Shifting the Focus From Variables to Substantive Domains When Modeling Homicide Case Outcomes

• Date: 01 May 2019
• Author: Shila Rene’ Hawk,Dean A. Dabney
• DOI: 10.1177/1088767918807253
• Published date: 01 May 2019
• Subject Matter: Articles

https://doi.org/10.1177/1088767918807253

Homicide Studies

2019, Vol. 23(2) 93 –125

© 2018 SAGE Publications

Article reuse guidelines:

sagepub.com/journals-permissions

DOI: 10.1177/1088767918807253

journals.sagepub.com/home/hsx

Article

Shifting the Focus From

Variables to Substantive

Domains When Modeling

Homicide Case Outcomes

Shila Rene’ Hawk1 and Dean A. Dabney2

Abstract

To advance our understanding of the factors that predict homicide investigation

outcomes, this study systematizes measures into five substantive domains of inquiry

(involved subjects, incident circumstances, case dynamics, ecological characteristics,

and investigator factors), drawing attention to the significance of each block as a

possible conceptual model for subsequent clearance research. The domain-focused

approach was tested using 2009 to 2011 homicide cases (N = 252). Findings suggest

such comprehensive models may provide practical flexibility to researchers when

confronted with data access restrictions while preserving a sense of conceptual

sophistication.

Keywords

clearance, policing, investigation, homicide, framework

The modern homicide clearance rate (59% in 2016) is at a historic low. Poor outcomes

continue to mount, as is evidenced by the 3% drop in the national rate observed from

2015 to 2016 and a collective clearance rate of less than 50% in cities with half-a-

million to a million residents (Federal Bureau of Investigation [FBI], 2015, 2016).

Based upon this current clearance rate and number of known murders (17,250 in

2016), we can estimate that roughly 7,000 homicides in 1 year resulted in the offenders

“literally getting away with murder” (Carter, 2013, p. i). Given the high-profile nature

of the offense, unsolved murders can undermine police legitimacy and erode the

1Applied Research Services, Inc., Atlanta, GA, USA

2Georgia State University, Atlanta, USA

Corresponding Author:

Shila Rene’ Hawk, Applied Research Services, Inc., 663 Ethel St. NW, Atlanta, GA 30318, USA.

Email: shawk@ars-corp.com

807253HSXXXX10.1177/1088767918807253Homicide StudiesHawk and Dabney

research-article2018

94 Homicide Studies 23(2)

intended specific and general deterrence underlying criminal sanctions (Dawson,

2018; Hinds, 2009; Riedel & Jarvis, 1999). Furthermore, unsolved homicide cases

leave affected communities without closure, foster retaliation from co-victims, and

afford perpetrators the opportunity to commit additional crimes (Cronin, Murphy,

Spahr, Toliver, & Weger, 2007; Jacobs & Wright, 2010; Vaughn, DeLisi, Beaver, &

Howard, 2009).

Notwithstanding these and other substantial public health concerns, gaps remain in

our understanding of the factors that predict whether homicide cases get solved, thus

limiting the prospects of improving clearance rates (Wellford & Cronin, 1999/2000).

While scholars have shed significant light on the patterned nature and subtleties of

homicide behavior, Jarvis and Regoeczi (2009) observed that much less research has

been focused on law enforcement’s efforts to identify and arrest homicide suspects.

Faced with falling homicide clearance rates and an incomplete picture of the factors

that predict case clearance (Alderden & Lavery, 2007; Keel, Jarvis, & Muirhead, 2009;

Puckett & Lundman, 2003; Regoeczi & Jarvis, 2013; Riedel, 2008), the current study

proposes a holistic and practical conceptual model to attend to the “many questions

regarding correlates of homicide clearance [that] remain unanswered” (Rydberg &

Pizarro, 2014, p. 2).

The present inquiry is shaped by a series of observations about the extant scholar-

ship on homicide clearance. First, data access yields significant measurement issues in

clearance research (Hawk, 2015; Regoeczi & Miethe, 2003). Agencies tasked with the

firsthand documentation and investigations of a homicide event often limit researcher

access to review department files. Data on socioeconomic factors or bureaucratic mea-

sures are less difficult to access but present challenges in the matching of geographic

or institutional units of analysis (e.g., matching Census tracts to patrol beats to murder

location). Second, data access can hinder the theoretical development of homicide

clearance literature (Puckett & Lundman, 2003; Xu, 2008). The availability of new

and predictive measures in prior studies affects the design and results of predictive

models (Addington, 2006; Loftin & McDowall, 2010). Third, clearance studies that

rely on secondary data to predict clearances across numerous jurisdictions (i.e., multi-

site research) are not entirely comparable to single-site studies (i.e., data analyses of

one jurisdiction). The former offers more transferable and trend-focused results, but

often requires aggregates that omit the idiosyncrasies between sites and must contend

with data reporting inconsistencies (Davies, 2007), while the latter allows for better

measurement and analyses of individual case outcomes, but has limited generalizabil-

ity (Schroeder & White, 2009). Finally, the extant clearance literature is marked by

mixed, incongruent, and even contradictory findings. This results in uncertainty about

the importance or directionality of key factors among case outcomes.

Collectively, the above hindrances compromise the ability of researchers to provide

much needed guidance to police practitioners seeking to develop best practices that

might reverse the trend of falling clearance rates (Carter, 2013). To extend conceptual

sophistication on what is known about homicide clearances, this study systematizes

measures into five substantive domains of inquiry, drawing attention to the signifi-

cance of each grouping as a possible conceptual model for subsequent clearance

Hawk and Dabney 95

research. We maintain that doing so provides practical flexibility to researchers when

confronted with data access restrictions while preserving a sense of conceptual sophis-

tication. Measures from each domain were included in a single statistical model pre-

dicting homicide case outcomes, and then the contributions of each group were

validated with supplemental stepwise and “whodunit” case analyses. Furthermore, we

pursue a single-site design of homicide case data to maximize measurement quality

and insights into variances among micro-places and investigators.

It is argued that further understanding of homicide case outcomes should include

the totality of preceding characteristics of the homicide event, as well as the subse-

quent case configuration when examining clearances. Incomplete or misspecified pre-

dictive models can yield erroneous conclusions and dampen the development of

subject matter consensus. As such, the purpose of the current research was to move

toward consensus by accounting for all five substantive domains of a homicide case in

examination of case outcomes.

Homicide Clearance Research

Knowledge about homicide clearances in the United States originates from two con-

nected, yet distinctly different types of research: multisite and single-site studies.

Multisite studies (e.g., Addington, 2006; Roberts, 2007; Wellford & Cronin, 1999/2000)

consolidate homicide data from two or more police jurisdictions toward understanding

rates; these studies tend to use jurisdiction-wide measures of variables such as racial

composition or socioeconomic status. Such an approach allows researchers to examine

case closure trends but does not lend itself to empirical questions regarding investiga-

tion outcomes. While there have been important findings across cities, the aggregated

approach requires the researcher to make assumptions about the uniformity of prac-

tices, procedures, and data quality across the jurisdictions. Those assumpotions can

introduce measurement error into the models and obscures variation of smaller geo-

graphical units and among investigators that likely effect case solvability. Conversely,

single-site studies (e.g., Litwin, 2004; Puckett & Lundman, 2003; Regoeczi & Jarvis,

2013; Xu, 2008) focus on homicides occurring within one jurisdiction. These targeted

inquiries afford researchers heightened access to original data and familiarity with

department resources, thus producing the best possible conditions to operationalize the

factors predictive of case clearance. A single-site (i.e., within-city) frame of reference

is best suited for our study’s goals.

Five Domains of a Homicide Case

Our analysis of the clearance literature suggests the factors that predict homicide case

closures operate within five substantive domains: involved subjects, incident circum-

stances, case dynamics, ecological characteristics, and investigator factors. Table 1

provides an overview of the measures found in published single-site homicide clear-

ance studies within each domain. Most of the studies have sought to account for case

closure by concentrating on the impact of the victims and offenders (involved