A Case Control Study of Non-Domestic Solved and Un-Solved Homicides in Trinidad and Tobago

Table of Contents

Abstract

Acknowledgements

List of Figures

List of Tables

List of Abbreviations

Introduction
Overview
Why is this Study Important?
Thesis Outline

Literature Review
Introduction
Definition of Homicide Detection Rate
Homicide Detection Rates as a Measure of Performance
Measurement Issue about Detection Rates
Theoretical Frameworks on Homicide Detection
Reasons for Decline in Homicide Detection Rates
Changes in the nature of homicides
Changes in the nature of police resources
Changes in the Behavior of Witnesses
The Investigation of Homicide
Forensic Evidence in Homicide Investigations
Political Influence on Homicide Investigation
Case Characteristics and Solvability Factors
The Wellford and Cronin (1999) Study
Summary of Literature Review

Methods
Introduction
Research setting
Definitions
The Choice of Research Design
Data Sources
Data Extraction, Cleaning and Exclusion Rules
Data Limitations and Issues
Other Limitations of the Data
Selection of the Case Control
Investigative and Solvability Factors Checklists
Populating the Checklists
Data Analysis
Data Protection and Ethics
Summary

Results
Introduction
Description of the Data
Non-Domestic Homicides by Motive
Spatial Distribution
Average Geographic Detection Rates
Temporal Distribution
Homicide Victims by Ethnicity
Homicides by victims per homicide
Homicide by offenders per homicide
Homicide by Weapon Type
Frequency of Investigative Variables
Frequency of the Solvability Factors

META-ANALYSIS RESULTS
Investigative Variables
Forrest Plot of Investigative Variables
Solvability Factors
Forest Plot of Solvability Factors
Results of Latent Class Analysis
Summary of Results
Discussion
Introduction

INVESTIGATIVE VARIABLES
Positive Association
Negative Association
Uncertain
Overall Results

SOLVABILITY FACTORS
Positive Association
Negative Association
Uncertain
Latent Classes
Other Considerations
Implications
Recommendations for Future Research
Limitations of the study
Summary of Discussion

Conclusion

References

Abstract

A case-control research design was used in this study with the aim of identifying the factors which differentiate solved and unsolved non-domestic homicides in Trinidad and Tobago for the seven year period 2008-2014. Two research questions guided the study:

1. Is the use of elements of a structured approach to investigate homicides associated with more solved cases?
2. Are certain characteristics of homicide cases associated with a higher likelihood of detection?

The data set for this study comprised all the non-domestic homicides reported to the Trinidad and Tobago Police Service from January 1st 2008 to December 31st 2014 was analyzed against 21 investigative variables and 16 solvability factors. Of these, only 2 investigative variables and 9 solvability factors were found to be strongly associated with solved cases. The two investigative variables were (a) 1st officer secured crime scene; and(b) investigator present at post-mortem. This study confirmed previous research findings in identifying the following variables as solvability factors (a) weapon recovered; (b) suspect named, (c) projectile recovered, (d) offender vehicle identified, (e) fingerprint found, (f) eye-witness (1vs. 0), (g) ballistic on weapon found, (h) 3 investigators versus 1, and (i) 4 investigators versus 1. The magnitude of these effects, measured with an odds-ratio meta-analysis suggested that the presence of these variables is positively associated with solved cases.

A model with three latent classes was found to best describe both the investigative variable and solvability factor variables. Based on the variation in the observed distribution, specifically the ratio of solved to unsolved cases, the classes were given thelabels “Easy Cases”, “Normal Cases” and “Difficult Cases”. A similar variation was not observed in the detection status for the investigative variables - the proportions of solved to unsolved for each class were equally 50%. Consequently those results were discarded as they provided no useful insight into the homicide cases.

Key Words: Non-domestic homicides, case-control research design, odds ratio metaanalysis, latent class analysis, investigative variables, solvability factors.

Acknowledgements

I would like to express my heartfelt appreciation to my joint supervisors Dr. Katrin Mueller-Johnson and Mr. John Parkinson for their valuable and constructive insight and guidance throughout the planning and subsequent development of this research work. Special thanks also to the entire academic faculty at the Institute of Criminology for their inspiring lectures and in particular Professor Lawrence Sherman for providing me with the opportunity to attend this prestigious institution.

I wish to also thank Commissioner Stephen Williams who facilitated my time away from work to attend all of my residential blocks. To Mrs. Lucinda Bowditch and Mr. Glenn Garner your handling of my administrative matters created a feeling of comfort. I am hugely indebted to Mr. Kenrick Antoine of the Special Branch department whose moral support and encouragement kept me focused throughout the entire journey.

Finally I would like to thank my wife Janel for her patience, forbearance, sacrifice and unfaltering support throughout this journey. She was my source of inspiration when things became difficult.

Andre A Norton

List of Figures

Figure 2. 1 Statistically significant investigative and solvability variables (Wellford and Cronin, 1999)

Figure 4. 1: Non-Domestic Homicides in Trinidad & Tobago (2008-2014)

Figure 4. 2: Percentage of non-domestic homicides by division

Figure 4. 3: Average homicide detection rate by division (2008-2014)

Figure 4. 4: Frequency of Homicides by Time of Day

Figure 4. 5: Percentage of Homicides by Day of the Week

Figure 4. 6: Homicide Victims by Ethnicity

Figure 4. 7: Percentage homicide offenders by ethnicity

Figure 4. 8: Percentage of homicides by victim per homicide

Figure 4. 9: Percentage homicide by offender’s per-homicide

Figure 4. 10: Homicides by weapon type and by means committed

Figure 4. 11: Graph of frequency on investigative variables in solved and unsolved homicides

Figure 4. 12: Frequency of solvability factors Bar-graph

Figure 4. 13: Forest plot of the meta-analysis results of the investigative variables

Figure 4. 14: Forest plot of the meta-analysis results of the solvability factors

Figure 4. 15: Detection status for the homicide cases belonging to the different classes of the solvability factors latent class model

Figure 4. 16: Overview of solvability factors for each class, by detection status of the corresponding cases in each class

List of Tables

Table 3.1 Check-list of investigative variables (adopted from Wellford and Cronin, 1999)

Table 3.2: Checklist of solvability factors (adopted from Wellford and Cronin, 1999)...

Table 4.1: Non-Domestic homicide by motive in Trinidad & Tobago (2008-2014)

Table 4.2: Frequency of investigative variables in solved (cases) and unsolved (controls) homicides

Table 4.3: Frequency of solvability factors present in (cases) solved and (controls) unsolved homicides

Table 4.4: Mean odds-ratio values of investigative variables

Table 4.5: Mean odds-ratio values of solvability factors

List of Abbreviations

illustration not visible in this excerpt

Introduction

Overview

While occupying a unique place in the consciousness of every citizen, homicide unleashes horror and fear in communities whenever it occurs and citizens expect the police to do all within their power to apprehend offenders (Schramm, 2001). Arresting an offender is one of the most important events in the criminal justice system because in its absence none of the accepted forms of punishment can be applied (Wellford & Cronin, 1999). One of the hallmarks of unsolved homicides is the fact that offenders remain at large to commit more crimes or become victims through retaliation (Wellford and Cronin, 1999).

Solving homicides is fundamental to one of the underlying principles of law enforcement known as deterrence (Beccaria, 1963; Bottoms and Von Hirsch, 2010). Deterrence theory predicts that when the perceived probability of being caught is high, unacceptable behavior is less likely to occur (Bottoms and Von Hirsch, 2010). Socially desirable behavior is more likely to occur if the certainty of getting caught for non- compliance is intensified (Von Hirsch et al., 1999). Therefore a low homicide detection rate produces low certainty of capturing offenders and signals law enforcement’s inability to prevent future homicides. Low detection rates also impacts the contiguity of sanctions being administered to offenders (certainty) since less persons will be prosecuted.

Unsolved homicides hamper the healing process of families and friends of victims and often lead to public dissatisfaction that can negatively affect police legitimacy especially when citizens lose trust and confidence in the moral authority of police to exercise their legitimate power (Tankebe, 2008).

Why is this Study Important?

During the period 1st January 2008 to 31st December 2014, Trinidad and Tobago (TT) experienced high homicide rates which averaged around 400 annually with a peak of 547 occurrences in 2008. This was beyond the normal threshold and represented a ‘tipping point’ for local law enforcement. High numbers of annual homicides attracted wide media attention both at home and abroad and resulted in the online postings of negative travel advisories deterring tourists from choosing TT as a potential destination.

Locally, the central Government identified the task of reducing homicides as a priority and at the beginning of 2013 mandated the Commissioner of Police (COP) to develop solutions to the growing homicide problem. In response, the COP operationalized several crime reduction initiatives. Some of these included significant investments in training, multi-agency arrangements and specialist units, restructuring and increasing the manpower of the Homicide Bureau of Investigations (HBI) and an island-wide implementation of hotspots patrols in homicide hotspots.

Despite the genuine efforts of the TTPS to effectively reduce homicides and other violent crimes and improve detection rates through these initiatives, the organization continues to struggle with high homicide and low detection rates. The latter has stubbornly remained low at around 16%. However, despite not achieving success at effectively reducing homicides, other serious crime such as robberies (10%), burglaries (15%) and larcenies (12%) have experienced minor reductions (CAPA, 2014).

The United Nations Office on Drugs and Crime (UNODC) and the World Health Organization (WHO) 2014 global status report on violence prevention ranked TT as the 10th most violent country in the world based on a murder per capita ratio for the 403 local homicides occurring that year (UNODC/WHO, 2014). The data used in the compilation of this report was obtained from 133 countries representative of 6.1 billion people or 88% of the global population (UNODC/WHO, 2015).

In May and November 2014 respectively, Trinidad and Tobago experienced the tragic murders of eminent attorney-at-law Dana Seetahal, and an elderly German couple, Hubertus and Birgit Kiel (both 70+ years old). Both cases drew sustained media attention that elevated the public profile of these cases and resulted in a large portion of the local homicide bureau’s resources being assigned to the investigations On the aftermath of both murders, the level of public concern about homicides is potentially driving an ever thickening wedge between police and citizens as evidenced by continuous public expressions of dissatisfaction with the agency’s perceived inability to solve crimes, especially homicides. In the Dana Seetahal incident, there were continuous calls coming from family members and friends for the case to be solved. A local crime stoppers anonymous tip line agency offered $TT3.5 million as an incentive for providing information leading to the arrest of the offender(s).

Approximately fifteen months after Seetahal’s murder, eleven persons were eventually arrested and charged for the crime. However, this did not bring solace to her family members. Her brother, Omar Seetahal, speculated that the eleven persons charged had no motive for wanting to kill his sister and questioned the timing of the arrests. He felt that TTPS’ celebration of having solved the case might be premature and further speculated whether his sister’s case was being used by TTPS to boost its dwindling image in lieu of justice for his sister’s death (Trinidad Express Newspaper, 2015). He holds the view that more needs to be investigated.

To date, the Kiel double murder remains unsolved despite assistance given to local homicide investigators by two members of the German police agency and a forensic pathologist who volunteered assistance on the request of the German government as part of a bi-lateral agreement with Trinidad and Tobago (Fraser, 2014 [Associated Press article]).

To the best of this researcher’s knowledge, there has been no homicide research done to date which has analyzed the factors which differentiate solved and unsolved nondomestic homicides within a single jurisdiction. All of the studies carried out thus far have included domestic-related homicides in their analysis. This thesis therefore seeks to fill the gap in the homicide literature and thereby add to the body of knowledge of criminological research on homicide solvability.

An odds ratio meta-analysis of investigative variables and solvability factors including their summary effects will be used to measure the magnitude of the differences between them in both the solved and unsolved homicides in terms of the presence of these variables in both groups. It is hoped that this approach will allow for identifying those variables with strongest predictive capability for solving homicides which may be later tested.

Thesis Outline

This thesis is presented in six chapters. After this introductory chapter, a review is carried out on the existing literature on homicide relevant to this research which also identifies the limitations of the research to date. The methods chapter sets out the two research questions and outlines the proposed research methodology for identifying, acquiring and analyzing homicide data. The fourth section provides a description of the data to place the results into context and also presents the results of the odds ratio meta- analysis. The discussion section provides comments on the research findings with particular reference to existing research and their implications for policing. Everything is then drawn together in the concluding chapter whereby a progressive and reasoned conclusion is presented.

Literature Review

Introduction

When considering the literature in relation to homicides, there is a considerable amount of research to employ. Though the scope of these literatures covers a wide range of topics, in order to tackle the declared aims of this current study, this chapter will focus primarily on three major themes, namely: homicide detection rates, the homicide investigative process and the characteristics of homicide cases and their investigation which favors detection (solvability factors).

In order to achieve the intended goal three things will be done: an exploration of previous research on homicide detection rates and its offshoots, an examination of how police departments in developed countries successfully investigate homicides and an analysis of the characteristics of homicide cases and their investigation thereby increasing the likelihood of more solved cases.

By conducting research into how homicides are investigated in first world countries as opposed to what is being done by the Trinidad and Tobago Police Service, this researcher hopes to: identify and propose solutions to problematic issues, copy successful strategies, methods and actions and avoid failed ones. These recommendations could form the basis of a new approach to investigating homicides in Trinidad and Tobago and begin the process of solving more homicide cases thereby increasing the existing homicide detection rates which have remained stably low at around 16% over the past seven years.

However, it is important to note that because the three themes interconnect with each other there would be interplay between them throughout the literature review.

Definition of Homicide Detection Rate

The homicide detection rate is calculated by dividing the number of homicides that are solved in a calendar year (1st January - 31st December) by the number of homicides which were reported in that year (Greenwood et al., 1977; FBI, 2004). For the purpose of this review the term detection rate will be used instead of clearance rate despite the latter term being used more frequently in published literature.

Homicide Detection Rates as a Measure of Performance

Homicide detection rates have often been used as a yardstick to measure overall police performance (Maguire et al., 2010). Solving a crime represents the most clear-cut quantifiable indicator of police effectiveness (Litwin and Xu, 2007). Simon (1991:39) also noted that “in homicide, the detection rate is the litmus test, the beginning and end of all debate”. Various analytical techniques have been used in detection rate studies which include individual, city level and multi-level analyses. The majority of the quantitative studies devoted to homicide detection rates have employed a multivariate approach using solved homicide as the dependent variable and case and victim characteristics as the predictor variables (Maguire et al., 2010). Logistic regression has frequently been used to explore relationships between predictor variables and case detection as evidenced in a study by Alderden and Lavery (2007).

Measurement Issue about Detection Rates

Despite popularity in the use of homicide detection rates as a tool for measuring overall police performance, there are those who consider it to be a biased and inadequate measure (Greenwood et al., 1977; Regeoczi et al., 2000; Litwin, 2004). Greenwood et al (1977) argued against its use because of the varying methods used by police agencies to define and record detections. The inadequacy arises when a crime from the previous year is solved in the current year and added to the detection rate of the latter.

Maguire et al (2010) believes that arriving at firm conclusions about factors related to homicide detections is problematic in part because homicide detection rate studies have used different levels of analysis, examined different jurisdictions and utilized differing analytical and exploratory frameworks. The severity in measurement errors in official police data also creates difficulty in determining whether observed increases and decreases in crime reflect true changes or mere artefacts of recording and reporting practices (Hoffman, 1971; Nadel, 1978; Poggio et al., 1985; Cordner, 1989, Alpert and Moore, 1993; Riedel and Jarvis, 1998). Despite dissenting views among researchers of the use of homicide detection rates as a barometer for police performance, it continues to be the most repeatedly used measurement (Maguire et al., 2010).

Theoretical Frameworks on Homicide Detection

Two competing frameworks have been advanced by scholars regarding the factors affecting detection in homicide cases, with the role of extralegal characteristics being pivotal to the debate (Schroeder and White, 2009). The first perspective known as the discretionary perspective (extra-legal characteristics) — based on Quinney’s (1977) conflict theory and Black’s (1976) behavior of law theory — alludes to the notion that police use discretion when investigating homicides and the degree of effort put into these investigations are influenced by certain extralegal characteristics of victims such as their ethnicity and social status (see also Puckett and Lundman, 2003; Litwin, 2004). This means that police investigate homicides involving victims with lower societal status less vigorously (Roberts, 2007).

The second framework known as the non-discretionary (solvability) perspective purports the opposite, viz, that because of the seriousness of homicides (Gottfredson and Hinderlang, 1979; Meyers, 1997) and the pressures exerted on detective to solve cases (Puckett and Lundman, 2003; Simon, 1991), all incidents receive vigorous efforts by investigators notwithstanding the extralegal characteristics of the victims involved (Roberts, 2007). Exponents of this framework include Klinger (1997) and Wolfgang (1958).

Previous research has consistently supported the non-discretionary framework (full investigation framework) underscoring the characteristics of homicide incidents and how police respond to these incidents as key predictors of detection (Roberts, 2007).

Reasons for Decline in Homicide Detection Rates

Homicide detection rates vary across jurisdictions and over time. The US national homicide detection rate dropped from 90% in 1965 to 64% in 2008 (FBI, 2008). Commenting on this decline, Fox (2000) stated that there is little likelihood of detection rates returning to 1960’s values. The majority of research on detection rates in the English language has been conducted in the US with a handful of other studies in a few other developed countries; for example Mouzos and Muller (2001) exploring solvability factors of homicides in Australia, Innes (2003) examining homicide detection rates in England and Wales, and two cross national studies comparing the US and Canada (Regeoczi et al., 2000) and the US and Japan (Roberts, 2008).

Despite little consensus, the researchers who have speculated about reasons for the overall decline in homicide detection rates have offered three distinct explanations for these declines —changes in the nature of homicides, changes in the nature of police resources and changes in the behavior of witnesses (Riedel and Rinehart, 1994; Caldarella and Cavanaugh, 1992).

Changes in the nature of homicides

Homicides have traditionally been viewed as crimes of passion involving intimate partners (husband and wife) or family members and have also been referred to as acquaintance homicides (Mouzos and Muller, 2001). In this type, the victim and offender are known to each other in some way before the crime occurred which usually results from either an argument or some other personal conflict between the victim and the offender (Richardson and Kosa, 2001).

Puckett and Lundman (2003) described these crimes as self-solvers (dunkers) because they usually have an obvious suspect and sufficient evidence to arrest the suspect (e.g. a husband calling the police and confessing to having killed his wife). These offences are usually characterized by substantial amounts of forensic evidence as well as witnesses who can corroborate elements of the offender’s story. In these circumstances the basic evidence required to solve these offences are present in the initial investigation and very little investigative effort is needed (Puckett and Lundman, 2003). The social nature of these ‘self-solvers’ resulted in high rates of detection for these offences.

Research has shown that over the past few decades there has been a shift away from acquaintance homicides toward an increase in stranger to stranger homicides (Richardson and Kosa, 2001). A stranger to stranger homicide is a type of homicide in which the victim and offender were unknown to each other prior to the crime occurring. Gilbert (1983) speculated that the increase in stranger to stranger homicides was linked to the proliferation of guns and drugs - two social stressors which Richardson and Kosa (2001) believe have contributed to the decline in homicide detection rates. Puckett and Lundman (2003) described stranger to stranger homicides as ‘whodunits’ or general mysteries (e.g. a detective is summoned to visit a homicide crime scene at some godforsaken location without any other information). These offences are more likely to be unwitnessed thereby resulting in offenders having to be selected from a much wider population of individuals when compared to domestic homicides (Richardson and Kosa, 2001). The absence of a pre-crime victim-offender relationship increases the difficulty in discerning a motive for the killing and if also linked to the illegal drug market these offences become less likely to be solved since there is greater difficulty in identifying suspects (Riedel and Rinehart, 1994; Caldarella and Cavanaugh, 1992). A US-Canada cross national study (Regeoczi et al., 2000) attributed the increase in stranger to stranger homicides to the widespread availability of guns and propensity of blacks and native minorities to engage in violence.

The 2005 International Association of Chief Police Officers (IACP) Murder Summit concluded that the decline in detection rates was as a result of an increase in stranger to stranger homicides, drug-related homicides, gang-related homicides and the availability of handguns. Gilbert’s (1983) study of homicide trends in San Diego between 1970 and 1980 also attributed the decline in homicide detection rates to the increase in stranger to stranger homicides.

Changes in the nature of police resources

The ability of police agencies to devote substantial and experienced detectives to investigations has diminished and has changed ways in which police agencies respond to crimes resulting in a negative impact on detection rates (Riedel and Rinehart, 1994; Cardarelli and Cavanaugh, 1992). Others have attributed the decline in homicide detection rates to resource constraints within police agencies is (Richardson and Kosa, 2001).

Fiscal challenges now require that police agencies balance the need to combat crime with the cost of policing (Bueerman, 2012). Decrease funding for police agencies in recent years is now making it considerably difficult to increase manpower. Satisfying the demands of more globalized, culturally diverse and technologically connected societies with fewer resources has been found to limit law enforcement’s ability to protect communities and enforce the law (Bueerman, 2012).

Changes in the Behavior of Witnesses

Successful homicide investigations have been found to be heavily reliant on eye witness testimony (Riedel and Rinehart, 1996; Geberth, 1996; Corwin, 1997; Wellford and Cronin; 1999) and information provided by individuals who live within the vicinity of the crime (Litwin, 2004; Riedel and Rinehart, 1996; Greenwood et al., 1977, Reiss, 1971). In the absence of physical evidence at a crime scene or a clearly defined motive, cases with low solvability can only be solved by witness cooperation (Rinehart, 1993; Keppel & Weis, 1994).

The biggest factor affecting detection rates is the reluctance of witnesses to cooperate with investigators (Riedel, 1995). Witnesses who fear retaliation by offenders or their associates (Riedel and Jarvis, 1999), those who lack trust in the police (Kane, 2005; Puckett and Lundman, 2003; Regeoczi and Jarvis, 2013; Warner, 2007), or those who feel socially marginalized (Corwin, 1997; Simon, 1991) were found to be less likely to cooperate with a police investigation. Witnesses intimately involved with offenders at the time the homicide occurred (Riedel, 1995) or witnesses who themselves have actively provoked the homicide might also be disinclined to assist in an investigation (Black, 1983). A US study of the impact of crime stoppers (a program which offers confidentiality and monetary rewards to witnesses) on solving cases concluded that the lack of involvement by citizens in police investigations as one of the major factors affecting detection rates (Riedel, 1995).

Brookman and Innes (2013) have concluded that witness reluctance resulting from lack of trust in police can be enhanced by utilizing a community policing approach. Community policing provides a foundation for reinforcing legitimacy, reducing the fear of crime and building public trust and confidence in policing. Community policing also improves police investigations (Skogan et al., 1999) and helps prevent violence (Kenney et al., 2010; White et al., 2003).

While changes in the nature of homicides, police resources and behavior of witnesses provide some compelling anecdotal hypotheses aimed at explaining reasons for the decline in homicide detection rates, there are other possible explanations that may have not yet been exposed to systematic research (Carter and Carter, 2015).

The Investigation of Homicide

While homicides per se have been the subject of many researches, the mode of investigation has not (Keppel and Weis, 1994). Much of the information on investigating homicides is based more on anecdotal evidence handed down by word of mouth (Keppel and Weis, 1994). Most text books devoted to homicide investigations relies heavily on the subjective experience of their respective authors. The paucity of confirmed knowledge in this area is not surprising since police investigations is an area which academics and criminologists have traditionally neglected (Keppel and Weis, 1994).

In a more recent study, Keel et al (2009) suggested that insufficient academic work has gone into determining law enforcement best practices that provide substantive insights for solving homicides. Most of the empirical work in this area has focused more on responses by the criminal justice system to homicide (e.g. Jarvis and Regeoczi, 2007; Roberts, 2007; Litwin and Xu, 2007; Alderden and Lavery, 2007).

Some researchers have associated the skills possessed by detectives coupled with their experience (length of years on the job) with homicide detection (Marche, 1994; Keel et al., 2009). Greenwood et al (1977) argued that an investigator’s experience is unrelated to homicide detection but asserted that detective caseload and the quality of homicide investigations do influence detection rates (i.e. as detective caseload increases, probability of detection decreases).

After analyzing 802 homicides that occurred between 1984 and 1992 in Columbus Ohio Puckett and Lundman (2003) discounted claims by other researchers of an association between detective experience and homicide detection. The results of the study revealed that homicide units where new homicide detectives are assigned to work under the supervision of an experienced detective with primary responsibility for a homicide case, exhibited no notable difference in detection rates as compared to the detectives with more experience in the homicide department (Puckett and Lundman, 2003). A possible explanation for this finding could be that homicide investigators working in large cities were experienced police officers who progressed from being regular patrol officers to being assigned to less conspicuous detective departments prior to them being transferred to the homicide unit (Rachlin, 1995).

What seems interesting about the above line of research is the notion that the caseload of homicide detectives could be a reflection of other resource limitations, such as inadequate staffing, which exists within the police department. An inadequately staffed department could result in a reduction in the response time to homicide crime scenes as well as limit the number of detectives that are available for evidence gathering and eliciting witness testimony (Litwin and Xu, 2007; Keel et al., 2009).

According to Wellford and Cronin (1999:17) “There are few homicide cases that given the right initial response, the right timing and the right dedication of resources cannot be solved.” However, Keel et al (2009) later argued that not all homicides can be solved. As uninviting as this contention may sound, Keel and colleagues (2009) provide some compelling examples in support of their argument. They stated that some homicides may suffer from identification problems, evidentiary deficiencies or environmental factors such as weather conditions (e.g. rain) and, as such, are unlikely to be solved despite the efforts put in by investigators; for example, a homicide in which the crime scene is discovered long after the incident would have occurred. Physical evidence, if any, left at the crime scene may have deteriorated due to the length of time it took to be discovered; at which time it may be of little investigative value (Keel et al., 2009). The disorderly circumstances surrounding the violence associated with some homicides make some crime scenes extremely difficult to view, evaluate and comprehend what had actually happened (keel et al., 2009). In light of these statements Keel (2008) had earlier argued that despite the confidence exhibited by homicide supervisors in the ability of their detectives to solve cases, there is need for a determination whether their current practices and procedures allow for the highest detection rate. This argument might suggest the need for these agencies to look within to determine if their investigative procedures and processes are at a standard which fully supports effective homicide investigations.

Research has shown that police forces in the United Kingdom (UK) utilize a highly organized structure for investigating homicides (ACPO, 2006). Each homicide investigation is sub-divided into three strategic phases referred to as: (1) the instigation and initial response phase, (2) the investigation phase and (3) the case management phase. These structured procedures have been officially published in three investigative manuals known as the Murder Investigation Manual (MIM), the Core Investigative Doctrine and the Major Incident Room Standardized Administrative Procedures (MIRSAP) Manual (ACPO, 2005). These provide clear guidelines for investigating murders and other major incidents in the UK (ACPO, 2005; 2006).

Similar to the UK, other jurisdictions such as the US and Australia have also adopted standardized procedures for investigating homicides. The Federal Bureau of Investigations (FBI) frequently publishes updated homicide investigative guidelines in online bulletins which provide standardized guidelines to assist homicide investigators in the US. Similar to their US and UK counterparts, police departments in Australia also employ structured procedures for investigating homicides (Mouzos and Muller, 2001).

The use of case screening by polices forces in the UK allows homicide detectives to make an upfront and objective assessment of the solvability factors present in a homicide case. For example, an eye-witnessed homicide with a corporative witness would allow investigators to proceed differently from one in which no witnesses were present (ACPO, 2006). By determining which solvability factors are present in a case during the initial investigation of a homicide influences how resources will be deployed for each case. Her Majesty’s Inspectorate of Constabulary (HMIC, 2012) endorsed this approach and asserted that effective investigation can be facilitated through a process of case screening, allocation and finalization of crimes.

Support for police agencies implementing systems which facilitate standardized and structured management of investigations have also been advocated by the Bureau of Justice Assistance (BJA, 2013). Similar to ACPO, the BJA also endorsed the need for thorough, consistent, high quality and well-managed investigation to improve homicide solvability. Any other approach would tend to produce investigative inconsistencies resulting in investigations becoming purely subjective; a concept strongly opposed by Kahneman (2011) in his book titled “Thinking Fast and Slow”. Investigations based purely on subjectivity (intuition) potentially open them to scrutiny and skepticism from critics (ACPO, 2006; BJA, 2013).

Forensic Evidence in Homicide Investigations

Fisher (2004) argued that forensic evidence is an integral part of the criminal justice system and especially essential to homicide investigations. Fisher (2004) identified the following ways in which forensic evidence can assist homicide investigators:

- Place the suspect in contact with the victim or with the crime scene;
- Prove a crime has been committed or establish key elements of a homicide;
- Establish the identity of persons associated with a homicide;
- Accelerate the process since suspects confronted with physical evidence (or results from its analysis) may make admissions or even confess;
- Establish the facts of what happened, as forensic evidence may be more reliable than eyewitnesses;
- Exonerate the innocent.

The use of DNA evidence came in the spotlight during the O.J. Simpson murder trial in the early 1990’s and has since become tremendously popular especially in television shows such as the CSI series (McEwen, 2013). These media portrayals have shaped public perception of how investigations are to be conducted as well as the importance of forensic evidence in solving crimes. While the CSI series enforces the need for crime laboratories, they have also created an illusion within the minds of viewers that forensic analysis is accomplished almost instantly and that forensics have unlimited capabilities; which in reality is not the case (McEwen, 2013).

There is a paucity of empirical research on DNA’s role and influence in the criminal justice process especially at the pre-arrest stage (Schroeder and White, 2009). In one such rare study, Pratt et al (2006) national survey of law enforcement agencies in the US discovered a backlog of more than 96,000 unsolved homicide cases of which over 50% contained workable DNA evidence awaiting testing. An analysis of their survey responses identified time delays in DNA processing, lack of expertise in DNA processing and cost factors associated with DNA analysis as formidable barriers affecting DNA evidence routine use within police departments.

Although these techniques are available, the frequency of their use and the way they are used (or the way evidence is collected for their use) varies across different jurisdictions. This may have also contributed to the differences in detection rates across different agencies (Keel et al., 2009).

Political Influence on Homicide Investigation

With the exception of a study by Davies (2007), there has been little empirical work which has scrutinized homicide investigations in which political interference (otherwise known as the ‘authorizing environment’) has taken place. As described by Davies (2007) political interference refers to oversight and interaction by mayors or city and town councils, managing relationships with coroners and medical examiners, collaborating and coordinating with court prosecutors, dynamics of the relationships with forensic lab personnel and also the politics which reside internally within the police agency.

Davies’ (2007) study highlighted the importance of political variables in two ways. Firstly, the emergence of political dynamics from local political figures and prosecuting attorneys and secondly, from the media who exerts influence not only on police procedures and practices but also on the investigative decision-making process. Davies (2007) further argued that homicide incidents which garner much political and media attention may attract more resources thereby improving their chances of being detected. Examining homicides committed in Los Angeles between 1990 and 1994, Lee (2005) found that cases were solved quicker if the amount of media attention given to a case was extensive (e.g. stories in the Los Angeles times).

Case Characteristics and Solvability Factors

In one of the early homicide studies Wolfgang (1958) had discovered back then that homicides involving white and elderly victims were less likely to be solved if the offender and victim were strangers and if the motive for the homicide was robbery. Puckett and Lundman’s (2003) Ohio study, which analyzed factors affecting homicide detection rates from 1984 to 1992, concluded that homicide detection is influenced by the quantity of physical evidence which is created in the process of committing the murder.

Research has shown that homicides committed in a more up-close and personal manner such as by strangulation/asphyxiation or with sharp instruments such as knives are more solvable than homicides committed with firearms (Addington, 2006, Mouzos and Muller, 2001; Puckett and Lundman, 2003; Litwin, 2004). In close contact type homicides, offenders usually leave more physical evidence behind for discovery by investigators which may be linked to the Locard ‘exchange principle’ which states that “whenever two objects come into contact, a mutual exchange of matter will take place between them” (Miller, 2003:116-117).

The results for weapon types have been mixed. Some studies have found that homicides committed with firearms are less likely to be solved than homicides involving the use of other weapon types (Regeoczi et al., 2000; Litwin, 2004; Addington, 2006; Alderden and Lavery, 2007; Roberts, 2007). However, there are other studies which findings provide opposite results and concluded that weapon type does not have any notable effect on detections (Riedel and Rinehart, 1996; Jiao, 2008). Wellford and Cronin (1999) did however find that cases where the weapon used to kill the victim was found were more likely to be solved.

Discoveries have indicated that the circumstances surrounding the homicide matters (Maguire et al., 2010), murders committed during the course of another (concomitant) crime (Riedel and Rinehart, 1996; Regeoczi et al., 2000; Lee, 2005; Mouzos and Muller, 2001), and those classified as either gang or drug-related have lower solve probability than other types of homicides. Surprisingly though, Roberts (2007) found higher detection rates for homicides with these three characteristics.

Riedel and Rinehart (1996) concluded that in homicides committed during the course of another (concomitant) crime, factors such as age, ethnicity and gender of the victim disappear when issues surrounding the accompanying crime are considered. Despite being unable to determine victim-offender relationship in homicides involving a concomitant crime, one might reasonably suggest that many of these homicides involve strangers (Riedel and Rinehart, 1996). It was also noticed that homicides committed in private locations are more likely to be solved than homicides committed at public locations (Wellford and Cronin, 1999; Litwin, 2004).

Regeoczi et al (2000) in a cross-national study between the US and Canada found that in homicides involving female victims, African-American or children (under 10 years) had a higher probability of being solved. The findings about female victims could be possibly linked to domestic homicides which have high detection rates and children under 10 years usually have restricted routine activities which may result in them often being killed by someone they know (Riedel and Rinehart, 1996). The influence of victim’s ethnicity has achieved mixed reviews among researchers. The results of two studies conducted by Lee (2005) and Alderden and Lavery (2007) found higher detection rates when homicide victims were white, while other studies (Puckett and Lundman, 2003; Litwin, 2004; Roberts, 2007) have dissimilar findings.

The Wellford and Cronin (1999) Study

To inform this research the work of Wellford and Cronin (1999) will be used as the empirical foundation for a conceptual replication. This research does not have Wellford and Cronin’s level of details, but the former being a major study among the homicide literature provides some potential predictors across the three main themes that are of focus in this review.

[...]