation in certain communities is a major deterrent to cooperation with
law enforcement officials (Riedel 1999). In particular, victim and witness intimidation appears to
be especially pervasive amongst organized crime and domestic violence cases (Healey 1995). The
prevalence of witness intimidation suggests that one solution to improving victim and witness
cooperation may be the expansion of police protection programs, particularly in cases where police
investigators suspect that an individual may be the victim of a gang-related or domestic crime.
Healey (1995) proposes that emergency relocation for witnesses, improved courtroom security,
segregation in correctional facilities, and greater community outreach may all be effective means
of eliciting greater victim and witness cooperation.
Other police methods and strategies may also be associated with greater citizen
cooperation. According to Dawson and Dinovitzer (2001), in domestic violence cases, the two
most important predictors of victim cooperation are the videotaping of testimony and the meetings
between victims and their assistance workers. The findings suggest that though traditional
approaches to victim and witness cooperation include more aggressive tactics such as threats of
obstruction of justice, forming a closer personal connection with the victim may be more effective
at encouraging cooperation in domestic violence cases. Such a trend may extend to other violent
crimes as well.
Lastly, the prospect of monetary incentives may prove a useful tool in prompting greater
citizen cooperation. Across the United States, many victim compensation programs, which provide
monetary relief to victims of crimes in exchange for their cooperation, exist. However, most
10
victims are either poorly informed or unaware of the services (Elias 1984). Since most victim
compensation programs are contingent on the full cooperation of the victim with the police, such
as in North Carolina, police investigators could presumably improve cooperation rates through
fully informing victims about the services and potentially leveraging the monetary incentive.
Similarly, the expansion of a CrimeStoppers-type program may be utilized to improve the
willingness of witnesses to come forward through anonymous tips. CrimeStoppers is an
increasingly popular program whereby anonymous citizens provide information to the police in
exchange for monetary rewards. The true impact of the program on community crime levels
remains largely unknown, though some scholars, such as Elias (1984), suggest that the true impact
of compensation programs may be more symbolic than tangible. In addition, though some
prospective witnesses may be motivated by the monetary incentive of CrimeStoppers, the program
may also have the effect of undercutting intrinsic motivations in witnesses who would otherwise
willingly cooperate with the police.
Ultimately, a combination of factors may determine victim and witness cooperation. On
the individual level, a variety of demographic features, including race, ethnicity, and sex, may
predict the individual’s eventual willingness to cooperate. Nonetheless, several investigative
strategies may also assist in greater cooperation. For instance, improved police protection and
monetary compensation, factors potentially within the control of a law enforcement official, may
also prove valuable predictors of eventual citizen cooperation.
Methods
For the analysis, I utilized data regarding all criminal gunshot cases in Durham, NC during
the 2015 calendar year – a total of 169 cases with 205 victims, of whom 24 died. The first models
examined the factors that predict the odds of a police clearance by arrest, and the second models
11
examined the factors associated with the odds of cooperation. In addition, I conducted a qualitative
analysis of data collected from fifteen interviews with police investigators in order to identify
investigator experiences and beliefs about citizen cooperation and potential means of reducing
noncooperation.
The data for the study come from a compilation of Durham Police Department records,
investigative case notes, and data coded from semi-structured interviews with police investigators.
The data includes extensive information about each gunshot case, including case details,
demographic information about the victim, information about the police investigation, and victim
and witness cooperation details.
The research setting for the study is Durham, NC, and thus, the results of the study are only
representative of the city. Nonetheless, the crime profile of Durham indicates that results from the
study may offer broader insight into police clearance and victim cooperation. Though different
from the crime profile of the United States as a whole, Durham bears a remarkably similar crime
profile when compared to other large cities in the United States, defined as cities with populations
greater than 250,000 (Appendix A). Amongst large cities, the violent crime rate in 2014 was 730.1
per 100,000 as compared to 724.4 per 100,000 in Durham (Federal Bureau of Investigation 2014).
More specifically, in Durham, aggravated assaults occurred at a rate of 434.9 per 100,000 while
murder and nonnegligent manslaughter occurred at a rate of 8.4 per 100,000. The average amongst
large cities in the same time period was 412.1 per 100,000 and 9.3 per 100,000, respectively
(Federal Bureau of Investigation 2014). Similarly, in 2014, the clearance rate for violent crimes in
Durham was approximately 44%, compared to 41% amongst large cities (Federal Bureau of
Investigation 2014, Durham Police Department 2015).
12
I selected the year 2015 for two primary reasons. First, the 2015 criminal cases were the
most recently available cases at the time of the study, thus providing the greatest relevance to any
results drawn from my analysis and easier recollection when interviewing investigators about
specific case details. Secondly, the 2015 investigative year experienced an unprecedented increase
in violent crimes in Durham. When compared to 2014, homicides increased by 68%, and
aggravated assaults increased by 23% (Durham Police Department 2015). Given the striking
elevation in crime, analysis of the 2015 investigative year could offer insight into potential means
of remedying and addressing this concern.
Data
Sources of Data
The majority of the data was compiled from two separate databases in the Durham Police
Department. Project Safe Neighborhood, a Durham initiative to partner law enforcement with
residents to reduce gun violence, keeps a consistent record of gunshot crimes throughout Durham,
defined as incidences where at least one individual was the victim of a shooting crime. The Project
Safe Neighborhood database was then cross-referenced with the digitized Durham Police
Department database on all criminal cases, which included more detailed information about the
cases, including the case notes. Through matching of the two sources, I was able to construct a
dataset of all 2015 shooting crimes and the details about those crimes. The final sample consisted
of 169 shooting crimes and 205 shooting victims.
In addition, another researcher, Sara Shilling, conducted semi-structured interviews with
Durham Police Department investigators that handle aggravated assault and homicide cases.
Overall, she conducted thirteen interviews with respondent representatives from each district in
Durham as well as the homicide unit. She conducted eight interviews with aggravated assault
13
investigators and five interviews with homicide investigators. Prior to the interviews, she
researched one or two cases from 2015 in which the respondent served as the lead investigator. In
the first section of the interview, she explored the specifics of the cases and asked about the victim
and witnesses, their extent of cooperation, and any attempts made by the investigator to reach out
or elicit further cooperation (Appendix B). The first section of the interview served two purposes.
First, it helped identify specific and tangible actions taken by the investigator to encourage
cooperation. Secondly, it helped render the interviewee to be more comfortable in the interview
by helping to demonstrate familiarity with the topic and the investigator’s case history.
In the second section of the interview, she asked the police investigators general questions
about their perceptions of citizen cooperation and the reasons for noncooperation. In addition, she
explored methods for working with victims and witnesses to elicit cooperation and discussed
recommendations for improving cooperation (Appendix B). Subsequently, with the notes taken
from the semi-structured interviews, I analyzed the larger themes and ideas in the investigator
responses to questioning. Specifically, I identified common responses to the questions in order to
synthesize the different investigator opinions on the obstacles to cooperation and the potential
solutions. I coded similar ideas to help determine points of agreement and disagreement between
the different investigators. Examining the investigator opinions provided insight into the factors
affecting clearance and cooperation that may not necessarily have been evident in a purely
quantitative regression analysis.
Clearance Model Variables
For the purpose of the statistical analysis, I included eight explanatory variables that could
serve as potential factors that influence police clearance, either from evidence in existing literature
or trends illustrated in the Durham Police Department case files.
14
Three demographic characteristics about the shooting victim were measured:
race/ethnicity, sex, and age. Race and ethnicity were coded as a categorical variable with white,
non-Hispanic as the baseline and dichotomous variables (0 or 1) for black, non-Hispanic and
Hispanic. Similarly, sex was coded as a categorical variable (1=female). Age was modified to be
a categorical variable with three categories: under 20 years (omitted category), between 20-30
years, and over 30 years of age. In cases with multiple victims, the demographics of the oldest
victim were used. I chose the oldest victim because presumably, the oldest victim is the victim that
usually interacts with police investigators the most and may have the greatest influence over the
behavior of the other victims.
In addition to the victim’s demographic characteristics, I also examined the degree of
victim cooperation. For the variable of victim cooperation, which was not directly included in the
Durham Police Department crime records, I instead examined the case notes and coded my own
variable. The victim cooperation variable required some subjective judgment, but I applied the
same judgement to all the case files. I coded the variable of victim cooperation as a nominal
variable with four categories: active, passive, uncooperative, and incapable. Active cooperation
occurred when a victim in a given shooting crime worked dynamically and cooperatively with the
police investigator to identify or arrest a suspect, such as providing the name of the suspect,
providing a comprehensive description of the suspect individual or vehicle, or seeking out the
investigator for outreach and questioning. In contrast, passive cooperation occurred when the
victim complied with investigator requests but was not actively involved in the investigation.
Passive victims generally provided initial statements to investigators but were not able to provide
a description of the suspect or otherwise provide useful information to the investigator.
Noncooperation (uncooperative victims) indicated when the victim directly impeded or prevented
15
the continuation of the investigation, such as directly lying to the police, refusing to make a
statement or otherwise disappearing from the investigation, or specifically telling the police that
the victim had no intention of pressing charges. Lastly, incapable victims indicated crimes in which
a shooting victim was not capable of cooperating in a police investigation. Incapable victims
existed in homicides (the victim was killed) and severe aggravated assaults when the victim was
injured beyond the point of being responsive in an investigation. In the latter instance, the victim
was either rendered incapable permanently, such as a comatose patient, or the investigation
concluded before the victim could become responsive.
For the clearance model, I initially ran the regression without the cooperation variables to
analyze the model with purely the exogenous, predetermined variables. Afterwards, I included the
cooperation variables to examine the impact of cooperation on clearance by arrest. Noncooperation
served as the baseline, and I included dichotomous variables for each of the other two types of
cooperation. I did not include an indicator for incapable victims because the incapable victim
variable was highly correlated with a homicide indicator, and only one variable would need to be
included.
In addition, four case circumstance variables about the crime were included in the analysis.
Given the marked disparity between the clearance rates of aggravated assaults, homicides, and
robberies, the type of crime was also included in the regression model, with aggravated assaults
serving as the baseline. Similarly, the model controlled for the time of day of the crime. Time of
day was simplified into three categories: morning, afternoon, and night (omitted category). In
regards to crime location, three categories were included in the analysis: inside and non-vehicular,
outside and non-vehicular, and vehicular (omitted category). Inside and outside crimes referred to
any crimes in which the victim was either indoors or outdoors at the time of the shooting,
16
respectively. Vehicular crimes included any shooting crimes in which the victim was in a car, the
offender was in a car, or both were in a car. In particular, vehicular crimes included drive-by
shootings, crimes in which the perpetrator fired a gun from a vehicle and fled the scene of the
crime. I categorized such crimes as vehicular, even if they could otherwise be categorized as either
inside or outside.
Lastly, after analyzing the investigator case files and coding the aforementioned variables,
I noticed another variable that could potentially play a role in crime clearance: call origin. In the
shooting crimes that I examined, the call to the police department either originated from the crime
scene or, less frequently, from the Duke University hospital. Crimes with calls that originated from
the hospital could be related to the cooperation of the victim, who chose to visit the hospital prior
to asking someone to call the police, and the capacity of the police to investigate the crime. As
such, I also included a variable indicating the call origin, with the call originating from the crime
scene serving as the omitted category.
As the purpose of the model was to identify underlying factors associated with crime
clearance, clearance served as the most valuable response variable of interest. However, cases
cleared by “exceptional means” were not included because such instances did not ultimately lead
to an arrest. Thus, only clearances through arrest were included in the indicator variable, and
clearance was coded as a dichotomous variable, with a “1” indicating that the case was cleared
through an arrest.
In total, I ran three different logistic regression models with clearance by arrest as the
response variable. The first model examined the exogenous factors (all variables excluding the
cooperation variables) associated with clearance by arrest in all shooting crimes. The second model
examined the same sample of shooting crimes but included the cooperation variables. Lastly, the
17
third model examined all the variables, including cooperation, but only examined cases with a
capable victim.
Cooperation Model Variables
The cooperation models examined the relationship between the same predictive variables,
apart from victim cooperation, and the response variables of victim cooperation in aggravated
assaults. Only aggravated assaults and robberies were examined in the cooperation models because
the variable of victim cooperation was nonexistent in homicides. Similarly, the cooperation models
also excluded aggravated assaults with incapable victims.
I examined two different cooperation models. In the first model, the response variable was
noncooperation, a variable that was coded as “1” if the victim was perceived as directly impeding
the police investigation, and the default was a combination of the variables for active and passive
cooperation. In the second model, the response variable was active cooperation, a variable that was
coded as “1” if the victim actively assisted in the police investigation, and the default was a
combination of the variables for noncooperation and passive cooperation. The two different
models were examined because the factors associated with active cooperation may differ from
those that are related to noncooperation.
Data Analysis
The logistic regressions were fit to the data in order to illuminate the relationships between
the explanatory variables and the response variables of clearance and cooperation. For the
regression models, dichotomous variables were utilized for all explanatory variables. Given the
default values for these variables, the baseline group for the clearance model consisted of shooting
incidents with a white, non-Hispanic, male victim under the age of twenty who was uncooperative,
was the victim of an aggravated assault vehicular shooting at night, and the call to the police was
18
made from the crime scene. The baseline group for the cooperation model was the same with the
exception that the victim’s degree of cooperation was not indicated.
The equations below present some of the logistic regression models for the analysis.
β1,…,βn serve as the coefficients for each of the explanatory indicator variables identified. Both
the magnitudes and the significance levels of the coefficients provide insight into the relationships
between the explanatory variables and the response variables.
𝑃(𝑐𝑙𝑒𝑎𝑟𝑎𝑛𝑐𝑒)
log (
)
1 − 𝑃(𝑐𝑙𝑒𝑎𝑟𝑎𝑛𝑐𝑒)
= 𝛽0 + 𝛽1 𝑏𝑙𝑎𝑐𝑘. 𝑛𝑜𝑛ℎ𝑖𝑠𝑝𝑎𝑛𝑖𝑐 + 𝛽2 ℎ𝑖𝑠𝑝𝑎𝑛𝑖𝑐 + 𝛽3 𝑓𝑒𝑚𝑎𝑙𝑒 + 𝛽4 𝑎𝑔𝑒. 20.30
+ 𝛽5 𝑎𝑔𝑒. 𝑜𝑣𝑒𝑟. 30 + 𝛽6 𝑝𝑎𝑠𝑠𝑖𝑣𝑒 + 𝛽7 𝑎𝑐𝑡𝑖𝑣𝑒 + 𝛽8 ℎ𝑜𝑚𝑖𝑐𝑖𝑑𝑒 + 𝛽9 𝑟𝑜𝑏𝑏𝑒𝑟𝑦
+ 𝛽10 𝑚𝑜𝑟𝑛𝑖𝑛𝑔 + 𝛽11 𝑎𝑓𝑡𝑒𝑟𝑛𝑜𝑜𝑛 + 𝛽12 𝑜𝑢𝑡𝑠𝑖𝑑𝑒 + 𝛽13 𝑖𝑛𝑠𝑖𝑑𝑒
+ 𝛽14 𝑐𝑎𝑙𝑙. 𝑓𝑟𝑜𝑚. ℎ𝑜𝑠𝑝𝑖𝑡𝑎𝑙 + 
𝑃(𝑛𝑜𝑛𝑐𝑜𝑜𝑝𝑒𝑟𝑎𝑡𝑖𝑜𝑛)
log (
)
1 − 𝑃(𝑛𝑜𝑛𝑐𝑜𝑜𝑝𝑒𝑟𝑎𝑡𝑖𝑜𝑛)
= 𝛽0 + 𝛽1 𝑏𝑙𝑎𝑐𝑘. 𝑛𝑜𝑛ℎ𝑖𝑠𝑝𝑎𝑛𝑖𝑐 + 𝛽2 ℎ𝑖𝑠𝑝𝑎𝑛𝑖𝑐 + 𝛽3 𝑓𝑒𝑚𝑎𝑙𝑒 + 𝛽4 𝑎𝑔𝑒. 20.30
+ 𝛽5 𝑎𝑔𝑒. 𝑜𝑣𝑒𝑟. 30 + 𝛽6 𝑟𝑜𝑏𝑏𝑒𝑟𝑦 + 𝛽7 𝑚𝑜𝑟𝑛𝑖𝑛𝑔 + 𝛽8 𝑎𝑓𝑡𝑒𝑟𝑛𝑜𝑜𝑛 + 𝛽9 𝑜𝑢𝑡𝑠𝑖𝑑𝑒
+ 𝛽10 𝑖𝑛𝑠𝑖𝑑𝑒 + 𝛽11 𝑐𝑎𝑙𝑙. 𝑓𝑟𝑜𝑚. ℎ𝑜𝑠𝑝𝑖𝑡𝑎𝑙 + 
Results and Discussion
Crime Patterns in Durham, NC
As illustrated in Table 1, in 2015, over three-quarters of the shooting crimes in Durham,
NC were aggravated assaults, and homicides only composed 13.6%. Out of the 169 gunshot cases,
only 21.9% were cleared (15.4% were cleared by arrest), and the remaining 78.1% of cases were
either designated as inactive or remain as active, open cases. The 21.9% clearance rate for shooting
19
crimes is substantially lower than the 44.3% for all violent crimes in Durham, NC, thus suggesting
that shooting crimes are particularly difficult to investigate and clear compared to other types of
violent crimes (Durham Police Department 2015). The particularly low clearance rate for shooting
crimes is especially alarming and may be related to the increase in shooting crimes in Durham
from 2014 to 2015. Lower clearance rates weaken the criminal justice system because they can
portray the perception that law enforcement is incapable of apprehending perpetrators, thus
reducing the deterrence for future crimes.
The demographics of the shooting victims also provided additional insight into the
individuals that are often the victims of shootings. For instance, as seen in Table 2, a vast majority
(87.8%) of shooting crime victims were black and non-Hispanic. Similarly, 86.8% of victims were
male. The type of cooperation across victims was fairly distributed between the four cooperation
categories. Passively cooperative victims were the most common (39.5%), though active and
uncooperative victims occurred at similar rates (21.5% and 23.9%, respectively).
Though not included in the clearance by arrest variable in the regression models, the
shooting crimes cleared by exceptional means nonetheless offer interesting insight, particularly as
they relate to victim cooperation. Though only eleven cases were cleared by exceptional means,
the majority of exceptionally cleared cases were designated as such because of the absence of
victim cooperation. In these cases, police investigators confidently identified the offender and
gathered sufficient evidence to bring the case to trial, but because of a victim’s noncooperation,
the district attorney decided against bringing the case to trial. For instance, in one such case, a
victim’s mother positively identified the offender and provided the police investigators with
sufficient evidence to build a case. However, when interrogated, the victim denied the statements
20
of his mother and refused to cooperate with the investigation, and thus, the case was designated as
cleared by exceptional means.
As illustrated in Table 3, the cases that were cleared by arrest and those that were not
differed on the majority of the variables. For instance, cases with white, non-Hispanic victims were
more likely to be cleared by arrest than cases with black, non-Hispanic victims or Hispanic victims.
Though sex did not appear to be related to clearance by arrest, clearance by arrest rates differed
across age groups. Cases with older victims had a higher rate of clearance by arrest than cases with
younger victims.
Furthermore, consistent with Paré, Felson, and Ouimet (2007) and Roberts (2008), crimes
regarded as more serious by the police had higher arrest rates than crimes regarded as less serious.
Homicides had the highest clearance rate by arrest (52.2%), followed by aggravated assaults
(10.0%) and robberies (6.3%). Time of day was also a differentiating factor, with gunshot crimes
that occurred later in the day less likely to result in a clearance by arrest than those that occurred
earlier.
Perhaps most striking about the shooting crimes in Durham was the prevalence of vehicular
shootings. Indeed, 57.4% of all the victims’ cases were automobile related (Table 1), and of those,
nearly all the crimes were drive-by shootings. The drive-by shooting with a witness who could not
identify the shooter was a particularly recurring situation that could be found in many of the case
files. As illustrated in Table 3, the consequences of an unidentified shooter were quite evident,
with vehicular crimes associated with much lower clearance rates.
As hypothesized, clearance by arrest differed drastically depending on the origin of the
police call. Calls from the hospital had a clearance by arrest rate of 0%, while calls from crime
scenes had clearance rates of 19.0%.
21
Cooperation Patterns in Investigations
As with clearance by arrest, victim demographics also appeared to be highly associated
with cooperation rates (Table 4). Ethnic minorities, racial minorities, and male victims were more
likely to be uncooperative in police investigations. Likewise, victims over the age of 30 were the
least likely to be uncooperative and the most likely to be actively cooperative in an investigation.
Active cooperation rates were also generally higher amongst robberies when compared to
aggravated assaults. Whereas 62.5% of robbery victims were actively cooperative, only 21.5% of
the aggravated assault victims were likely to be as helpful in a police investigation.
Crime circumstance variables also appeared to be closely related to cooperation rates.
Victims of vehicular crimes were slightly more likely to be uncooperative and substantially less
likely to be actively cooperative. Similarly, victims of crimes in which the call to the police was
made from the hospital were only actively cooperative in 14.3% of cases and uncooperative in
40% of all cases.
Victim cooperation also shared a strong association with clearance by arrest rates. As
expected, clearance rates increased with greater cooperation (Table 5). Active cooperation had the
highest clearance rate (16.2%) while noncooperation had the lowest clearance rate (2.3%).
However, the disparity when comparing cases with capable victims and incapable victims is
especially concerning. The clearance rate for shooting crimes with incapable victims was 48.2%,
substantially higher than the clearance rate of crimes with active victims. The difference suggests
that crimes with any capable victim, even an actively cooperative victim, were significantly less
likely to be cleared by arrest.
22
Factors Influencing Clearance Rates
The results of the multivariate logistic regressions for clearance by arrest are presented in
Table 6. In general, the analysis provides some evidence for the hypothesis that clearance rates
differ based on victim demographics but does strongly suggest that crime circumstances are
predictive of crime clearance by arrest.
Consistent with Black’s theory of law (1976), when controlling for other variables, crimes
with black and Hispanic victims appear to be cleared at a lesser rate than those with white, nonHispanic victims, particularly when the victim was capable. Specifically, compared to a baseline
of white, non-Hispanic victims, the odds ratio of a crime being cleared was 0.82 for Hispanic
victims in crimes with capable victims (p=0.20), though the disparity was not as evident in all
shooting crimes.
According to the regression model, neither sex nor age group appeared to be a particularly
predictive factor in crime clearance. When controlling for other variables, the rates of clearance
by arrest did not differ from the baseline when indicators for sex and age group were included in
the model.
However, the logistic regression model provides evidence to support the hypothesis that
crime circumstances may be predictive of eventual clearance. The crime type, time of day, crime
location, and call origin were all fairly predictive of a crime’s eventual clearance by arrest.
The time of a crime appears to be related to its eventual clearance. Shooting crimes that
occurred in the afternoons and in the evenings were solved at fairly similar rates, but crimes in the
mornings were substantially more likely to be solved when controlling for all other confounding
factors. The results suggest that crime scenes, witnesses, and evidence may be more easily
23
investigated during early hours. Time-sensitive police resources may also be more readily
assessable during the day.
Likewise, crime location was also predictive of clearance by arrest. Vehicular crimes were
the most difficult to solve, with both crimes that occurred inside and outside cleared at higher rates.
Amongst crimes with capable victims, the odds ratio of a crime being cleared, when compared to
a vehicular crime, was 1.15 and 1.09 for outside and inside crimes, respectively (p=0.03 and
p=0.22). The lower clearance rates for vehicular crimes, which were predominantly drive-by
shootings, is not especially surprising. In most such cases, the victim was unable to identify the
shooter. The reasons for the lower rate of shooter identification are less clear, but two primary
reasons could provide an explanation. First, the nature of drive-by, vehicular shootings lends it to
be more difficult to positively identify a perpetrator. In such cases, it may be difficult for a victim
to even see, let along remember, a shooter because of how quickly the offender can flee and remain
concealed within a car. Secondly, drive-by shootings are more characteristic of gang related
crimes, whereas crimes of passion and other common motivations for crimes are rarely manifested
as vehicular shootings. Thus, since gang-related crimes are also commonly associated with fear
and victim intimidation, even if victims of vehicular shootings could identify their shooters, they
may choose to avoid sharing the details out of fear of gang violence and retaliation.
Consistent with the trends in the descriptive crime statistics, the origin of the police call
was also a strong predictor of eventual clearance. Shooting crimes in which the call to the police
was placed from the Duke Hospital were solved at a significantly lesser rate than crimes in which
the call was placed from the crime scene. Two likely reasons provide explanations for the
differences. First, when the call to the police does not occur from the crime scene, the investigators
may be substantially inhibited in their capacity to investigate the crime because they must first
24
identify the crime scene, and the evidence at the crime scene may already be compromised.
Secondly, in such crimes, the victims chose to visit the hospital to be treated for their gunshot
wounds prior to asking someone to call the police. Such behavior could be suggestive of a lack of
trust in the police by either the victims or their acquaintances and thus an unwillingness to later
cooperate in the police investigation.
Lastly, consistent with extensive existing literature (Paré, Felson, and Ouimet 2007,
Roberts 2008), when controlling for other potential factors, homicides did appear to be solved at a
much greater rate than aggravated assaults. Indeed, the odds ratio of a homicide being cleared was
1.38 times the odds of a similar aggravated assault being cleared (p<0.01). The results of the model
suggest that even when the examined covariates were taken into consideration, homicides still
have a significantly higher rate of clearance than aggravated assaults. Thus, various characteristics
about homicide investigations, such as the additional resources, the more experienced
investigators, or the added emotional motivation to solve a killing, may all play a role in
encouraging higher homicide clearance rates.
Another such explanation is, as hypothesized, the degree of a victim’s cooperation. Higher
degrees of cooperation appeared to increase the rate of clearance by arrest when compared to an
uncooperative victim. In particular, active cooperation was especially predictive of higher crime
clearance, and the odds ratio of clearance for crimes with victims that exhibited active cooperation
was 1.13 times the odds ratio of similar crimes with uncooperative victims (p=0.10). Thus, the
models strongly suggest that victim cooperation and noncooperation are some of the primary
explanations for the lower overall clearance rates in aggravated assaults when compared to
homicides.
25
Many of the interviewed investigators also echoed the idea that the influence of citizen
cooperation plays a role in the differences between homicide and aggravated assault arrest rates.
Over half of the respondents explained that witnesses were generally more cooperative in homicide
investigations than in aggravated assault investigations for two reasons. First, due to the extended
timeline of homicide investigations, homicide investigators have more time to develop
relationships with witnesses. Homicide investigators can visit witnesses multiple times and spend
time building rapport with them. Secondly, the death of an individual plays a large role in the
willingness of a prospective witness to share information with the police. When an individual is
the victim of an aggravated assault shooting, community members do not feel the need to come
forward and get closure, especially since gunshot wounds can even be seen as a status symbol in
the community. However, death is a more permanent loss, and community members often feel
stronger emotions as a result of a homicide and want to get justice. Indeed, family members are
typically more involved in homicide investigations than in aggravated assault investigations.
In addition, several interview subjects suggested that in homicides, the death of a victim
actually allows for an easier investigation because the victim is not able to be uncooperative in the
investigation. Homicide investigations grant investigators access to victims and details about their
lives, but such information may not be accessible in aggravated assaults when a victim chooses
not to cooperate.
Respondents also pointed to general resources and the caseload of homicide investigators,
compared to aggravated assault investigators, as a potential reason for the different clearance rates.
For instance, the homicide unit has substantially more resources than the violent crime units, such
as more contact with the district attorneys, longer time to work on investigations, more lab tests,
more assistance from other investigators (all homicide investigators work collaborative on an
26
investigation whereas in an aggravated assault, only one investigator may be at the crime scene),
and an overall lighter caseload. In addition to facilitating an easier investigation, several
investigators also suggested that greater resources and a lighter workload contributes to spending
more time with potential witnesses and thus greater citizen cooperation in investigations.
Factors Influencing Victim and Witness Cooperation
The second logistic regression model analyzes the association between the explanatory
variables and the response variable of victim cooperation, identified as one of the strongest
predictors of crime clearance by arrest (Table 7). Based on the analysis, victim demographics (and
one crime circumstance variable) appear to be the primary predictors of noncooperation with the
police, but crime circumstances appear to be predictors of active cooperation.
All four crime circumstance variables appear to be significant predictors of active
cooperation, whereas all three victim demographics were not strongly associated with active
cooperation. For instance, robberies were more likely to have actively cooperative victims. The
pattern may be due to the nature of robbery crimes. In most robberies, victims are not acquainted
with the perpetrator, and thus, they may have fewer reservations about assisting in the police
investigations. Moreover, victims of robberies may also have the added incentive of attempting to
reobtain their lost property. Likewise, crimes that occurred in the morning or afternoon were more
likely to have cooperative victims than crimes in the evening, and crimes that occurred outside or
inside were more likely to have actively cooperative victims than vehicular crimes. Both evening
crimes and vehicular crimes present circumstances that may prevent active cooperation. Victims
in such crimes may be limited in their capacity to assist the police in such crimes because of an
inability to see or identify the offender.
27
Ultimately, the results of the active cooperation regression model suggest that the
likelihood of active victim cooperation depends not on who the victim is but on the circumstances
of the crime. Thus, implementing police tactics to mitigate the underlying factors that contribute
to reduced active cooperation may assist in improving cooperation and thus overall clearance rates.
However, the associations appear to be the opposite for the noncooperation model. Of the
crime circumstance variables, only call origin was a strong predictor of eventual victim
noncooperation. The odds ratio of noncooperation for crimes in which the call to the police was
made from the Duke Hospital when compared to crimes in which the call was made from the crime
scene was 1.15 (p=0.11). A qualitative analysis of the investigator case files presents some
potential insight into the difference in cooperation. In all the cases in which the police call was
made from the hospital, the victim arrived at the hospital via car without anybody having called
the police. The police were later called by the hospital due to mandatory reporting, but in many
such cases, presumably, the victim or their acquaintances chose not to call the police because they
did not want police involvement in an investigation. Thus, such victims may have been more likely
to be actively uncooperative in subsequent police questioning.
All of the examined victim demographic variables offered some degree of predictive value
in the cooperation model. Racial and ethnic minorities were more likely to be uncooperative with
police investigations. Compared to those of a similar white, non-Hispanic victims, the odds of a
black, non-Hispanic victim not cooperating was 1.23 times (p=0.20), and the odds of a Hispanic
victim not cooperating was 1.34 times (p=0.20). The results indicate that minorities, in regards to
race and ethnicity, were more likely to actively impede police shooting investigations. In contrast,
female victims were significantly more likely to cooperate with investigators. Indeed, when
compared to the odds of a male victim not cooperating, the odds ratio of a female victim electing
28
to not cooperate with a shooting investigation was just 0.83 (p=0.05). Similarly, the model
provides evidence that cooperation differed across age groups. Older victims were the least likely
to be uncooperative, whereas victims between the ages of 20 and 30 were the most likely to be
uncooperative in investigations. Compared to those under twenty, the odds ratio of a victim
between the ages of 20 and 30 not cooperating was 1.07 times as large (p=0.39).
Several scholars suggested, but did not examine, that victim demographics may play a role
in a victim’s cooperation (Litwin 2004, Riedel 1999). The results of the cooperation logistic
regression model strongly support this hypothesis. Based on the results of the model, racial and
ethnic minorities, males, and those between the ages of 20 and 30 were the most likely to be
uncooperative in police investigations. One potential explanation for the relationship between
victim demographics and cooperation is that, rather than victim demographics directly affecting
cooperation, both the demographic variables and the cooperation variable may be highly associated
with a victim’s gang involvement, a variable that could not be measured in this analysis due to
limitations in the Durham Police Department case files. For instance, from an analysis of the case
files and the investigator interviews, gang involvement in Durham, NC appeared to be especially
prevalent amongst racial and ethnic minorities, males, and individuals between the ages of 20 and
30. Consequently, deterrents that may exist to cooperation due to gang involvement, such as the
fear of retaliation, may be reflected in the noncooperation of these demographics.
Reasons for Victim and Witness Noncooperation
Multiple reasons may provide explanations for the causes of victim noncooperation,
particularly amongst different demographics. Previously literatures (Riedel 1999, Healey 1995)
suggested that the fear of retaliation and victim intimidation is the driving factor behind victim and
witness noncooperation. The results of our interviews are consistent with the hypothesis but also
29
suggest that numerous other reasons may provide explanations for noncooperation, including
general indifference, lack of trust in the police, and complications in the court system.
Consistent with findings in Riedel (1999) and Armstrong, Plecas, and Cohen (2013), all of
the interviewees mentioned the fear of retaliation as a primary reason for the absence of victim or
witness cooperation. The investigators suggested that many victims and potential witnesses fear
retaliation in the form of personal injury or injury to one’s family. Retaliation is often used as a
tool used by gang affiliated individuals to deter cooperation with the police, and for many victims
and witnesses, the costs of potential harm outweigh the marginal benefits of working with
investigators.
A variant of the retaliation explanation, the pervasiveness of the “no snitching” culture in
the target communities, was also frequently cited. In contrast to the fear of retaliation, the “no
snitching” culture refers to a general cultural norm in both gang and non-gang affiliated
communities against informing and working with the police. Those caught speaking with the
police are labeled as “snitches,” which carries social and safety repercussions. Moreover, the
phrase “snitches get stiches” was mentioned by several investigators as an added deterrent to
working with the police.
A few interviewees pointed to a general indifference towards crime, suggesting that for at
least some individuals, no reason exists to care about solving crimes. Similarly, some victims and
witnesses may refuse to work with the police in solving crimes because in their given communities,
the preferred method of handling disputes is through street justice. Hence, since investigators
would typically complicate attempts at street justice, members of the community, primarily the
gang-affiliated, choose to not discuss shootings with the police in hopes of retaliating on their own.
30
Furthermore, a general mistrust of the police, particularly amongst the demographics least
likely to cooperate, may be an underlying factor in a victim or witness’s refusal to cooperate. The
investigators suggested several possible explanations for general mistrust in a particular victim or
witness. First, community members may be skeptical of the police because of previous experiences
of law enforcement members lying to the individuals, such as promising to keep a witness’s name
anonymous and confidential but later sharing the name. Similarly, community members may lack
faith in the efficacy of the criminal justice system if they have had prior experience of working
with investigators and not perceiving direct results. The lack of results because of certain
circumstances, such as insufficient evidence to arrest a potential suspect, could foster a belief that
police investigators are not interested in assisting the community or not attempting to help to the
best of their ability in the investigations.
Likewise, many respondents pointed to complications in the court system, along with the
expectation that witnesses testify during trial, as a significant impediment to garnering cooperation
from a victim or witness. Some individuals may refuse to provide information to the police because
of the expectation that they later serve as witnesses because of their testimony. Other community
members willingly share information with the police for the purpose of an investigation but refuse
to have their name recorded, thus complicating attempts to later prosecute a suspect. The
investigators also suggested several explanations for individual reservations about testifying in a
public court. First, the public nature of the trial process, whereby all witness names are openly
available, could lead to concerns about safety and reputation. Moreover, the court process is
lengthy, arduous, and expensive for members of the community. Most trials require witnesses to
make several appearances in court over a long period of time, a difficult task for community
members that lack reliable forms of transportation or the capacity to take extensive time off work.
31
Another underlying theme is a general misunderstanding about criminal investigations. An
investigator mentioned the difficulty in acquiring cooperation in undocumented immigrant
communities due to a fear of deportation. According to the respondent, the fear stems from an
inherently false understanding of the criminal justice system because the police would never report
a victim of a crime to be deported. The concept of a misunderstanding of the criminal justice
system was echoed by several other respondents. The respondents emphasized that many
community members do not recognize the court system and the police as distinct entities. Thus,
when the court elects to make a certain decision that may not be favorable to the community, such
as choosing not to prosecute or setting a low bail, the police receive the blame for the decision,
thus leading to greater police mistrust and skepticism.
Investigator Strategies for Eliciting Cooperation
When asked about strategies for encouraging victim and witness cooperation, the homicide
and violent crime investigators lacked a strong consensus over preferred methods. Instead, there
was a general belief that no single strategy can most effectively encourage an individual to
cooperate with the police, and approaches and tactics should be tailored to a given individual or
situation. For instance, a majority of the investigators mentioned that in certain circumstances,
attempting to find common ground with the victim or witness is a particularly useful tactic.
Through referencing commonalities with the potential witness, such as race, being a single mother,
or growing up in the same community, some investigators find success in helping community
members see the investigators beyond their status as police officers, thus allowing the individuals
to more easily open up.
Similarly, respondents cited that in some circumstances, they resort to emotional appeal to
encourage cooperation, through the use of phrases such as “think of their family,” “this is
32
somebody’s child,” or “you can help stop this from happening to someone else.” However, many
felt that appealing to emotion is not a particularly effective method, and even if emotional appeal
proves useful temporarily, its effect is often fleeting, and victims and witness return to being
uncooperative.
Several respondents also mentioned leveraging pending criminal charges against witnesses
to elicit their cooperation. In certain circumstances, the threat of further criminal prosecution can
encourage a prospective witness to work with the police despite initial reservations. Some
investigators also indicated that on occasion, they would pit victims and witnesses against one
another to convince them to talk and share what they knew.
Additionally, respondents mentioned techniques that often involved accommodating the
witnesses. For example, some investigators make the witness feel comfortable, such as offering a
snack or drink, and others suggest meeting the victim or witness at a convenient location. For some
witnesses who fear being seen with the police in their communities, the best location is at the police
station, while for other witnesses who are uncomfortable or stressed at the police station, the best
meeting location may be at their homes. Likewise, a few interviewees indicated that sometimes
the best police investigator to conduct an interview is the one that is most relatable to the victim
or witness, be it the same gender or the same race, thus allowing the investigator and the witness
to share common ground.
A few investigators also indicated that face-to-face interviews, compared to talking over
the phone, are particularly effective because they allow for the officers to better read body
language, and individuals are less likely to lie or fabricate stories when in person. In addition,
several investigators mentioned that using two investigators in a witness interview is markedly
more effective at gathering information than using just one. The use of two investigators allows
33
one investigator to observe body language and responses while the other drives the line of
questioning. Two investigators can also play off one another’s questioning and language to prevent
lying or encourage the witness to be more cooperative.
Conclusion and Recommendations
This study aimed to accomplish three primary questions as they pertained to shooting
crimes: identify the circumstances most predictive of a violent crime’s eventual clearance by
arrest, determine the reasons for victim noncooperation in police investigations, and propose
measures to encourage greater victim cooperation. Through a combination of quantitative and
qualitative methods, I have identified and elucidated the factors most predictive of crime clearance
by arrest and victim cooperation in shooting crimes in Durham, NC.
Based on an analysis of two logistic regressions that predict the odds of a crime’s clearance
by arrest, in general, crime circumstances appear to be stronger predictors of clearance by arrest
than victim demographics. Particularly, crimes that occurred during the morning were more likely
to be cleared, possibly due to the availability of resources and the accessibility of the crime scene
and witnesses during earlier hours. Moreover, vehicular crimes were less likely to be cleared due
to the difficulty of identifying shooters in drive-by shootings. One possible reason for the reduced
clearance rates in vehicular shootings is that such crimes were typically representative of gangrelated activity, and victims may be reluctant to cooperate and share shooter details out of the fear
of retaliation. Similarly, crimes in which the call to the police was made from the Duke Hospital
also were less likely to result in arrest. When the same variable was included in the subsequent
noncooperation model, it became evident that victims who chose to go to the hospital for treatment
rather than call the police were more likely to be uncooperative and directly impede police
investigations.
34
The results of the clearance model thus suggest that victim cooperation, and factors which
may be indicative of victim behavior, play a major role in police investigations and their likelihood
of being solved. Thus, it becomes particularly important to identify the reasons behind victim
noncooperation in investigations. From the cooperation model examined in the study, victim
demographics appear to be most predictive of noncooperation. The demographics most likely to
be uncooperative in an investigation included racial and ethnic minorities, males, and individuals
between the ages of 20 and 30. One potential explanation for the demographic trends is the
prevalence of gang activity in these categories, which, according to the investigator interviews I
conducted in the study, serves as one of the driving motivators behind noncooperation.
Indeed, when asked about the reasons for noncooperation in investigations, police
investigators primarily pointed to two reasons: the fear of retaliation and a general skepticism of
the police and the court system. Thus, any attempts at improving victim and witness cooperation
in investigations should aim to address the identified underlying causes, such as making the
witness feel more comfortable or attempting to find common ground, two particularly effective
tactics outlined by investigators during their interviews.
The investigators provided mixed responses when asked about ideas and suggestions for
improving systemic victim and witness cooperation. Most of the interviewees provided multiple
suggestions and recommendations, but some respondents struggled to come up with responses.
Those who struggled to provide responses stated that they believe that there is little within the
control of the police that can be done to improve cooperation with the community. Others
mentioned that as long as gang culture continues to exist, and the fear of retaliation and “no
snitching” culture that come with it, little can be done to encourage greater cooperation with
investigators.
35
The results of both the quantitative and qualitative analyses of the thesis thus suggest that
an improvement in citizen cooperation, and consequently clearance rates, could occur through a
shift in culture and norms. Restarting the gang unit at the Durham Police Department could assist
in encouraging cooperation since gang behavior and intimidation is typically responsible for
developing a culture that encourages victim and witness noncooperation. Likewise, increasing
programs that work with youths and children in the communities could help accomplish a similar
objective and combat the pervasive fear of organized crime in the communities.
Moreover, relocation services for victims and witnesses, a suggestion proposed by a few
respondents, may also prove effective. Since one primary reason for noncooperation, according to
interviewees, is the fear of being seen speaking to the police, victims and witnesses may be more
willing to share with the police if their safety is guaranteed or they are given the option to relocate
outside of their neighborhoods and communities. In general, improved police protection could
assist in assuaging the fears of prospective victims and witnesses. If police departments can
provide the necessary protection to victims and their families, thus mitigating any fears about
retaliation in their communities or from gangs, they may be more effective in subsequently
encouraging greater cooperation.
Furthermore, as indicated by several investigators, many Durham communities are likely
to be affected by the pervasiveness of a culture that is jaded and indifferent to crime due to its
regularity and the inefficacy of police investigations. Such indifference renders some community
members to be unmotivated and unwilling to work with investigators. Thus, increasing cooperation
starts with building trust between the police and the community. Because one of the driving factors
in noncooperation is mistrust of the police and the court system, developing trust between
36
investigators and members of the community could serve as a stepping stone to a better partnership
and thus more cooperation.
Several strategies may result in developing greater trust. For instance, expanding
community policing, especially amongst police investigators in the violent crimes and homicide
units, may improve perceptions of the police as well as expose investigators more directly to the
cultures of their communities. While community policing already exists to a certain degree at the
Durham Police Department, detective engagement in community policing should be particularly
expanded. Interviewees viewed events like National Night Out, an annual campaign that aims to
promote the police-community relationship through celebration and community events, as widely
successful. Improving trust may occur if police investigators become more regularly involved in
the communities they work with, not just when they are responding to an incident. Such activity
helps individuals identify with the police and view the investigators as community members rather
than distant cops. Seeing investigators frequently involved in the community also lends credibility
to the belief that the police hold the community’s best interest at heart. Several investigators
mentioned that the results of such strategies can be seen empirically; in general, if an officer
already has rapport with community members, they are far more likely to share important
information when it is needed.
Similarly, due to the efficacy of investigators sharing common ground with the witnesses
they interview, another potential police reform could involve focusing on developing the police
force to be more representative of the communities they work with. For instance, police
departments could invest more resources into recruiting investigators directly from the local
communities. Such recruits would not only already be more familiar with the community, but they
37
would also likely be more successful in garnering the trust of potential victims and witnesses
during investigations because they could reference commonalities in upbringing.
Education, particularly as it relates to the role of the police, may also facilitate in
accomplishing a similar goal. For example, educating the public on the role and responsibilities of
the police could help illustrate the distinctions between the police and the court system. At least a
few investigators felt that some criticism and lack of trust in investigators stems from flaws in the
court system. Therefore, more explicitly teaching the differences between the two entities could
help illustrate to civilians that even when investigators work to the best of their abilities,
shortcomings can still occur due to limitations in the courts.
Accordingly, a change in the current court system may also be a necessary precursor to
improving victim and witness cooperation. In general, the respondents who pointed to
complications in the court system as an influencing factor in noncooperation also likely believed
that a change in the system could improve cooperation. Specifically, two major improvements
were mentioned. First, interviewees believed that there should be a change to witness testimony
and information as it pertains to privacy. Instead of witness information being publically available,
witness testimony and information should remain private and only available to judges and lawyers.
The investigators felt that more individuals would come forward with information pertinent to an
investigation if they could be reassured that their information would remain private. Secondly,
interviewees emphasized that a reduction of court appearances could lessen the burden placed upon
witnesses. In the status quo, witnesses are required to appear in court numerous times, and the
investigators felt the multiple appearances should not be necessary. Though both changes may
result in greater witness cooperation, the feasibility of such initiatives is less certain given present
limitations in the courts.
38
However, overall, one of the most effect ways of improving cooperation may simply be
through increased staffing and resources, a sentiment shared by many investigators. The majority
of respondents felt that an increase in staffing would consequently lead to lighter caseloads,
increased police presence, and greater community policing. With lighter caseloads and greater
presence, investigators could dedicate more time to working with victims and witnesses and
encouraging their cooperation.
Another potential police strategy that was hypothesized and explored in this study was the
expansion of monetary incentives to encourage cooperation. However, because the lack of
financial payments was not frequently cited by investigators as an influencing cause in
noncooperation, the effects of such a program may be limited.
For instance, investigators shared mixed responses to the State Victim Compensation
(SVC) program, and many expressed that they were unfamiliar with the specifics of the program.
Only a few respondents mentioned ever receiving contact from the SVC, and several respondents
stated that the SVC program should be better explained to both police investigators and victims.
One large limitation of the SVC is that the program typically takes weeks to get in touch with
victims, and often times, cases are no longer active at that point. Hence, the long transaction period
for the SVC program renders it difficult to use in encouraging cooperation. Consequently, only a
few respondents stated that they believe that the program can be helpful, more respondents
indicated that they were not sure if the program is helpful, and some respondents even indicated
that they did not believe the program is helpful. In addition to administrative limitations of the
SVC program, the respondents also suggested that if victims want to cooperate, they will often
cooperate from the beginning, and money will not incentivize a victim that does not want to
cooperate in the first place.
39
Another monetary program that I examined was the use of CrimeStoppers. All of the
investigators indicated that they mention CrimeStoppers during an investigation, especially if a
prospective witness appears to be knowledgeable about a case but hesitant to speak with the police
due to concerns about privacy. Furthermore, for all respondents, amongst the communities they
worked with, CrimeStoppers is a well-known program, though admittedly, not all people fully
understand how CrimeStoppers works. However, the investigators were slightly more divided on
their opinions of CrimeStoppers. Most interviewees did agree that on the whole, CrimeStoppers is
a useful program to gather information, particularly due to its anonymity that assuages a primary
fear of community members. Nonetheless, several respondents raised potential issues and
limitations of the program. Some indicated that the existence of CrimeStoppers leads some
individuals to not share information directly with officers because they know that they can be paid
for the information if they instead used CrimeStoppers. Others mentioned that CrimeStoppers tips
generally are information that investigators already know, though sometimes they can be useful in
corroborating existing information. They stressed that while CrimeStoppers is useful for providing
leads and tips, it is rarely possible to solve a case purely from a CrimeStoppers tip.
Therefore, the true merits of the CrimeStoppers program may lie less in the monetary
component of the program and more in the anonymity it provides to prospective witnesses. Since
CrimeStoppers allows victims and witnesses to provide information without revealing their private
information, it can be a particularly useful tool in combatting the pervasive fear of retaliation in
communities.
Ultimately, the results of this thesis suggest two major policy conclusions. First,
encouraging victim cooperation ought to be a primary objective in police investigations because
of the strong relationship between victim behavior and crime clearance. If a police department can
40
increase rates of victim cooperation, and more importantly decrease incidences of noncooperation,
it may be able to improve its overall clearance rates, particularly in nonfatal shooting crimes.
Secondly, attempts to improve systemic cooperation should aim at reforming the cultures of fear
and mistrust that often discourage cooperation. For instance, providing more credible protection
for prospective witnesses or expanding anonymous tip programs, such as CrimeStoppers, may lead
to victims and witnesses feeling more comfortable about opening up to investigators. Similarly,
department programs that encourage the development of trust, such as community policing,
education programs, and recruiting directly from the surrounding areas, may prove valuable in
building trust between the police and the communities they work with directly. Particularly amidst
current events, such tactics may prove especially effective in working towards the ultimate goal of
improving the safety of all individuals.
41
References
Armstrong, Jennifer, Darryl Plecas, and Irwin M. Cohen. 2013. The Value of Resources in Solving
Homicides: The Difference Between Gang Related and Non-Gang Related Cases. Centre
for Public Safety and Criminal Justice Research. University of the Fraser Valley.
Black, Donald J. 1970. “Production of Crime Rates.” American Sociological Review 35 (4): 733–
48. http://www.jstor.org/stable/2093948.
Black, Donald J. The Behavior of Law. New York: Academic Press, 1976.
Braga, Anthony A., and Desiree Dusseault. "Can Homicide Detectives Improve Homicide
Clearance Rates?." Crime & Delinquency (2016): 0011128716679164.
Cook, Philip J., and Gregory W. Fischer. 1976. Citizen Cooperation with the Criminal Justice
System. Durham, NC: Duke University.
Dawson, Myrna, and Ronit Dinovitzer. 2001. "Victim Cooperation and the Prosecution of
Domestic Violence in a Specialized Court." Justice Quarterly 18 (3): 593-622.
doi:10.1080/07418820100095031.
Durham Police Department 2015 Annual Report. 2015.
http://durhamnc.gov/DocumentCenter/View/9455, accessed March 12, 2016.
Elias, Robert. 1984."Alienating the Victim: Compensation and Victim Attitudes." Journal of
Social Issues 40 (1): 103-16. doi:10.1111/j.1540-4560.1984.tb01084.x.
Federal Bureau of Investigation. 2014. Crime in the United States: Uniform Crime Reports.
Washington, DC: Government Printing Office.
Healey, Kerry M. 1995. Victim and Witness Intimidation: New Developments and Emerging
Responses. Washington, DC: National Institute of Justice.
42
Keel, Timothy G., John P. Jarvis, and Yvonne E. Muirhead. 2009. "An Exploratory Analysis of
Factors Affecting Homicide Investigations: Examining the Dynamics of Murder Clearance
Rates." Homicide Studies 13 (1): 50-68. doi:10.1177/1088767908326903.
Litwin, Kenneth J. 2004. "A Multilevel Multivariate Analysis of Factors Affecting Homicide
Clearances." Journal of Research in Crime and Delinquency 41 (4): 327-51.
doi:10.1177/0022427803260269.
Paré, Paul-Philippe, Richard B. Felson, and Marc Ouimet. 2007. "Community Variation in Crime
Clearance: A Multilevel Analysis with Comments on Assessing Police Performance."
Journal of Quantitative Criminology 23 (3): 243-58. doi: 10.1007/s10940-007-9028-0.
Pfuhl, Erdwin H. 1992. "Crimestoppers: The Legitimation of Snitching." Justice Quarterly 9 (3):
505-28. doi:10.1080/07418829200091501.
Pittman, D. J., & Handy, W. 1964. “Patterns in criminal aggravated assault.” J. Crim. L.
Criminology & Police Sci. 55 (462).
Pogue, Thomas F. 1975. "Effect of Police Expenditures on Crime Rates: Some Evidence." Public
Finance Review 3 (1): 14-44. doi:10.1177/109114217500300102.
Puckett, Janice L., and Richard J. Lundman. 2003. "Factors Affecting Homicide Clearances:
Multivariate Analysis of a More Complete Conceptual Framework." Journal of Research
in Crime and Delinquency 40 (2): 171-93. doi:10.1177/0022427803251125.
Riedel, Marc. 1999. "The Decline of Arrest Clearances for Criminal Homicide: Causes, Correlates,
and
Third
Parties."
Criminal
Justice
doi:10.1177/088740349900900302.
43
Policy
Review
9
(3):
279-305.
Roberts, Aki. 2008. "The Influences of Incident and Contextual Characteristics on Crime
Clearance of Nonlethal Violence: A Multilevel Event History Analysis." Journal of
Criminal Justice 36 (1): 61-71. doi: 10.1016/j.jcrimjus.2007.12.007.
Wellford, C., & Cronin, J. 1999. An Analysis of Variables Affecting the Clearance of Homicides:
A Multistate Study. Washington, DC: Justice Research and Statistics Association.
Appendix A
44
Appendix B
DPD INVESTIGATORS QUALITATIVE INTERVIEW GUIDE
INTRODUCTION
1. Please tell me about your history in law enforcement.
2. How long have you been in your current position and department?
CASE FILES
Asked per case (3-5 cases per investigator)
CASE FILE INTRODUCTION
1. Tell me about this case.
VICTIM QUESTIONS
(Do not ask for homicide cases, skip to witness questions)
1. How cooperative was the victim in the police investigation?
2. Do you believe that the victim was being truthful?
3. What strategies did you use to encourage the victim to cooperate?
4. How crucial was the victim’s cooperation to your investigation?
WITNESS QUESTIONS
1. Who were the key witnesses in this case?
2. Was this witness willing to speak with you about the case?
3. Do you believe that the witness was being truthful?
4. What did you do to encourage key witnesses to speak with you?
5. How crucial was the witness’s cooperation to your investigation?
6. How vital is victim/witness cooperation to continue/re-open this investigation? (If case is
inactive/open)
45
GENERAL QUESTIONS
1. What are the challenges that you feel are associated with witness and victim cooperation?
2. Why do you think homicides have a higher arrest rate than nonfatal assaults?
3. How is witness behavior different in fatal vs. non-fatal cases? (If investigator has worked
both homicide and aggravated assault cases)
4. What tools do you utilize to help with victim and witness cooperation?
5. Does the state victim compensation program work to encourage participation?
6. How often do you tell individuals about Crime Stoppers during an investigation?
a. Do you tell everyone you speak with about the program?
b. Do you feel that most people in the community know about the program?
c. How helpful do you feel this program is for solving gunshot crimes?
7. What do you think would help to improve victim and witness cooperation?
8. If you had more resources what would you do to improve victim and witness
cooperation?
PERCEPTIONS OF POLICE ROLE
1. Who do you see yourself serving in your duties as an investigator?
WRAP UP
1. Is there any other information that you would like to share on this topic?
ADDITIONAL QUESTIONS
(if not already covered through other sections)
1. What features of a case make it most challenging to close?
2. What are some features of cases that making for better closure rates?
3. What typically causes you to stop pursuing a case?
46