The Impact of Investigation Strategies and Tactics on Homicide Clearance

Abstract

This study examines whether changes made to the investigation procedures within the Rochester Police Department’s homicide unit have an effect on homicide clearance net of theoretically significant variables related to victim and police devaluation, event characteristic, and victim lifestyle. It examines 132 homicides investigated by the homicide unit over 4 years consisting of the pre-intervention period and post-intervention period. It is hypothesized that investigation tactics will affect the odds of clearance, even after controlling for other theoretically relevant variables. The findings support this hypothesis, suggesting that the approaches and tactics police departments implement within their investigative units are important.

Keywords

homicide clearance investigation detectives

Introduction

The aftermath of a homicide event poses many negative consequences to the deceased’s family, friends, and society as a whole. Homicide often results in psychological distress experienced by the victim’s family and friends (Armour, 2002). At the societal level, some scholars estimate that the average cost per homicide is approximately 17.25 million dollars (DeLisi et al., 2010). When offenders are not apprehended, the potential deterrent effect of sanctions is diminished (Braga & Dusseault, 2018) and police legitimacy may be undermined (Roberts & Lyons, 2011). This can result in an increase of fear, legal cynicism, and self-help violent behaviors within communities (Pizarro, 2017; Riedel, 2008; Roberts & Lyons, 2011). As a result, apprehending homicide offenders, and clearing these cases, is of paramount importance to the police and society.

In the United States, cases are considered cleared when the police arrest and charge offenders with a criminal offense and turn them over to the court for adjudication, or when exceptional circumstances such as the death of the offender preclude apprehension (Riedel, 2008). Nationally, clearance rates have declined from the 90% range in the early 1960s to the 60% range presently (Braga & Dusseault, 2018). Studies that have examined clearance often do so from the theoretical lens of victim devaluation and event characteristics, and more recently by employing neighborhood processes to measure police devaluation, as well as victim lifestyle explanations. While these theoretical perspectives have contributed to the literature, the examination of investigatory policies as they relate to homicide clearance is comparatively rare. The present study seeks to fill this void by examining the effect on clearance of investigatory strategies and tactics implemented by the Rochester Police Department (RPD) in 2012.

Homicide Clearance Research

The bulk of research on homicide clearance has examined this issue from a victim devaluation, event characteristics, police devaluation, and/or victim lifestyle theoretical lens. Victim devaluation, which is also known as discretionary or extra-legal perspective, has its roots in Black’s (1976) theory of the behavior of law. This perspective posits that the amount of law one receives is a function of the social space one occupies. As applied to homicide clearance, this approach is based on the assumption that the police exercise discretion when investigating homicides based on extralegal characteristics such as victim socioeconomic status and demographics (Riedel, 2008). As a result, homicides are less likely to be cleared if the victim was a member of a group with low social status because officers are not as motivated and do not put in the necessary effort when investigating these cases.

Conversely, the event characteristics perspective, which is also known as the nondiscretionary or solvability perspective, posits that because of the seriousness and visibility of homicides, work group norms motivate investigators to work diligently and aggressively to clear all cases, regardless of victim social status (Klinger, 1997; Riedel, 2008). According to this perspective, cases are not cleared due to event characteristics that make homicides harder to investigate and clear. For instance, lack of witnesses, firearm usage, gang and instrumental motives, and the location in which the homicide incident takes place make the investigation process more difficult, and thus less likely to be cleared.

Police devaluation focuses on the perceptions of community residents. According to this perspective, the negative association between clearance rates and non-White low socioeconomic status victims is more attributable to devaluation of the police as effective agents for resolving conflicts, and not the devaluation of victims as posited by victim devaluation. As described by Keel, Jarvis, and Muirhead (2009), in areas where residents do not trust the police, witnesses, family members, and other parties with valuable information may choose not to participate in the homicide investigation. In their view, homicides go unsolved because persons who could provide valuable information to solve the crime “view the police with mistrust, [and] also view the conflict as one in which informal street justice may be a more effective resolution” (p. 63). The areas that traditionally experience these types of mistrust of the police are often those suffering from structural disadvantages and racial isolation (Mancik, Parker, & Williams, 2018). As a result, neighborhood social structural characteristics such as socioeconomic status, racial composition, and, more recently, collective efficacy are used as proxies to measure this perspective (Keel et al., 2009; Mancik et al., 2018; Petersen, 2017; Puckett & Lundman, 2003). This perspective is also referred to as neighborhood processes.

The victim lifestyle perspective overlaps both victim devaluation and event characteristics and posits that homicide cases are more difficult to solve when they involve victims who lead deviant lifestyles. In this perspective, deviant lifestyle is defined as being a drug dealer, being a gang member, and/or having a criminal record. Rydberg and Pizarro (2014) found that victim participation in a deviant lifestyle is associated with the distribution of clearance impeding event factors across homicide incidents. Specifically, victims enmeshed in criminal lifestyles were more likely to be killed on a public street and also killed in a confrontation or drive-by with a firearm, both event characteristics that make cases harder to clear. As a result, they suggest that cases involving victims who are known-criminals are more difficult for detectives because they are harder to locate credible witnesses and to retrieve physical evidence from.

These four perspectives have each found some support in the homicide clearance literature (Riedel, 2008). Some studies reinforce the assumptions of victim devaluation (Hawk & Dabney, 2014; Keel et al., 2009; Lee, 2005; Litwin & Xu, 2007), while other studies do not (Alderden & Lavery, 2007; Litwin, 2004; Puckett & Lundman, 2003). Similarly, examinations of police devaluation have also found support in some studies with clearance decreasing in socioeconomic disadvantaged neighborhoods (Litwin & Xu, 2007; Ousey & Lee, 2009; Petersen, 2017), while other studies have found no relationship (Borg & Parker, 2001; Litwin, 2004; Puckett & Lundman, 2003; Regoeczi & Jarvis, 2013; Xu, 2008). Studies that have examined the racial composition of neighborhoods as a proxy for police devaluation have also found mixed results (Litwin & Xu, 2007; Petersen, 2017; Puckett & Lundman, 2003; Xu, 2008). Conversely, the event characteristics perspective has found more robust support in the literature. Research has found weapon type, incident location, motive, and the availability of witnesses often affect whether an investigation is cleared (Alderden & Lavery, 2007; Jiao, 2007; Litwin, 2004; Mancik et al., 2018; Marché, 1994; Puckett & Lundman, 2003; Rydberg & Pizarro, 2014; Wellford et al., 1999). Finally, a recent examination of victim lifestyle found support (Rydberg & Pizarro, 2014), but given the newness of this approach, more research is needed.

While such prior work has greatly contributed to the literature, a line of research that has received significantly less attention is that which focuses on the investigation process. After a homicide is reported and the police respond, whether a suspect is apprehended may be determined by the actions investigators take or do not take. As a result, to better understand the dynamics that increase the likelihood of clearance, investigation processes should be incorporated and examined simultaneously with the prevalent theoretical perspectives.

The Homicide Investigation

The main objective of the homicide investigation is to clear the case with the arrest of the offense perpetrator. The investigative process is generally comprised of two stages: preliminary and supplementary (Geberth, 1990; Hough & McCorkle, 2016; Miletich, 2003). The preliminary investigation is initiated when the first officer arrives at the homicide location (Hough & McCorkle, 2016). During the preliminary phase, the responding officer secures the scene to preserve evidence, identify potential witnesses, and, if possible, apprehend the offender. Responding officers also photograph the scene to record its appearance and condition at the time of their arrival, and document important times to develop a timeline of the event. The supplemental investigation is conducted by the investigator(s) assigned to the case (Hough & McCorkle, 2016). During this stage, the investigator(s) follows up on the information prepared by responding officers. This includes canvassing the incident area, locating and contacting witnesses, conducting interviews, and examining evidence retrieved by responding officers.

Studies that have examined the investigation process suggest that it is structured and complex (Hawk & Dabney, 2014). According to Stelfox (2011), most of the work is cognitive in nature as it involves collection of information, followed by evaluation, interpretation, and deciding what action to take. Investigators often work backward from the event to gain an understanding of the precipitating circumstances and create a timeline of the lethal act (Cronin, Murphy, Spahr, Toliver, & Weger, 2007). The victim’s lifestyle is important in this process as it gives the investigator a glimpse of potential motives and offenders.

The Investigation Process and Clearance

The majority of studies examining the effect of investigation tactics on clearance either focus on identifying best practices in investigations by comparing high and low clearance police departments or by adding investigation relevant variables as a control. Studies that have examined high and low clearance agencies suggest that investigation tactics have, at best, a marginal effect on clearance. One of the first studies in this area found that supplementary investigation tactics did not significantly influence clearance (Chaiken, Greenwood, & Petersilia, 1977). Instead, arrest clearance was due primarily to the work of patrol officers during the preliminary investigation, information provided by the public, and routine clerical processing. Other researchers have also found limited effects between investigation tactics and clearance. For example, Eck (1992) found that clearance is often beyond the control of investigators and is an outcome based on incident event characteristics. Worrall’s (2016) analyses of clearance rates for property and violent crimes by 570 law enforcement agencies over a 13-year period found that investigation tactics and resources have only marginal effects on clearance of violent crimes.

Studies that have examined various measures of investigation tactics as control variables support these findings. Investigator experience and departmental enforcement efforts (i.e., number of officers per capita and monetary expenditures) are not significant covariates of clearance according to Marché (1994), as well as Cloninger and Sartorius (1979). Similarly, Pucket and Lundman (2003) found no effect between investigator workload and experience on homicide clearance. Rydberg and Pizarro (2014) examined investigator workload with measures that tapped into time since last case, number of open cases, and whether the last case on which the officer worked was open or closed. They found time from last case to be the only significant variable, with longer periods between investigation assignments resulting in longer clearance times for the current case. They attributed this finding to a possible “getting rusty” effect.

Other researchers, however, have found that investigation tactics do matter for a successful case outcome. As reported by Braga and Dusseault (2018), changes in homicide investigation tactics and protocols by the Boston Police Department, through the use of a problem-oriented policing approach, resulted in an increase in clearance rates. This included the addition of investigators, extensive training on investigation best practices and forensics, the development of standardized investigation protocols, and the addition of monthly peer-review meetings to discuss open cases. Following these changes, clearance of homicide cases increased by 18%.

The discrepancy of findings across studies might be a product of the type of investigation variables examined, as distinct investigative variables might not be tapping into the specific factors that influence clearance. Indeed, qualitative and mixed method studies that have examined the association between best practices in investigations and agency clearance rates find that specific tactics do indeed increase the probability that a homicide is cleared. One of the most important predictors of investigation success is the cooperation of witnesses to the crime (Keel et al., 2009; Wellford et al., 1999). Other studies have found that the preliminary investigation is also a critical factor in clearance as this is the most important time for officers to locate and secure witnesses and collect key material evidence (Horvath, Meesig, & Lee, 2001; Wellford et al., 1999). The speed with which homicide detectives, evidence technicians, and medical examiners are notified and the time it takes them to arrive to the scene are also associated with clearance of the case (Wellford et al., 1999). Carter and Carter (2016) found that the investigation that takes place during the first 48 hr is pivotal to successful case outcomes. Agencies that secure witnesses and material evidence and narrow potential motives and offenders within this period have higher clearance rates.

Management oversight, providing optimal training to investigators, and the availability of resources such as DNA analysis and other forensic tools are also important in clearing cases (Carter & Carter, 2016; Keel et al., 2009; McEwen, 2011, 2013). Keel and colleagues (2009) suggested that training detectives on proper investigation techniques improves clearance rates in agencies. They also found a marginal but significant positive effect from using scientific evidence analyses such as blood spatter, voice stress analysis, and other similar analytic tools. Nonetheless, Schroeder and White (2009), in their examination of homicide clearance from 1996 to 2003 in the New York Police Department, found that detectives use these tools only after other investigation options were deemed fruitless. McEwen’s (2013) examination of the use of forensic specialists and evidence in the investigation of Phoenix homicides found that although forensic specialists are effective in the collection of evidence, their expertise was only effective in clearing the most difficult investigations. Clearance rates in the Phoenix Police Department did not increase despite their success in clearing difficult cases, however. Although forensic evidence might not be as fruitful in the clearance of cases, these tools can help post-arrest in the conviction of arrested homicide offenders. Indeed, McEwen (2011) found in his analyses of the use of forensic evidence in Denver, Colorado, and San Diego, California, that these tools often result in probative evidence, which is useful in the court adjudication stages.

Finally, staffing is an important covariate. Horvath et al. (2001), as well as Wellford and colleagues (1999), found that assigning more than one investigator to a case increases clearance. Carter and Carter (2016) suggested the optimal level of staffing is one sergeant and four investigators per squad, with investigators rotating as “lead” and serving in that capacity in no more than three homicide investigations per year. Relatedly, Wellford et al. (1999) found how investigators approach an investigation also affects clearance. Following up with witnesses in a timely manner, attending autopsies, and detailed note taking increases case clearance. Based on prior research, it is hypothesized that investigation tactics will indeed affect the odds of clearance, even after controlling for other theoretically relevant variables.

Present Study

The current inquiry seeks to add to the literature by further exploring whether police investigatory tactics affect case clearance. To do so, this study examines a dataset of homicide incidents before and after the RPD established a new investigative protocol in 2012. More specifically, this study examines whether changes made to the investigation strategies and tactics within the RPD’s homicide unit have an effect on homicide clearance net of theoretically significant variables. Based on prior research, it is hypothesized that investigation tactics will affect the odds of clearance, even after controlling for other theoretically relevant variables.

Research Site

Rochester, New York, is the third largest city in New York State, with a population of 210,565 as of the 2010 Census. Whites comprised 37% of the population, African Americans 39%, Hispanics 18%, and 6% Other. In 2013, Federal Bureau of Investigation (FBI) statistics show that Rochester’s murder rate per capita of 19.9 per 100,000 residents exceeded that of both New York City (3.9) and the national rate (4.5) (FBI Uniform Crime Reports, 2013). During the study, the city had an average of 34 murders per year during the pre-intervention period (39 murders in 2010 and 29 in 2011), and an average of 32 murders post-intervention (37 in 2013 and 27 in 2014). Rochester is also among the poorest cities in the country. The U.S. Census estimates that Rochester’s median income was US$30,875 in 2013, with 32.9% of Rochester residents living below the poverty level.

Homicide investigation process

Homicides in Rochester are investigated by the RPD’s Major Crimes Unit (MCU), which is housed in the Central Investigations Division at RPD Headquarters. Homicide investigators begin their careers as uniform patrol officers and gain valuable investigative experience on the street. Officers are eligible for an investigator position once they have 3 years of experience in patrol. Promotion to investigator requires completion of a written examination and a role-play interview with a victim, witness, or suspect played by an actor trained on how to react to questions posed by the candidate. Once promoted, investigators are assigned to a patrol section in a plain clothes assignment.

Prior to 2012, the MCUs’ Homicide Squad was comprised of two sergeants, who each supervised from four to six investigators. A team of two investigators and a sergeant rotated being on call each week, from 4:00 p.m. Friday to 4:00 p.m. the following Friday. The “on-call” team was assigned all homicides that occurred during their on-call week, which occasionally resulted in the on-call team being assigned multiple homicides in 1 week. Consequently, on-call homicide detectives would occasionally have to temporarily drop an investigation during those crucial, initial hours of the case and shift their focus to another homicide. Even if the on-call team acquired only one case during their on-call week, they worked the case alone, without assistance from other investigators in the unit. The “team” consisted of only those two investigators who responded to the scene, were briefed by patrol command or investigators, and then took over all investigative tasks for the case. There was little supervisory oversight of whatever work was subsequently done or not done.

In January 2012, MCU was expanded with the addition of seven investigators transferred from decentralized patrol division assignments. Three of these investigators are initially assigned to the Homicide Squad, one was assigned to the Impact Team (sexual assault investigations against children), and three had no permanent assignment but were used to help out where needed in Homicide, Impact, Arson, Sex Offenders, and Bank Robbery. In addition, another sergeant was added to MCU.

Following the 2012 expansion, the manner in which homicides were investigated underwent important changes. First, more personnel were available to assist with homicides in general, and especially when a homicide investigator had a day off or was on vacation. This resulted in various combinations of homicide investigators handling new cases with someone other than their regular partner. Second, and most importantly, the two primary case investigators began to be supported by other investigators who could assist in important, but time-consuming tasks, such as writing search warrants (for premises, vehicles, social media sites, and cell phone data), assisting in the retrieval of video, and interviewing witnesses.

In January 2013, the change in tactics was formalized when the homicide unit was re-organized to reflect the way homicides were being investigated. The unit now had three sergeants, each with a squad of four investigators. Under this reorganization, a sergeant and two investigators are primarily responsible for being called in during their “on-call” week if the case occurs after duty hours. However, the sergeant is given discretion to call in additional team members if case dynamics (e.g., multiple victims, witnesses, and/or suspects, as well as the complexity of the incident itself) warranted the use of additional personnel. Even if additional personnel are not initially called in after Major Crimes is notified, non-primary team members are assigned case tasks (e.g., writing search warrants, redoing the neighborhood canvass) after reporting for duty the following morning.

If a case came in during duty hours, all team members and even other on-duty investigators are expected to go to the scene and assist the primary team. Most importantly, once a team “catches” a homicide, they are no longer on-call for the remainder of the week and a new team volunteers to go up on call. The only exception to this rule occurs if the on-call team “catches” a “ground ball,” meaning a case that essentially solved itself, such as a domestic murder with a suspect in custody or a murder-suicide.

Between the two primary investigators, one functions as the “lead” on the case. As a practical matter, however, being the “lead” matters little. The two primary investigators, as well as the two non-primary team members, attend briefings with their sergeant, unit command personnel, district attorneys, and other RPD personnel providing support to the investigation. While the two primary investigators handle more of the investigatory load in each case, investigatory tasks are split between them without reference as to who is the lead.

Other post-2012 changes include more supervisory oversight. Progress meetings are regularly scheduled during which the team assigned to the investigation reported to MCU’s captain and lieutenant. These meetings include personnel from the District Attorney’s office and the Monroe County Crime Lab. A “lab meeting” is scheduled at the crime lab to present a power point describing the case and the evidence that needs to be tested. In addition, the RPD invested in a “command vehicle” that contain a conference room, video, Wi-Fi, and a white board. The command vehicle, which essentially is a large recreational vehicle (RV), is driven to every homicide scene. Briefings are now held inside the command vehicle and tasks assigned immediately after the homicide is reported. See Table 1 for a summary of pre- and post-2012 tactics.

Table 1.

Investigation Tactics Summary—Pre- and Post-Intervention 2012.

Pre-intervention	Post-intervention
Two investigators are responsible for all homicides that occurred during their “on-call week”	On-call investigators handle only one new case that occurs during their on-call week
A homicide “team” consists of two investigators	A homicide “team” consists of four investigators
The two investigators assigned to the homicide are solely responsible for all investigator tasks	The two “lead” investigators are supported by other team members who assist in time-consuming tasks such as writing search warrants for cell phone data and obtaining video
Investigators control the case with little supervisory oversight	Investigators meet regularly with MCU command personnel and the assistant district attorney assigned to the case to discuss case progress and receive guidance as well as resources from other RPD departments, including the SIS (“SIS,” aka “the Vice Squad”) and the Tactical Unit, or federal agencies, as needed.
	Initial stages of the investigation are more organized and controlled. Responding MCU Homicide investigators and supervisors are briefed in the Command Vehicle, and tasks are assigned. Video collected may be viewed on the video system inside the Command Vehicle.
	An ATF special agent is embedded in MCU and assists with investigatory tasks, as well coordinating federal resources when needed.
	A presentation is routinely made to the Crime Lab after each homicide to discuss evidence collected and whether it needs to be tested
	When it is decided that an arrest can be made, a meeting is scheduled with the First Assistant District Attorney and other RPD specialized units, such as Special Investigations and the Tactical Unit, to present the case and coordinate locating and arresting the suspect(s).

Note. ATF = Bureau of Alcohol, Tobacco, Firearms and Explosives; MCU = Major Crimes Unit; RPD = Rochester Police Department; SIS = Special Investigation Section.

Method

The current study centers on 132 homicides investigated by the RPD’s Homicide Unit over 4 years consisting of the pre-intervention period (January 1, 2010, and December 31, 2011) and post-intervention period (January 1, 2013, and December 31, 2014). Data were obtained directly from RPD’s homicide files compiled by the investigators assigned to the case and were collected by research assistants with approval from departmental administrators. A data collection protocol instrument was employed to ensure uniformity between the four researchers who were trained by the principal investigator overseeing all data collection.¹

Data collected include variables associated with clearance in the literature (Puckett & Lundman, 2003; Riedel, 2008; Rydberg & Pizarro, 2014), including investigator-related measures (e.g., investigator experience, caseload), victim characteristics (e.g., gender, race, age, lifestyle, and criminal history), event characteristics (e.g., weapon used, mode, witnesses, inside/outside residence), and neighborhood characteristics (income level, race). Neighborhood characteristics were collected from the U.S. 2010 Census. Similar to Puckett and Lundman (2003), census tracts were used as proxies for neighborhoods.

Dependent variables

Two dependent measures of homicide clearance are used. The first, clearance, measures whether the case was cleared by arrest.² RPD detectives consider a case closed once an offender is arrested, even if they anticipate making further arrests. The second dependent variable, time to clearance, captures the amount of days between the date on which the investigation began and the date on which the first arrest was made. Open cases, which were not cleared with an arrest during the study period, were right censored to avoid missing data on that variable.³

Independent variables

This intervention variable captures the investigation tactics and procedures (as a whole) adopted by MCU in 2013 and utilized throughout the remainder of the study period (1 = post-investigation, 0 = pre-intervention). The remaining measures consist of investigator, victim, event, and neighborhood variables (see Table 2).

Table 2.

Variable Coding.

Variable	Coding
Dependent
Cleared	1 = case was cleared by arrest; 0 = case not cleared by arrest
Time to clearance	Continuous duration of time elapsed (in days) between the date investigation began and the date of the clearance or right-censoring
Independent
Intervention	1 = post-intervention (2013-2014); 0 = pre-intervention (2010-2011)
Control
Investigator
Years of experience	Years of experience as lead investigator
Number of open cases	Number of cases lead investigator has open
Victim
Gender	1 = female; 0 = male
Race	1 = Non-African American; 0 = African American
Age	1 = <15 years old or >44 years old; 0 = 15-44 years old
Deviant lifestyle	Summative scale of gang membership, drug dealing, other illegal activity, and prior criminal record (0-3)
Event
Weapon	1 = non-firearm; 0 = firearm
Mode	1 = face-to-face; 0 = other
Witnesses	1 = witness; 0 = no witnesses provided information
Residence	1 = victim or offender residence; 0 = other
Neighborhood
Income
Low	1 = low income area; 0 = other
Average	1 = average income area; 0 = other
High	1 = high income area; 0 = other
Race
Mostly non-White	1 = mostly non-White area, 0 = other
Integrated	1 = integrated area; 0 = other
Mostly	1 = mostly White area; 0 = other

Investigator variables capture years of experience and number of open cases. Given the prior literature on effective homicide investigation, these variables are meant to capture workload and adequacy of the investigation personnel. Years of experience is a continuous variable of the lead detective’s total number of years of experience as an investigator. Number of open cases is a continuous variable of the total number of cases the lead investigator had pending (i.e., open) since January 1 of the calendar year in which the homicide occurred. Number of cases was captured within calendar year instead of a rolling number of overall cases because although an investigator may keep a case on the books for a year, once all leads dry up, they stop actively investigating it, and instead shift their focus to new investigations.

The second set of variables tap into the victim devaluation and lifestyle perspectives and serve as proxies of social status and importance, which have been utilized in prior research to measure these concepts (Alderden & Lavery, 2007; Litwin, 2004; Litwin & Xu, 2007; Puckett & Lundman, 2003; Regoeczi, Jarvis, & Riedel, 2008; Roberts & Lyons, 2011; Wellford et al., 1999). These categorical variables include victim gender (female = 1, male = 0) and race (1 = Other, 0 = Black). In addition, age captures whether victims were younger (<14) or older (>44), which prior research has shown to increase the odds of an arrest (coded as 1), or were between the ages of 14 and 44 (coded as 0) (Alderden & Lavery, 2007; Litwin, 2004; Puckett & Lundman, 2003; Roberts, 2007). Finally, deviant lifestyle is modeled after Rydberg and Pizarro (2014) and captures the victim’s involvement in several measures of criminal lifestyle. A summative scale was created to approximate how enmeshed the victim was in a deviant lifestyle, which captured whether the victim was a drug dealer, was a gang member, and/or had a criminal record at the time of the homicide, resulting in a score ranging from 0 to 3, with higher values indicating higher levels of involvement in deviance.

The next set of variables tap into the event perspective, and focus on incident-level situational factors that have been shown in previous studies to impact the probability of clearance (Alderden & Lavery, 2007; Jiao, 2007; Litwin, 2004; Marché, 1994; Puckett & Lundman, 2003; Wellford et al., 1999). Weapon indicates whether a firearm or some other weapon was used (1 = no firearm; 0 = firearm). Mode captures how the homicide was carried out (1 = face-to-face, 0 = other) referring to incidents where the victim and offender were communicating and were aware of each other’s presence at the time of the incident. The witnesses variable (1 = yes, 0 = no) captures whether an incident had one or more witnesses to the incident who gave a statement to the police about what they witnessed. Finally, residence indicates whether the incident took place in the victim’s and/or offender’s residence or some other location (1 = yes, 0 = no).

The final set of variables tap into neighborhood characteristics, which are used as proxies to measure the police devaluation perspective. Similar to Puckett and Lundman (2003), two sets of dummy variables are used to measure census tract income and race. Income captures whether the average median household income in the tract is low, average, or high. Low income is coded if the average median household income is one standard deviation below the mean. Average is coded if the median household income is within one standard deviation above or below the mean. High income is coded if the median household income is one standard deviation above the mean. The same approach is used in the coding of census tract race composition, where dummy variables measure whether the census tract is mostly non-White, integrated, or mostly White. Mostly non-White is coded when the percentage of non-White is one standard deviation above the mean, integrated when the percentage of non-White is within one standard deviation of the mean, and mostly White when the percentage of non-White is one at least one standard deviation below the mean.

Findings

Table 3 presents descriptive statistics for the full 4-year study time frame, as well as the pre- and post-intervention periods. None of the variables included in the analyses have unknown missing values. Overall, of the 132 homicide incidents, 68% were cleared by an arrest during the study period. When examining pre- and post-intervention clearance specifically, 54% of incidents were cleared pre-intervention, while 83% were cleared post-intervention. Although over half of the cases were cleared within 7 days, on average, homicide incidents were cleared within 57 days as the average days to clearance is affected by the presence of various incidents that took over a year to clear. Similar to the examination of pre- and post-intervention specific clearance, time to clearance also varies pre- and post-intervention. On average, investigators cleared a case within 75 days pre-intervention and 44 days post-intervention. However, the average time to clearance of right censored cases increases to 510 days (841 days pre-intervention, 141 days post-intervention).

Table 3.

Descriptive and Bivariate Statistics.

Variable	All	Pre-intervention	Post-intervention
	(N = 132)	(n = 68)	(n = 64)
	M (SD)	M (SD)	M (SD)
Dependent
Cleared	0.68 (0.468)	0.54 (0.502)	0.83 (0.380)^a
Time to clearance	57.17 (161.160)	75.43 (226.263)	44.42 (92.775)
Independent
Intervention	0.48 (0.502)	—	—
Control
Investigator
Years of experience	15.39 (4.919)	15.87 (3.867)	14.89 (5.823)
Number of open cases	1.34 (1.377)	1.88 (1.588)	0.77 (0.850)
Victim
Gender	0.15 (0.360)	0.16 (0.371)	0.14 (0.350)
Race	0.23 (0.421)	0.22 (0.418)	0.23 (0.427)
Age	0.24 (0.430)	0.21 (0.407)	0.28 (0.453)
Deviant lifestyle	1.05 (0.790)	1.00 (0.792)	1.09 (0.791)
Event
Weapon	0.33 (0.470)	0.38 (0.490)	0.27 (0.445)
Mode	0.68 (0.468)	0.71 (0.459)	0.66 (0.479)
Witnesses	0.66 (0.476)	0.60 (0.493)	0.72 (0.453)
Residence	0.33 (0.473)	0.34 (0.477)	0.33 (0.473)
Neighborhood
Income
Low	0.13 (0.336)	0.12 (0.325)	0.14 (0.350)
Average	0.67 (0.470)	0.68 (0.471)	0.67 (0.473)
High	0.20 (0.399)	0.21 (0.407)	0.19 (0.393)
Race
Mostly Non-White	0.17 (0.381)	0.21 (0.407)	0.14 (0.350)
Integrated	0.68 (0.468)	0.66 (0.477)	0.70 (0.460)
Mostly White	0.14 (0.352)	0.13 (0.314)	0.16 (0.366)

Statistically significant difference between pre- and post-intervention; χ² = 12.258, p < .001.

Most victims were male (85%), African American (77%), and between the ages of 15 and 44 years of age (76%). Furthermore, homicides were most likely to be carried out with a firearm (67%), in a face-to-face manner (68%), and involve victims and offenders who knew each other at the time of the murder (64%). Moreover, most homicide incidents had at least one witness (66%) who gave a statement to the police, and the majority of incidents occurred somewhere other than the victim/offender residence (67%). Finally, the bulk of incidents occurred in average income and racially integrated census tracts.

Two sets of multivariate analyses were undertaken. The first examined if the change in investigation tactics had an effect on whether incidents were cleared with an arrest or not. Due to the dichotomous nature of the dependent measures, logistic regression was employed to identify the covariates that differentiate cleared versus not cleared cases. The advantage of using this analytic technique is that it tests the goodness-of-fit of the entire model and provides odds ratio calculations to determine the relative importance of each independent variable (Pampel, 2000). Changes in time to clearance pre- and post-intervention are also examined. In addressing time to clearance, a Cox proportional hazard model was utilized. Cox proportional hazard modeling is the optimal procedure for investigating time to clearance because it can incorporate observed information on the timing of clearance into parameter estimates, as well as differentiating between cleared and right-censored cases (i.e., those that remained open at the end of the data collection; Roberts, 2007; Rydberg & Pizarro, 2014).

Table 4 presents the results for the initial logistic regression analyses. Model 1 in Table 4 assesses whether the intervention is related to case clearance. Confirming the hypothesis, the intervention significantly affects case clearance, with cases occurring after the intervention having a higher probability of clearance. This model, however, only explains a modest amount of variation in the dependent variable (r² = .091), which is expected considering the model does not control for other theoretically relevant variables. Model 2 includes theoretically relevant control variables in addition to the intervention main independent variable. The purpose of this analysis is to assess if the intervention remains significant net of other theoretically relevant variables. As expected due to the number of variables included in the model, the model explains substantially more variation in the dependent variable (r² = .409). Supportive of the hypothesis, the intervention variable remains a significant covariate of clearance even after controlling for other relevant variables. The model shows that even after controlling for other variables, cases that occurred post-intervention had a 4.354 higher probability of being cleared.

Table 4.

Logistic Regression of Non-Cleared/Cleared Cases (N = 132).

Variable	Model 1			Model 2
Variable	B	SE	OR	B	SE	OR
Independent
Intervention	1.395***	.411	4.037	1.471*	0.621	4.354
Control
Investigator
Years of experience				−0.170*	0.066	0.844
Number of open cases				−0.580*	0.227	0.560
Victim
Gender				−0.597	0.849	0.551
Race				−0.515	0.683	0.598
Age				−0.292	0.773	0.747
Deviant lifestyle				−0.526	0.383	0.591
Event
Weapon				0.761	0.735	2.141
Mode				1.686**	0.646	5.399
Witnesses				1.719**	0.554	5.581
Residence				0.572	0.661	1.771
Neighborhood
Income
Average				−1.522	0.962	0.218
High				−0.816	1.082	0.442
Race
Integrated				−0.889	0.865	0.411
Mostly White				−1.027	1.110	0.358
Constant	0.177	.243	1.194	4.101*	1.976	6.404
Model χ²	12.659***			69.474***
−2 log likelihood	152.470			95.656
Cox & Snell R²	.091			.409

Note. OR = odds ratio.

p < .05. **p < .01. ***p < .001.

Several control variables are also significant in the predicted direction, although none of the variables that tap into the victim and police devaluation emerge as significant. Specifically, a greater number of investigator open cases decreased the likelihood of clearance, while incidents involving face-to-face homicides, and those where at least one witness was present and gave a police statement, were more likely to be cleared. Interestingly, investigator years of experience has an inverse relationship with clearance, whereas more experienced investigators were less likely to clear cases.

The results of the Cox regression models are displayed in Table 5. Similar to the logistic analyses, two models were estimated, one which only includes the intervention variable, and the second, a combination of intervention, investigator, victim, event, and neighborhood theoretically relevant variables. The interpretation of parameter estimates hazard ratios (HR) are presented and are interpreted analogously to odds ratios. HR over 1.00 indicate that the independent variable is associated with a homicide incident being cleared more quickly, and values under 1.00 indicate that the incident is cleared less quickly (Cox, 1972).

Table 5.

Cox Regression Time to Clearance (N = 132).

Variable	Model 1		Model 2
Variable	B	SE	B	SE
Independent
Intervention	−1.033***	.217	−0.975***	.244
Control
Investigator
Years of experience			−0.015	.019
Number of open cases			−0.156	.087
Victim
Gender			0.131	.289
Race			−0.277	.266
Age			0.682*	.291
Deviant lifestyle			−0.301*	.135
Event
Weapon			−0.585*	.246
Mode			−0.328	.221
Witnesses			−0.673***	.203
Residence			−0.491*	.235
Neighborhood
Income
Average			0.076	.306
High			0.213	.356
Race
Integrated			0.308	.270
Mostly White			0.158	.348
−2 log likelihood	1,013.283***		974.055***

p < .05. **p < .01. ***p < .001.

Model 1 only contains the intervention variable and indicates that cases post-intervention take significantly less time to clear. Model 2 introduces investigator, victim, event, and neighborhood characteristics indicators. Over and above the control measures, the intervention remains statistically significant, with cases post-intervention taking significantly less time to clear. Similar to the logistic models, neighborhood processes that tap into police devaluation are not significant; however, several variables that were not significant in the logistic model have a significant relationship with time to clearance.

Six control variables that measure a combination of victim devaluation and lifestyle, and event characteristics emerged as significant in Model 2. Homicides of 15 to 44 years old took longer to clear than those of other ages. Victim lifestyle is significant in the non-predicted direction, in that victims enmeshed in criminal lifestyles were cleared faster than those of victims not involved in deviant and criminal activities. Finally, consistent with prior research, homicides carried out with non-firearms were cleared faster, as well as those where there was at least one witness and that occurred in the victim’s and/or offender’s residence.

Discussion

The current study sets out to examine the effect of investigation tactics on homicide clearance net of theoretically significant variables that tap into victim devaluation, event, police devaluation, and victim lifestyle. This study is one of the few that simultaneously examines the robustness of the four clearance explanations explored in the literature. It is also one of the few natural experiments examining the effect of investigation tactics on clearance with a pre- and post-intervention time.

It was hypothesized that investigation tactics will affect the odds of clearance, even after controlling for other theoretically relevant variables. The findings presented here support this hypothesis and suggest that the approaches and tactics police departments implement within their investigative units are indeed important. The Rochester Homicide Unit increased staffing and management oversight, and become more efficient in their assignment of tasks immediately after the homicide was reported due to their addition of the mobile command RV. Following these changes, homicide clearance rates increased, and the days-to-clearance decreased showing that these shifts had a positive effect. These findings support previous research that show adequate staffing with competent, well-trained investigators, proper oversight by management, and the immediacy in which tasks are distributed are important covariates in homicide clearance (Carter & Carter, 2016; Horvath et al., 2001; Wellford et al., 1999). The results also underscore the importance of the investigative efforts undertaken within the “first 48-hr” period of the investigation.

Other important findings also emerged from the analyses. First, the findings presented here join prior works that show event characteristics are important. Incidents carried out in a face-to-face manner, in the victim’s and/or offender’s residence, and those involving weapons other than firearms were more likely to be cleared and cleared faster. This speaks to the possibility that these cases are easier to investigate given the presence of physical evidence. The findings also support prior research that witness cooperation in the form of providing statements to the police also has a significant effect on the clearance of cases (Carter & Carter, 2016; Keel et al., 2009; Mancik et al., 2018; Puckett & Lundman, 2003); hence, tactics that foster these practices also have the potential to be helpful. Interestingly, the significance of age may be more related to theoretically relevant event factors than to victim devaluation. Specifically, incidents involving very young or very old victims often occur in the victim/offender’s residence, and incident location (i.e., residence and inside locations) is positively related to clearance. Although, not examined here, these age groups are also more likely to be victimized by an individual known to them (in many cases, family members), and prior research has found that known victim–offender relationships are positively related with clearance (Rydberg & Pizarro, 2014).

Prior research suggests that investigator experience is unrelated to homicide clearance, as all homicide detectives are typically experienced police officers. Interestingly, the results show that cases post-intervention were more likely to be cleared by investigators with less experience. This paradoxical result may be explained by the fact that seven less experienced investigators transferred to MCU in 2012. These investigators, while having less experience, were more familiar and comfortable using investigative tools that had not been extensively utilized in MCU prior to their arrival. These tools included the use of search warrants to obtain historical cell phone data (including cell tower information) and the content of cell phones seized, scrutinizing social media (e.g., Facebook and Instagram), creating fake social media accounts to communicate with suspects anonymously, and canvassing for video. These tasks often were neglected prior to 2012 because the two case investigators were solely responsible for all investigative steps and also had a larger caseload. Beginning in 2012, more of a team approach was adopted which not only placed more importance on the use of cell phone technology, reviewing social media, and collecting video, but which “farmed out” these time-consuming investigative tasks to other team members and freed up the primary investigators for other important tasks.

Contrary to Rydberg and Pizarro (2014), cases involving victims more deeply enmeshed in a deviant lifestyle took less time to clear. It may be that in those cases, other variables such as the presence of non-involved witnesses, or the availability of video and cell phone GPS evidence, offset the effect of deviant lifestyle. Investigators in Rochester note that cases involving deviant victims are sometimes solved with the assistance of witnesses who are themselves involved in a deviant lifestyle. These witnesses, who are typically friends or associates of the victim or offender, will usually cooperate in a homicide investigation only after they are arrested for an unrelated offense, and agree to testify in exchange for a lighter sentence. This is very similar to what Rydberg and Pizarro (2014) found in their study of clearance in Newark, New Jersey.

Taken as a whole, the findings presented here suggest that investigation tactics matter in clearance. This has implications for theory, as it suggests that the victim devaluation, event, police devaluation, and lifestyle theoretical perspectives do not offer the only valid explanations for why homicides are solved. While important, these perspectives do not take into account the dynamic nature of police work and how the police go about investigating homicides, which is an important piece of the clearance framework. The findings show that homicide clearance is complex, involving multiple dynamics which may leave a case open on an agency’s books, or culminate in the apprehension of a murder suspect. As a result, to gain a true, more complete understanding of homicide investigation outcomes, multiple theoretical frameworks should be combined in the assessment of the covariates of clearance in future studies.

We would be remiss to ignore that the effects presented here may be a product of broader unrelated dynamics. Indeed, the findings may simply be a product of broader ecological effects and population shifts in the city of Rochester. While this may be a possibility, an examination of the events that occurred in the city and overall New York State suggests that this is not the case. There were no regional or state mandates/policy changes impacting homicide investigations in New York State during the study period, and to the authors’ knowledge, there were no policy changes in nearby jurisdictions. This suggests that the investigation tactic effects found in the present study do not appear to be the product of any broader systematic policy change, nor of diffusion of benefits from neighboring jurisdictions. In addition, while Rochester’s population decreased by roughly 9,000 residents from 2000 to 2010, Rochester’s population was stable during the 4 years of the study period. Given these observations and the findings of recent research (Braga & Dusseault, 2018; Carter & Carter, 2016) that have demonstrated the importance of investigation tactics, it is plausible that the increase in Rochester clearance rates during the post-intervention period was causally related to the change in investigative tactics and procedures implemented. Future research should further explore the findings presented here in light of clearance rates in nearby jurisdictions and other ecological variables.

Despite the important contributions, this study is not without limitation. Rochester is a relatively small, poor city with a relatively high murder rate per capita. The findings here may not be generalizable to cities with less homicide and poverty. In addition, the intervention here was treated as one nominal independent variable, without its component parts being parsed, coded, and measured for their respective contribution to case outcomes. As such, the resources, practices, and procedures of the MCU cannot be compared quantitatively. Future research should parse out the various investigative tactics in an effort to assess which are more influential. Given that the findings presented here suggest a heightened importance of tactics that focus on the reliance of digital technological advances such as video and cell phone data analyses, researchers should also explore the utility of these tactics. Relatedly, Carter and Carter (2016) suggested that future research should attempt to quantify effective tactics to uncover and account for variations in agency and neighborhood types. Quantifying tactics and procedures may not only allow for better comparisons between agencies with differing clearance rates but also help illuminate the tactics and procedures that matter most, such that agencies with limited budgets could adopt those which are most cost-effective.

Future research should also attempt to replicate the results of this study in communities that are demographically different than Rochester, paying particular attention to whether victim lifestyle and officer experience produce the same counterintuitive findings seen here. Finally, qualitative studies in which researchers are embedded within homicide units, and observe how and what detectives do from the initial call to the scene to the arrest of the offender, are sorely lacking and much needed to expand and amplify our knowledge of police tactics and procedures, and thereby “render [them] less opaque” (Puckett & Lundman, 2003, p. 178). Gaining the trust and confidence of experienced police officers is not easy, but efforts to do so, if successful, would provide another avenue to uncover the nexus between investigator performance and homicide clearance.

In conclusion, this study suggests that police investigation procedures and tactics should be viewed as an important perspective in the homicide clearance literature. Clearly, the manner in which police investigate homicides is relevant to the outcome of the investigation. In terms of its practical significance, this study also suggests that, in the world of policing, police departments can improve their homicide clearance rates by adopting some combination of the procedures and tactics described here.

Footnotes

Declaration of Conflicting Interests

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

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

Notes

Author Biographies

Jesenia M. Pizarro is an associate professor in the School of Criminology and Criminal Justice at Arizona State University. Her research focuses on the situational covariates of homicide and the reactions to this crime.

William Terrill is a professor in the School of Criminology and Criminal Justice at Arizona State University. His research centers on police behavior, with an emphasis on police use of force and police culture.

Charles A. LoFaso is a doctoral student in the Ohio State University Sociology Department. He earned a JD from the University at Buffalo Law School, State University of New York, an MS in criminal justice from Michigan State University, and an MA from the Ohio State University. Prior to enrolling in the Ohio State University, he was a member of the Rochester Police Department for 17 years, a detective for 11 and a homicide detective for 4 years. His research interests include homicide and policing.

References

Alderden

M. A.

Lavery

T. A.

(2007). Predicting homicide clearances in Chicago: Investigating disparities in predictors across different types of homicide. Homicide Studies, 11, 115-132.

Armour

M. P.

(2002). Journey of family members of homicide victims: A qualitative study of their post-homicide experience. American Journal of Orthopsychiatry, 72, 372-382.

Black

(1976). The behavior of law. Bingley, UK: Emerald Group Publishing.

Borg

M. J.

Parker

K. F.

(2001). Mobilizing law in urban areas: The social structure of homicide clearance rates. Law & Society Review, 35, 435-466. doi:10.2307/3185409

Braga

A. A.

Dusseault

(2018). Can homicide detectives improve homicide clearance rates? Crime & Delinquency, 64, 283-315. doi:10.1177/0011128716679164

Carter

D. L.

Carter

J. G.

(2016). Effective police homicide investigations: Evidence from seven cities with high clearance rates. Homicide Studies, 20, 150-176.

Chaiken

J.M.

Greenwood

P.W.

Petersilia

(1976). The criminal investigation process: a summary report [online]. Santa Monica, CA: Rand Corporation. Available from: http://www.rand.org/content/dam/rand/pubs/papers/2006/P5628–1.pdf

Cloninger

D. O.

Sartorius

L. C.

(1979). Crime rates, clearance rates and enforcement effort: The case of Houston, Texas. American Journal of Economics and Sociology, 38, 389-402.

Cox

D. R.

(1972). Regression models and life tables. Journal of the Royal Statistical Society, 34, 187-220.

10.

Cronin

J. M.

Murphy

G. R.

Spahr

L. L.

Toliver

J. I.

Weger

R. E.

(2007). Promoting effective homicide investigations. Community Oriented Policing Services, U.S. Department of Justice. Washington, D.C.

11.

DeLisi

Kosloski

Sween

Hachmeister

Moore

Drury

(2010). Murder by numbers: Monetary costs imposed by a sample of homicide offenders. The Journal of Forensic Psychiatry & Psychology, 21, 501-513. doi:10.1080/14789940903564388

12.

Eck

J. E.

(1992). Criminal investigation. In Cordner

Hale

(Eds.), What works in policing? Operations and administration examined (pp. 19-34). Cincinnati, OH: Anderson.

13.

Federal Bureau of Investigation (2013). Crime in the United States 2013, Retrieved from https://www.fbi.gov/about-us/cjis/ucr/crime-in-the-u.s/2013/crime-in-the-u.s.-2013/violent-crime/murder-topic-page/murdermain_final

14.

Geberth

V. J.

(1990). Practical homicide investigation: Tactics, procedures, and forensic techniques. New York, NY: Elsevier.

15.

Hawk

S. R.

Dabney

D. A.

(2014). Are all cases treated equal? Using Goffman’s frame analysis to understand how homicide detectives orient to their work. The British Journal of Criminology, 54, 1129-1147. doi:10.1093/bjc/azu056

16.

Horvath

Meesig

R. T.

Lee

Y. H.

(2001). A national survey of police policies and practices regarding the criminal investigation process: Twenty-five years after Rand. Washington, DC: National Institute of Justice.

17.

Hosmer

D. G.

Lemeshow

May

. (2008). Applied survival analysis: Regression modeling of time-to-event data. Hoboken, NJ: John Wiley.

18.

Hough

McCorkle

(2016). American homicide. Thousand Oaks, CA: SAGE.

19.

Jiao

A. Y.

(2007). Explaining homicide clearance: An analysis of Chicago homicide data 1965–1995. Criminal Justice Studies, 20, 3-14.

20.

Keel

T. G.

Jarvis

J. P.

Muirhead

Y. E.

(2009). An exploratory analysis of factors affecting homicide investigations: Examining the dynamics of murder clearance rates. Homicide Studies, 13, 50-68.

21.

Klinger

D. A.

(1997). Negotiating order in patrol work: An ecological theory of police response to deviance. Criminology, 35, 277-306.

22.

Lee

(2005). The value of life in death: Multiple regression and event history analyses of homicide clearance in Los Angeles County. Journal of Criminal Justice, 33, 527-534.

23.

Litwin

K. J.

(2004). A multilevel multivariate analysis of factors affecting homicide clearances. Journal of Research in Crime and Delinquency, 41, 327-351.

24.

Litwin

K. J.

(2007). The dynamic nature of homicide clearances: A multilevel model comparison of three time periods. Homicide Studies, 11, 94-114.

25.

Mancik

A. M.

Parker

K. F.

Williams

K. R.

(2018). Neighborhood context and homicide clearance: Estimating the effects of collective efficacy. Homicide Studies, 22, 188-213.

26.

Marché

G. E.

(1994). The production of homicide solutions. American Journal of Economics and Sociology, 53, 385-401.

27.

McEwen

(2011). The role and impact of forensic evidence in the criminal justice system, final report. Washington, DC: U.S. Department of Justice.

28.

McEwen

(2013). Executive summary evaluation of the phoenix homicide clearance project. Washington, DC: U.S. Department of Justice.

29.

Miletich

J. J.

(2003). Homicide investigation: An introduction. Lanham, MD: Scarecrow Press.

30.

Ousey

G. C.

Lee

M. R.

(2009). To know the unknown: The decline in homicide clearance rates, 1980–2000. Criminal Justice Review, 35, 141-158. doi:10.1177/0734016809348360

31.

Pampel

F. C.

(2000). Logistic regression: A primer. Thousand Oaks, CA: SAGE.

32.

Petersen

(2017). Neighbourhood context and unsolved murders: The social ecology of homicide investigations. Policing and Society, 27, 372-392.

33.

Pizarro

J. M.

(2017). Gang homicides in the United States: What we know and future research directions. In Brookman

Maguire

E. R.

Maguire

(Eds.), The handbook of homicide (pp. 73-88). Hoboken, NJ: Wiley Blackwell.

34.

Puckett

J. L.

Lundman

R. J.

(2003). Factors affecting homicide clearances: Multivariate analysis of a more complete conceptual framework. Journal of Research in Crime and Delinquency, 40, 171-193.

35.

Regoeczi

W. C.

Jarvis

J. P.

(2013). Beyond the social production of homicide rates: Extending social disorganization theory to explain homicide case outcomes. Justice Quarterly, 30, 983-1014.

36.

Regoeczi

W. C.

Jarvis

J. P.

Riedel

(2008). Clearing murders: Is it about time? Journal of Research in Crime and Delinquency, 45, 142-162.

37.

Riedel

(2008). Homicide arrest clearances: A review of the literature. Sociology Compass, 2, 1145-1164. doi:10.1111/j.1751-9020.2008.00130.x

38.

Roberts

(2007). Predictors of homicide clearance by arrest: An event history analysis of NIBRS incidents. Homicide Studies, 11, 82-93.

39.

Roberts

Lyons

C. J.

(2011). Hispanic victims and homicide clearance by arrest. Homicide Studies, 15, 48-73. doi:10.1177/1088767910397278

40.

Rydberg

Pizarro

J. M.

(2014). Victim lifestyle as a correlate of homicide clearance. Homicide Studies, 18, 342-362.

41.

Schroeder

D. A.

White

M. D.

(2009). Exploring the use of DNA evidence in homicide investigations: Implications for detective work and case clearance. Police Quarterly, 12, 319-342.

42.

Stelfox

(2011). Criminal investigation: Filling the skills gap in leadership, management, and supervision. Policing: A Journal of Policy and Practice, 5, 15-22. doi:10.1093/police/paq055

43.

Wellford

C. F.

Cronin

J. M.

Brandl

Bynum

Eversen

Galeria

(1999). An analysis of variables affecting the clearance of homicides: A multistate study. Washington, DC: Justice Research and Statistics Association.

44.

Worrall

J. L.

(2016). Investigative resources and crime clearances: A group-based trajectory approach. Criminal Justice Policy Review, 30, 155-175.

45.

(2008). Characteristics of homicide events and the decline in homicide clearance: A longitudinal approach to the dynamic relationship, Chicago 1966-1995. Criminal Justice Review, 33, 453-479. doi:10.1177/0734016808320693