Crime & Delinquency Evidence-Based © 2010 SAGE ... · Intimate Partner Violence in the...

23
Article Crime & Delinquency XX(X) 1–23 © 2010 SAGE Publications Reprints and permission: http://www. sagepub.com/journalsPermissions.nav DOI: 10.1177/0011128710362056 http://cad.sagepub.com Evidence-Based Prosecution of Intimate Partner Violence in the Post-Crawford Era: A Single-City Study of the Factors Leading to Prosecution Jill Theresa Messing 1 Abstract The Crawford v.Washington decision has prompted changes in the way that intimate partner violence (IPV) is prosecuted.This research uses logistic regression to exam- ine the victim, offender, and offense characteristics associated with the decision to prosecute a sample (N = 904) of domestic violence arrestees under an evidence- based prosecution strategy post-Crawford. Documentation of injury and police taking the perpetrator into custody at the scene of the crime have the greatest effect on the decision to prosecute, although the victim’s willingness to assist with prosecution is also a significant factor. Future researchers should seek to replicate these findings, better understand current prosecution strategies, and determine the criminal justice and social service interventions best equipped to combat IPV in the post-Crawford era. Keywords prosecution, evidence-based prosecution, intimate partner violence, domestic violence, post-Crawford 1 School of Social Work, Arizona State University, Phoenix, USA Corresponding Author: Jill Theresa Messing, School of Social Work, Arizona State University, 411 N. Central Ave, Suite 800, Phoenix, AZ 85004, USA Email: [email protected] at ARIZONA STATE UNIV on October 22, 2012 cad.sagepub.com Downloaded from

Transcript of Crime & Delinquency Evidence-Based © 2010 SAGE ... · Intimate Partner Violence in the...

Page 1: Crime & Delinquency Evidence-Based © 2010 SAGE ... · Intimate Partner Violence in the Post-Crawford Era: A Single-City Study of the Factors Leading to Prosecution Jill Theresa Messing1

Article

Crime & DelinquencyXX(X) 1 –23

© 2010 SAGE PublicationsReprints and permission: http://www. sagepub.com/journalsPermissions.nav

DOI: 10.1177/0011128710362056http://cad.sagepub.com

Evidence-Based Prosecution ofIntimate Partner Violence in thePost-Crawford Era:A Single-City Studyof the Factors Leadingto Prosecution

Jill Theresa Messing1

Abstract

The Crawford v. Washington decision has prompted changes in the way that intimate partner violence (IPV) is prosecuted. This research uses logistic regression to exam-ine the victim, offender, and offense characteristics associated with the decision to prosecute a sample (N = 904) of domestic violence arrestees under an evidence-based prosecution strategy post-Crawford. Documentation of injury and police taking the perpetrator into custody at the scene of the crime have the greatest effect on the decision to prosecute, although the victim’s willingness to assist with prosecution is also a significant factor. Future researchers should seek to replicate these findings, better understand current prosecution strategies, and determine the criminal justice and social service interventions best equipped to combat IPV in the post-Crawford era.

Keywords

prosecution, evidence-based prosecution, intimate partner violence, domestic violence, post-Crawford

1School of Social Work, Arizona State University, Phoenix, USA

Corresponding Author:Jill Theresa Messing, School of Social Work, Arizona State University, 411 N. Central Ave, Suite 800, Phoenix, AZ 85004, USAEmail: [email protected]

at ARIZONA STATE UNIV on October 22, 2012cad.sagepub.comDownloaded from

Page 2: Crime & Delinquency Evidence-Based © 2010 SAGE ... · Intimate Partner Violence in the Post-Crawford Era: A Single-City Study of the Factors Leading to Prosecution Jill Theresa Messing1

2 Crime & Delinquency XX(X)

The Centers for Disease Control and Prevention estimates that women are the victims in approximately 5.3 million cases of intimate partner violence (IPV) per year (NCIPC, 2003). Today, the primary response to IPV is via the criminal jus-tice system, through arrest and prosecution. This has not always been the case; because of legal statutes and cultural norms, IPV was long labeled a private mat-ter and state intervention was discouraged (Kurz, 1989; Pleck, 1987). Although there is copious literature on prosecutorial decision making where other types of criminal behavior are concerned (e.g., Albonetti, 1986, 1987), only relatively recently have arrest and prosecution become prevalent in cases of IPV, and rela-tively little is known about the factors that lead to the prosecution of these cases.

Rapid changes in policy and practice have combined with emerging case law on IPV prosecution. The Crawford v. Washington (2004) and Davis v. Washington (2006) decisions, which prohibit the use of victim statements made to police for the purposes of establishing a record of events for use in prosecution (called “tes-timonial” hearsay), have vastly changed the way that prosecutors respond to IPV. Although no-drop and mandatory prosecution have flourished in the past, it is likely that the use of evidence-based prosecution strategies—which focus on evidentiary variables such as medical reports, photographs, and 911 transcripts (excited utterances exempt from the hearsay prohibition), rather than victim report of the incident—will proliferate in this post-Crawford era. This research provides a preliminary examination of the victim, offender, and offense character-istics associated with the district attorney’s intent to move forward with the prosecution of domestic violence cases under a post-Crawford evidence-based prosecution strategy in one large northern California city.

Prosecution of Intimate Partner ViolenceMandatory, no-drop, and evidence-based prosecution strategies have spread throughout prosecutors’ offices in recent years. Mandatory and no-drop prosecu-tion strategies remove discretion from the victim in cases of IPV; because a crime has been committed against the state, prosecution occurs, with the victim as a wit-ness, regardless of the victim’s desire to prosecute (Ford, 2003). In 1996, these policies had been adopted in two thirds of U.S. prosecutors’ offices, although the vast majority of offices have flexible implementation policies (Mills, 2003; Rebovich, 1996). Hard no-drop prosecution strategies typically include subpoenas and crim-inal justice sanctions for victims who refuse to assist with the prosecution of their batterer, whereas soft or flexible policies continue to allow the victim some flexi-bility in determining whether the prosecution will move forward (Ford, 2003; Gewirtz, Weidner, Miller, & Zehm, 2006). The impetus for these policies was low rates of prosecution, although Garner and Maxwell (2009) argue that prosecution

at ARIZONA STATE UNIV on October 22, 2012cad.sagepub.comDownloaded from

Page 3: Crime & Delinquency Evidence-Based © 2010 SAGE ... · Intimate Partner Violence in the Post-Crawford Era: A Single-City Study of the Factors Leading to Prosecution Jill Theresa Messing1

Messing 3

is not necessarily infrequent in cases of domestic violence with 61.6% of those arrested for domestic violence in the United States prosecuted, deferred, or diverted.

There is conflicting evidence in regard to the efficacy of these policies. In Washington, D.C., for example, prosecution rates went from 15% before no-drop prosecution to 66% after a no-drop prosecution policy was implemented (Buzawa & Buzawa, 2003). It may be that mandatory filing and prosecution policies lead to an increase in cases that remain in the system and an increase in duration of court oversight but a decrease in rates of successfully completed prosecutions (Davis, Smith, & Taylor, 2003; Ford, 2003; Peterson & Dixon, 2005). Comparisons of domestic and nondomestic cases of aggravated and sexual assault in 15 counties across eight states in 2002 show that processing outcomes in domestic violence cases were consistently the same or more severe than the outcomes in nondomes-tic cases (Smith, Durose, & Langan, 2008). No-drop prosecution strategies were offered as one explanation for these data, along with domestic violence defen-dants having more active criminal justice status, higher rates of pretrial detention, and younger age (Smith et al., 2008).

In San Diego, California, domestic homicides decreased from 30 in 1985 to 7 in 1994, coinciding with the implementation of a no-drop prosecution policy in this county (Buzawa & Buzawa, 2003). However, the observational nature of this find-ing means that no threats to internal validity were controlled, and the change in policy cannot be definitively linked to the drop in homicide rates. Conversely, a National Institute of Justice study found that across 48 states, increased domestic violence prosecution led to increased domestic homicide among both Black and White women (Dugan, Nagin, & Rosenfeld, 2001).

Using an experimental design, Ford and Regoli (1993) found that there were no differences in violence at 6-month follow-up when perpetrators were randomly assigned prosecution strategies of pretrial diversion to counseling, prosecution with counseling recommended as a condition of probation, or prosecution with conviction and sentencing under a no-drop prosecution policy. However, when victims were randomized to a condition where they were able to drop charges, these victims reported less new violence than all other groups (Ford & Regoli, 1993). This finding led the authors to conclude that allowing the victim to drop charges provides her with power that she otherwise would not have to take control of relationship events. It also appears that mandatory policies stretch the limited resources of overburdened district attorneys and may alienate victims. This may reduce victims’ motivation to call the police when IPV occurs and expose them to future danger (Buzawa & Buzawa, 2003).

Evidence-based prosecution, although sometimes incorrectly conflated with no-drop or mandatory prosecution strategies, focuses on the successful prosecu-tion of domestic violence cases through evidentiary strategies that do not rely on

at ARIZONA STATE UNIV on October 22, 2012cad.sagepub.comDownloaded from

Page 4: Crime & Delinquency Evidence-Based © 2010 SAGE ... · Intimate Partner Violence in the Post-Crawford Era: A Single-City Study of the Factors Leading to Prosecution Jill Theresa Messing1

4 Crime & Delinquency XX(X)

victim testimony (Gewirtz et al., 2006). The tools of evidence-based prosecution include, among other things, the use of excited utterances, photographs, medical reports, and witness testimony. Evidence-based prosecution strategies are used far less often than no-drop or mandatory prosecution and, therefore, less is known regarding the implementation and efficacy of these policies (Gewirtz et al., 2006; Rebovich, 1996). A single study examining the efficacy of evidence-based pros-ecution where there were child witnesses to a domestic violence incident found that fewer cases were declined, trials increased in number, and trial convictions increased, although time to trial was lengthened (Gewirtz et al., 2006). No research to date has examined the differential impact of victim, perpetrator, and case char-acteristics on the decision to prosecute under an evidence-based policy in the post-Crawford era.

Factors Associated With Intimate Partner Violence ProsecutionRegardless of policy, there is little research examining what domestic violence cases are prosecuted (cf. Belknap et al., 2000; Dawson & Dinovitzer, 2001; Henning & Feder, 2005; Kingsnorth, Macintosh, Berdahl, Blades, & Rossi, 2001; Kingsnorth, Macintosh & Southerland, 2002; Worrall, Ross, & McCord, 2006). Table 1 summarizes research findings of the 11 articles published to date on variables influencing the decision to prosecute cases of domestic violence; research examining only judicial outcomes (Belknap & Graham, 2000; Cramer, 1999), as opposed to prosecutorial outcomes, has been excluded from the table and from most discussions of the literature. Data for each summarized study were gathered, typically in a single city, between 1978 (Rauma, 1984) and 2003 (Worrall et al., 2006).

As demonstrated in Table 1, there is some agreement among studies in regard to the effect of certain variables on the decision to prosecute pre-Crawford. When measured, victim injury (in 6 of 9 studies where the variable was measured, here-after represented as 6/9), victim cooperation at the prosecutorial stage or the victim desiring arrest (3/4), witness corroboration (3/5), the defendant having a criminal history (7/11; including domestic violence, nondomestic assault, nonviolent crime, prior arrest, prior prison term, misdemeanor conviction, felony charge, and/or fel-ony conviction, depending on the study), having an order of protection/restraining order in place (2/3), noncohabitation (4/6; in one study, this is an aggregate vari-able, combined with being married), separation (2/2), and the defendant being unemployed (3/3) led to an increase in the likelihood of prosecution in the majority of studies. Dual arrest (4/4) led to a decrease in the likelihood of prosecution in all studies examining this variable, although some authors noted that this is the effect of policy (Kingsnorth & Macintosh, 2007; Kingsnorth et al., 2001).

at ARIZONA STATE UNIV on October 22, 2012cad.sagepub.comDownloaded from

Page 5: Crime & Delinquency Evidence-Based © 2010 SAGE ... · Intimate Partner Violence in the Post-Crawford Era: A Single-City Study of the Factors Leading to Prosecution Jill Theresa Messing1

5

Tabl

e 1.

Var

iabl

es A

ffect

ing

the

Dec

isio

n to

Pro

secu

te D

omes

tic V

iole

nce

Cas

es b

y R

esea

rch

Stud

y

Sele

ct V

aria

bles

Schm

idt

Hir

sche

l &

Dav

is

Kin

gsno

rth,

D

awso

n &

K

ings

nort

h,

Kin

gsno

rth

&

Hen

ning

&

Wor

rall

Exam

ined

in

Rau

ma,

& S

teur

y, M

artin

, H

utch

inso

n,

et a

l., et

al.,

Din

ovitz

er,

et a

l., M

acin

tosh

, Fe

der,

et a

l., Pr

evio

us R

esea

rch

1984

a,b

1989

b 19

94

2001

b 20

03

2001

20

01c

2002

20

07b

2005

20

06

Dua

l arr

est

↓ ↓

Vic

tim in

jur y

↑ —

↑ ↑

↑Se

vere

vio

lenc

e

↑ —

W

eapo

n

↑ —

↑ —

Vic

tim c

oope

ratio

n

↑d

↑d

↑ ↑e

↑e

Witn

ess

corr

obor

atio

n

↑ —

—C

rim

inal

his

tory

↑ ↑

↑f —

↑f ↑

—R

estr

aini

ng o

rder

↓i

Mar

ried

—g

↑h

↓i

Non

coha

bita

tion

↑ —

Sepa

ratio

n

Su

spec

t dr

ug/a

lcoh

ol

↑ —

j

↑ —

Vic

tim d

rug/

alco

hol

↑ —

—Su

spec

t no

n-W

hite

↑k

↑k

—Su

spec

t un

empl

oyed

↑l

↑ M

ale

defe

ndan

t

Not

e: ↑

= in

crea

sed

likel

ihoo

d, ↓

= d

ecre

ased

like

lihoo

d; —

= v

aria

ble

nons

igni

fican

t.a P

revi

ous

rese

arch

stu

dy fi

ndin

gs a

re o

rgan

ized

from

left

to

righ

t: ol

dest

to

new

est

data

.b A

naly

sis

of m

ale

defe

ndan

ts o

nly.

c Res

earc

h w

as c

ondu

cted

in C

anad

a.d V

ictim

pre

senc

e at

com

plai

nt/c

harg

ing.

e Vic

tim d

esir

es a

rres

t.f S

ome

mea

sure

s in

crea

se li

kelih

ood;

som

e ar

e no

nsig

nific

ant.

g Vic

tim–o

ffend

er r

elat

ions

hip.

h Dec

reas

ed li

kelih

ood

for

boyf

rien

d/gi

rlfr

iend

or

com

mon

law

rel

atio

nshi

p.i O

ne v

aria

ble,

mar

ried

and

non

coha

bitin

g.j A

rgum

ent

drin

king

rel

ated

.k A

fric

an A

mer

ican

/Bla

ck d

efen

dant

s.l E

mpl

oym

ent

stat

us u

nkno

wn.

at ARIZONA STATE UNIV on October 22, 2012cad.sagepub.comDownloaded from

Page 6: Crime & Delinquency Evidence-Based © 2010 SAGE ... · Intimate Partner Violence in the Post-Crawford Era: A Single-City Study of the Factors Leading to Prosecution Jill Theresa Messing1

6 Crime & Delinquency XX(X)

Other variables show patterns that are less clear. The following variables have been found to increase the likelihood of prosecution in half or fewer of the studies in which they are examined and have no effect on the remainder: victim experi-encing severe violence (2/4), suspect use of a weapon (2/7), suspect drug or alco-hol use (4/8), victim drug or alcohol use (2/5), and the suspect being non-White (2/6). Relationship status appears to be more complicated, likely because of the many different ways that this variable is measured. Where marriage is examined specifically (in one case, being married is measured in combination with not cohabiting), it is found to decrease the likelihood of prosecution half of the time (2/4). Dawson and Dinovitzer (2001) examined the effect of having a boyfriend/girlfriend relationship and a common law marriage on the likelihood of prose-cution and found these relationships to decrease the likelihood of prosecution; although this does tend to indicate that married couples have an increased likeli-hood of prosecution, it does not appear that this was tested directly. With the excep-tion of victim injury and relationship status, few variables have been examined where research finds opposite effects (increase vs. decrease); however, the same variables are often found either to have an effect or to have no effect on the likeli-hood of prosecution, depending on the study.

Several other previously examined variables are not included in Table 1 because they have been found to significantly affect the likelihood of prosecution in only a single study. Victim receipt of medical treatment was not related to the decision to prosecute (Dawson & Dinovitzer, 2001; Kingsnorth & Macintosh, 2007; Schmidt & Steury, 1989), except in a single study where the victim's going to the hospital was found to increase the likelihood of prosecution (Kingsnorth et al., 2001). Although Schmidt and Steury (1989) speculated that the lack of a relationship between receipt of medical treatment and prosecution may be due to the presence of photos (negating the need for medical reports to document inju-ries), the two studies that examined the effects of photos on prosecution found no relationship (Dawson & Dinovitzer, 2001; Kingsnorth et al., 2001).

Being co-parents was found to decrease the likelihood of prosecution in one study (Kingsnorth & Macintosh, 2007) and to have no effect in another (Martin, 1994). Living in a census block with a lower median household income was found by Henning and Feder (2005) to increase the likelihood of prosecution. Worrall and colleagues (2006) found that suspect arrest at the scene increased the likelihood of prosecution. Finally, several studies examined the effect of victim and/or suspect age on the likelihood of prosecution. Younger suspect age was found to increase the likelihood of prosecution in one study (Henning & Feder, 2005) and to have no effect in two others (Martin, 1994; Worrall et al., 2006). Younger victim age, simi-larly, was found to increase the likelihood of prosecution in a single study (Kingsnorth & Macintosh, 2007) and have no effect in another (Worrall et al., 2006).

at ARIZONA STATE UNIV on October 22, 2012cad.sagepub.comDownloaded from

Page 7: Crime & Delinquency Evidence-Based © 2010 SAGE ... · Intimate Partner Violence in the Post-Crawford Era: A Single-City Study of the Factors Leading to Prosecution Jill Theresa Messing1

Messing 7

Although more recent research on the prosecution of domestic violence has focused on male and female defendants (see Table 1), for several reasons, this research focuses on prosecutorial decision making in cases of domestic violence with male defendants and female victims. First, women are victimized in intimate relationships at a higher rate than men (Black & Breiding, 2008; Dobash, Dobash, Wilson, & Daly, 1992; Tjaden & Thoennes, 2000) and this victimization is typi-cally more serious (Brush, 1990; Cantos, Neidig, & O’Leary, 1994; Melton & Belknap, 2003). Second, previous research has found that men and women are treated differently at each stage of the criminal justice system, particularly when their crimes are gendered in nature (Bernat, 2001; Chesney-Lind, 1987; Messing & Heeren, 2009; Simon & Landis, 1991). Third, in research on IPV, where gender is included as an independent variable, with few exceptions (Belknap & Graham, 2000; Kingsnorth et al., 2002), studies consistently find that the criminal justice system treats women more leniently than men at the prosecutorial phase (Davis et al., 2003; Dawson & Dinovitzer, 2001; Henning & Feder, 2005; Kingsnorth & Macintosh, 2007; Martin, 1994; Worrall et al., 2006) and the judicial phase (Belknap et al., 2000; Cramer, 1999; Davis et al., 2003; Henning & Feder, 2005). Indeed, Kingsnorth and Macintosh (2007) found that although gender showed interaction effects with various elements of the crime, female domestic violence offenders are treated more leniently at all stages of the criminal justice system, and the authors concluded that prosecutors take into account the gendered nature of IPV in their decision making.

MethodsTo examine the factors (perpetrator characteristics, victim characteristics, and incident characteristics) associated with a decision to prosecute in cases of male perpetrated IPV, this research used case files from the domestic violence unit in one large, urban northern California police department. The California penal code uses the term domestic violence and defines it as “any person who willfully inflicts upon a person who is his or her spouse, former spouse, cohabitant, former cohabi-tant, or the mother or father of his or her child, [battery or] corporal injury result-ing in a traumatic condition.” Data were collected from the year 2005 (January to December).

SamplingBased on the severity of the crime that was committed as determined by recorded penal code violations, a disproportional stratified sample was collected from a sampling frame (N = 2,598) of cases of domestic violence perpetrated by males

at ARIZONA STATE UNIV on October 22, 2012cad.sagepub.comDownloaded from

Page 8: Crime & Delinquency Evidence-Based © 2010 SAGE ... · Intimate Partner Violence in the Post-Crawford Era: A Single-City Study of the Factors Leading to Prosecution Jill Theresa Messing1

8 Crime & Delinquency XX(X)

18 years and older and responded to by police in this city. When one perpetrator was arrested for domestic violence multiple times within the year (325 offenders perpetrated 806 cases), one offense was randomly selected and included within the sampling frame. The major distinction is between sexual assault, corporal injury, and/or the use of weapons (high violence group) and the lesser charges of nonviolent offenses (restraining order violations, vandalism, criminal threats, robbery) or domestic battery (low-moderate violence group). Misdemeanor and felony cases are included in this sample; homicide cases are not included.

Within the final sampling frame, 19.5% (n = 507) of the men had crimes that qualified them for inclusion in the high violence group; the remaining 80.5% (n = 2,091) of the men were placed in the low-moderate violence group. All men in the high violence group (n = 507) and a random sample of men in the low-moderate domestic violence group were selected for inclusion (n = 507). More complete data on perpetrator, victim, and incident characteristics were collected from the standardized forms and narratives of these case files. Select case charac-teristics are presented in Table 2. Throughout the data collection process, cases were excluded because of inappropriate victim–offender relationships or substantial missing case information. The final sample consisted of the population of high violence domestic violence cases (n = 482) and a random sample of the low-moderate violence domestic violence cases (n = 422). Therefore, the final sample consists of 904 cases of male perpetrated domestic violence investigated by the domestic violence unit of one northern California police department.

ResultsStepwise logistic regression analysis is used to examine the relationship between the decision to prosecute a domestic violence offender and perpetrator character-istics, victim characteristics, and incident characteristics. The decision to prose-cute is reflected in police case files as the prosecutor’s agreement to move forward with a particular case upon presentation by the assigned detective. This is a binary outcome, with 0 being a decision to not move forward with prosecution and the value 1 representing a decision to move forward with prosecution. Of the domestic violence cases included in this sample, the prosecutor chose to move forward with prosecution in 39.7% of high violence cases and 9.8% of low-moderate violence cases (this equals 27.5% of all cases in the sample). Pursuant to California penal code section 1203.097, this prosecutor’s office does not divert cases. Given the sampling strategy, in all analyses of these data, the sample is weighted to reflect the proportion of cases in the sampling frame. The logistic regression model includes 11 variables and explains 31.9% of the variance in charging decisions (see Table 3).

at ARIZONA STATE UNIV on October 22, 2012cad.sagepub.comDownloaded from

Page 9: Crime & Delinquency Evidence-Based © 2010 SAGE ... · Intimate Partner Violence in the Post-Crawford Era: A Single-City Study of the Factors Leading to Prosecution Jill Theresa Messing1

Messing 9

Table 2. Case Characteristics

High Violence, Low-Moderate Variable n (%) Violence, n (%)

Prosecution (n = 904) Yes 186 (39.7) 40 (9.8)Offender race/ethnicity (n = 900) White 17 (3.6) 23 (5.5)African American 333 (69.7) 319 (75.6)Latino 109 (22.8) 66 (15.6)Other 19 (4.0) 14 (3.3)Victim race/ethnicity (n = 897) White 40 (8.4) 33 (7.9)African American 302 (63.2) 307 (73.3)Latina 108 (22.6) 64 (15.3)Other 28 (5.9) 15 (3.6)Cohabiting (n = 904) Yes 265 (55.0) 179 (42.4)No 217 (45.0) 243 (57.6)Alcohol and drug use (n = 902) None reported 362 (75.3) 345 (81.9)Alcohol only 98 (20.4) 66 (15.7)Drugs only 18 (3.7) 8 (1.9)Alcohol and drugs 3 (0.6) 2 (0.5)Court orders (n = 860) None 317 (69.9) 321 (78.9)Restraining order (current/previous) 75 (16.6) 67 (16.4)Emergency protective order 61 (13.5) 19 (4.7)Violence perpetrateda Weapon (n = 904) 136 (28.2) 10 (2.4)Punch (n = 894) 213 (44.5) 118 (28.4)Hit with object (n = 894) 69 (14.4) 6 (1.4)Bite (n = 894) 18 (3.8) 3 (0.7)Kick (n = 894) 52 (10.9) 34 (8.2)Injury documentationa Visible injuries (n = 901) 332 (68.9) 123 (29.2)Photos (n = 895) 236 (49.5) 73 (17.5)Medical treatment (n = 869) 153 (31.7) 36 (8.8)Perpetrator in custody (n = 897) Yes 241 (50.0) 160 (37.9)No 241 (50.0) 262 (62.1)Victim preferencesa (n = 904) Prefers arrest (n = 904) 356 (73.9) 253 (60.0)Agrees to prosecute (n = 904) 80 (16.6) 53 (12.6)

aNot mutually exclusive.

at ARIZONA STATE UNIV on October 22, 2012cad.sagepub.comDownloaded from

Page 10: Crime & Delinquency Evidence-Based © 2010 SAGE ... · Intimate Partner Violence in the Post-Crawford Era: A Single-City Study of the Factors Leading to Prosecution Jill Theresa Messing1

10

Tabl

e 3.

Log

istic

Reg

ress

ion

Mod

el, D

epen

dent

Var

iabl

e: P

rose

cutio

n

Pr

opor

tion:

Pros

ecut

ed,

Una

djus

ted

OR

A

djus

ted

OR

Va

riab

le

Not

Pro

secu

ted,

%:%

(9

5% C

I) (9

5% C

I) z

Any

thr

eats

to

kill

23.7

:12.

3 2.

36 (

1.44

-3.8

5)

2.25

(1.

11-4

.57)

2.

25*

Cou

ple

Latin

o/La

tina

25

.3:1

3.1

2.54

(1.

57-4

.11)

2.

63 (

1.48

-4.6

9)

3.29

**C

ohab

itatio

n 65

.9:2

8.8

2.72

(1.

82-4

.08)

2.

06 (

1.18

-3.5

9)

2.54

*V

ictim

des

ires

arr

est

78

.8:6

3.6

2.77

(1.

75-4

.37)

2.

23 (

1.20

-4.1

3)

2.55

*D

rug

or a

lcoh

ol u

se

30.2

:18.

8 1.

60 (

1.03

-2.4

7)

0.59

(0.

34-1

.01)

-1

.91

Perp

etra

tor

in c

usto

dy

81.8

:36.

6 5.

56 (

3.46

-8.9

4)

6.11

(3.

33-1

1.23

) 5.

84**

Vic

tim a

gree

s to

pro

secu

te

29.2

:9.9

3.

35 (

2.08

-5.3

9)

4.74

(2.

33-9

.64)

4.

53**

Res

trai

ning

ord

er

36.7

:18.

7 2.

48 (

1.63

-3.7

7)

2.16

(1.

18-3

.94)

2.

50*

Phot

os t

aken

at

scen

e 52

.0:2

8.7

4.20

(2.

80-6

.29)

1.

91 (

1.02

-3.5

7)

2.03

*In

juri

es v

isib

le

71.2

:43.

6 4.

34 (

2.86

-6.6

1)

3.42

(1.

84-6

.35)

3.

89**

Med

ical

tre

atm

ent

36.0

:16.

8 3.

32 (

2.12

-5.1

8)

2.56

(1.

45-4

.55)

3.

22**

Not

e: C

I = c

onfid

ence

inte

rval

; OR

= o

dds

ratio

.*p

< .0

5. *

*p <

.005

.

at ARIZONA STATE UNIV on October 22, 2012cad.sagepub.comDownloaded from

Page 11: Crime & Delinquency Evidence-Based © 2010 SAGE ... · Intimate Partner Violence in the Post-Crawford Era: A Single-City Study of the Factors Leading to Prosecution Jill Theresa Messing1

Messing 11

The strongest predictor of prosecution is whether the perpetrator is taken into custody at the scene of the crime (adjusted odds ratio [OR] = 6.11, p < .0005). This occurs in 50% of high violence cases and 37.9% of low-moderate violence cases (see Table 2). Three variables regarding injury documentation are included in the final model: visible injury (adjusted OR = 3.42, p < .005), victim receipt of medical treatment (adjusted OR = 2.56, p < .005), and photos taken by the police (adjusted OR = 1.91, p < .05). The odds of prosecution are also increased if the victim has a past or current permanent restraining order or a temporary restraining order or the responding officer issues an emergency protective order at the scene of the crime (all referred to as restraining orders; adjusted OR = 2.16, p < .05). If the victim can be located after the incident and states to the detective upon follow-up that she is willing to assist in the prosecution of the perpetrator, the likelihood of prosecution is 4.74 times greater (p < .005).

Two variables significantly associated with the likelihood of prosecution are related to the characteristics of the victim and perpetrator. If both members of the couple are Latino/Latina, there is a significantly greater chance of prosecution (adjusted OR = 2.63, p < .005), and if the victim and perpetrator cohabit (regard-less of marital status) there is a significantly greater likelihood of prosecution (adjusted OR = 2.06, p < .05). The final variables are related to past violence and/or the current violent incident. Any threats to kill the victim made by the perpetra-tor, with or without a weapon, lead to a higher likelihood of prosecution (adjusted OR = 2.25, p < .05). If the victim desires arrest at the scene of the crime, there is a significantly greater likelihood of prosecution (adjusted OR = 2.23, p < .05).

An officer’s report that the perpetrator is intoxicated or under the influence of drugs at the scene of the crime predicts a significant increase in the likelihood of prosecution when there is no adjustment for other variables in the model but is nonsignificant when there is adjustment for other significant variables. Other vari-ables tested and found to be nonsignificant within the final model included vio-lence perpetrated, severity of injury, the use of weapons, victim and perpetrator age, perpetrator demeanor, having children, children present at the scene of the crime, marital status, victim statements that she is scared of the perpetrator, whether the victim (or a neighbor, for example) called the police, and separation.

DiscussionViolence and Injury

Within the final model, types/acts of violence are not associated with the decision to prosecute. However, in bivariate analyses, and prior to the inclusion of injury variables in the final regression model, violent acts were significantly related to

at ARIZONA STATE UNIV on October 22, 2012cad.sagepub.comDownloaded from

Page 12: Crime & Delinquency Evidence-Based © 2010 SAGE ... · Intimate Partner Violence in the Post-Crawford Era: A Single-City Study of the Factors Leading to Prosecution Jill Theresa Messing1

12 Crime & Delinquency XX(X)

increased levels of prosecution. Violence variables significant at the p < .05 level in bivariate analyses included punching (the perpetrator hitting the victim with a closed fist), hitting with an object (e.g., baseball bat, broom, telephone, gun), kicking, and biting the victim (although biting is a relatively rare event, as seen in Table 2). Hitting the victim with any object increased the likelihood of prosecution 2.46 times in bivariate analyses and occurred in 14.4% of high violence and 1.4% of low-moderate violence cases. Punching the victim was the most common vio-lence variable and increased the likelihood of prosecution by 2.10 times, whereas biting had an OR of 3.72 and kicking had an OR of 1.99 in bivariate analyses.

The effects of these violent acts disappear when the variables examining injury were included in the model, suggesting that it is not violence per se that leads to prosecution but rather the visible and documented effects of violence (namely, injury). Within police case files, officers are asked to record both their determina-tions about whether the injuries to the victim were visible and whether these injuries were severe (vs. moderate, minor, complaint of pain, or none). Much like variables examining violence, variables examining severe injury were significant in bivar-iate analyses yet did not remain significant within the final model. However, visible injuries (n = 455, see Table 3) led to a 3.42 times increase in prosecution adjusted for other variables in the model. The majority of previous research has found that the presence of any victim injury (Henning & Feder, 2005; Hirschel & Hutchinson, 2001; Kingsnorth et al., 2002; Schmidt & Steury, 1989), an increased number of injuries (Kingsnorth & Macintosh, 2007), or the presence of severe injury (Hirschel & Hutchinson, 2001; Schmidt & Steury, 1989; Worrall et al., 2006) increases the likelihood of prosecution, although fewer studies have found no effect of victim injury (Davis et al., 2003; Kingsnorth et al., 2001). Two studies found severity of violence to increase the likelihood of prosecution (Kingsnorth & Macintosh, 2007; Schmidt & Steury, 1989), although only Kingsnorth and Macintosh (2007) found violence and injury to simultaneously increase the likeli-hood of prosecution in multivariate analyses (Schmidt & Steury used bivariate analyses).

With one exception (Kingsnorth et al., 2001), previous evidence regarding the effects of medical treatment (Dawson & Dinovitzer, 2001; Kingsnorth & Macintosh, 2007; Schmidt & Steury, 1989) and photos (Dawson & Dinovitzer, 2001; Kingsnorth et al., 2001) shows no effect. Within this research, medical treatment (n = 189) and having photos taken (n = 309) were significant predictors of prosecution, with adjusted ORs of 2.56 and 1.91, respectively. The receipt of medical treatment does suggest that the victim had injuries severe enough to war-rant medical attention. However, only 15.3% of those receiving medical treatment had injuries described by officers as severe; 44.7% of these injuries were described as moderate. It is reasonable, then, to conclude that the documentation of treatment

at ARIZONA STATE UNIV on October 22, 2012cad.sagepub.comDownloaded from

Page 13: Crime & Delinquency Evidence-Based © 2010 SAGE ... · Intimate Partner Violence in the Post-Crawford Era: A Single-City Study of the Factors Leading to Prosecution Jill Theresa Messing1

Messing 13

is the factor that leads this variable to be significant, rather than the severity of the injuries themselves.

Brandishing a Weapon and Threats to KillBrandishing a weapon follows a similar pattern as the violence variables. High violence perpetrators brandished weapons in 28.2% of cases, whereas low-moderate violence perpetrators brandished weapons in 2.4% of cases. Brandishing a weapon leads to increased odds of prosecution in bivariate analyses (OR = 2.8, p < .0005), but this variable does not remain significant in the final model. Previous research has found the use of a weapon to be related to increased domestic violence pros-ecution in bivariate (Schmidt & Steury, 1989) and multivariate (Kingsnorth & Macintosh, 2007) analyses.

Making threats against an intimate partner (either with or without a weapon) remains significant in the multivariate regression model. Threats to kill led to a 2.25 times increase in the likelihood of prosecution (p < .05). These threats occurred in 21.3% of high violence and 8.1% of low-moderate violence cases. In cases where the district attorney decided to move forward with prosecution, these threats were present in 23.7% of cases; among cases where prosecution did not occur, threats were present in 12.3% of cases (see Table 3). Threats, especially threats to kill and/or threats made with a weapon, have been related both to violent recidivism and to the escalation of IPV (Campbell, 2001; Campbell et al., 2003; Hilton et al., 2004; Kropp & Hart, 2000). However, previous research has not examined the effect of threats to kill on the decision to prosecute domestic violence offenders.

Perpetrator in CustodyThe perpetrator being taken into custody has the highest adjusted OR of all vari-ables and results in slightly greater than a six times increase in the likelihood of prosecution; this is the variable within the model that has the most predictive power. Nearly 82% of those prosecuted were taken into police custody at the time of the crime (see Table 3). Examining this relationship in another way, perpetra-tors who flee the scene of the crime are prosecuted in 8.8% of cases in this sample, whereas perpetrators who are at the scene when the police arrive are prosecuted in 42.8% of cases in this sample; this difference is quite large. The only other study to examine this found that arrest at the scene of the domestic violence incident increased the likelihood of prosecution by an OR of 11.92 in a parsimonious logistic regression model; similar to this research, arrest at the scene was the most predic-tive variable in the model (Worrall et al., 2006, p. 489).

at ARIZONA STATE UNIV on October 22, 2012cad.sagepub.comDownloaded from

Page 14: Crime & Delinquency Evidence-Based © 2010 SAGE ... · Intimate Partner Violence in the Post-Crawford Era: A Single-City Study of the Factors Leading to Prosecution Jill Theresa Messing1

14 Crime & Delinquency XX(X)

Victim Cooperation

Although there is a mandatory arrest policy in place in this jurisdiction, at the scene of the crime, the police ask the victim whether she would like the perpetrator arrested. Victims respond in the affirmative in 37% to 50% of cases (see Table 2). As has been found in previous research (Kingsnorth et al., 2002; Kingsnorth & Macintosh, 2007), when the victim states that she desires arrest at the scene of the crime, the perpetrator is 2.23 times more likely to be prosecuted. The victim’s desire for arrest does suggest that she is interested in receiving help from the crimi-nal justice system, at least at the time that the crime was committed.

In this jurisdiction, all police reports are routed to a domestic violence detective who completes a follow-up investigation. During the course of this investigation, if the detective is able to contact the victim, the detective asks whether the victim is willing to assist in the prosecution of the offender. The victim’s willingness to assist with prosecution at this point significantly increases the likelihood of prosecution—nearly five times. Previous research has found victim presence at complaint (Davis et al., 2003) or charging conference (Schmidt & Steury, 1989) and full cooperation with the prosecutor (Dawson & Dinovitzer, 2001) to increase the likelihood of pros-ecution. When a victim states or demonstrates that she is willing to prosecute, she is asking the criminal justice system to help her stop the violence that is occurring in her relationship and can act as a witness and bring further evidence against the per-petrator, strengthening the prosecution’s case. Although victim cooperation is impor-tant, it is not the only—or even the greatest, according to the ORs—consideration. In nearly 10% of cases in which there was no prosecution, the victim agreed to prosecute on follow-up. Additionally, in 70% of cases where prosecution occurred, the victim refused to assist with prosecution at follow-up.

Relationship CharacteristicsIf the victim and the perpetrator are cohabiting, regardless of their marital status, the perpetrator is 2.06 times more likely to be prosecuted. This is in contrast to previous research that has found the opposite effect of cohabitation; that is pre-vious research has found that when victims and perpetrators are not cohabitating, there is a greater likelihood of prosecution (Kingsnorth & Macintosh, 2007; Schmidt & Steury, 1989). Some variables regarding the nature of the relationship (married, ex-intimate partner) were significant in bivariate analyses, but when examined in conjunction with cohabitation, this effect disappeared. Previous research has not reported the effects of relationship status and cohabitation together in mul-tivariate analyses. Marriage increased the likelihood of prosecution in bivariate tests, an effect opposite from that found in previous studies where it was directly

at ARIZONA STATE UNIV on October 22, 2012cad.sagepub.comDownloaded from

Page 15: Crime & Delinquency Evidence-Based © 2010 SAGE ... · Intimate Partner Violence in the Post-Crawford Era: A Single-City Study of the Factors Leading to Prosecution Jill Theresa Messing1

Messing 15

measured (Martin, 1994) or measured as an aggregate variable with noncohabita-tion (Kingsnorth et al., 2002). Separation decreased the likelihood of prosecution in bivariate analyses, an effect opposite from that found in Schmidt and Steury’s (1989) and Kingsnorth and Macintosh’s (2007) research.

Drugs and AlcoholDrug or alcohol use by the perpetrator at the time of the crime significantly increases the likelihood of prosecution when entered into the model with no other variables (OR = 1.6). However, when drug or alcohol use is entered into the model with other significant variables, the sign on the coefficient changes, and the vari-able becomes nonsignificant (p < .06). Some previous research has found that the use of drugs or alcohol increased the likelihood of prosecution in bivariate (Martin, 1994; Schmidt & Steury, 1989) and multivariate analyses (Kingsnorth & Macintosh, 2007). Similar to this research, four studies found no effect of the use of drugs or alcohol in multivariate models (Henning & Feder, 2005; Hirschel & Hutchinson, 2001; Kingsnorth et al., 2002; Worrall et al., 2006).

Race and EthnicityAs shown in Table 2, perpetrators in this sample were predominantly African American (69.7%-75.6%), followed by Latino (15.6%-22.8%). The proportion of Whites (3.6%-5.5%) and those classified as Other (3.3%-4.0%) is comparatively low. Victims follow this same pattern, but with a slightly smaller number of African Americans (63.2%-73.3%) and nearly double the proportion of Whites (7.9%-8.4%). In 85.7% of cases, the race of the victim and perpetrator is the same. According to the 2000 census, in the city where this research was conducted, the racial composition is much more heterogeneous than this sample of arrestees: approximately 35% African American, 23% Caucasian, 22% Latino/Latina, and 15% Asian.

Similar to research conducted by Kingsnorth and colleagues (2001), this research examines the effect of racial dyads on the decision to prosecute; because of the homogeneity among couples, however, only couples in which both partners are of the same race/ethnicity can be examined. Within this sample, the perpetra-tors in Latino/Latina couples were 2.63 times more likely to be prosecuted than other offenders. Because of the racial and ethnic breakdown of this sample, this is essentially an effect of the perpetrators in Latino/Latina couples being prosecuted at higher rates than the perpetrators in African American couples. Some previous research has found that African American/Black perpetrators are more likely to be prosecuted (Henning & Feder, 2005; Hirschel & Hutchinson, 2001), but other

at ARIZONA STATE UNIV on October 22, 2012cad.sagepub.comDownloaded from

Page 16: Crime & Delinquency Evidence-Based © 2010 SAGE ... · Intimate Partner Violence in the Post-Crawford Era: A Single-City Study of the Factors Leading to Prosecution Jill Theresa Messing1

16 Crime & Delinquency XX(X)

studies have reported no effect of race/ethnicity (Kingsnorth et al., 2001; Kings-north & Macintosh, 2007; Martin, 1994; Worrall et al., 2006). With the exception of Kingsnorth and Macintosh (2007), the prevalence of Latinos in previous sam-ples is small (or nonexistent); comparisons between African American and Latino/Latina offenders and victims have not been made in past research on the prosecution of male perpetrated domestic violence. Furthermore, the race of both the perpetrator and victim is taken into account in this variable; the effect of eth-nicity is the strongest in the 16.13% (n = 144) of cases where both the perpetrator and the victim are Latino/Latina. Kingsnorth et al (2001) found racial dyads to have no effect on the decision to prosecute; however, their dyads were much more diverse and the sample consisted of only 8.8% (n = 40) Latino/Latina couples.

Restraining Orders and Criminal HistoryWhen the victim has a restraining order against the perpetrator or is issued an emergency protective order (a 5-day restraining order handed out at the scene of the crime by a responding officer after gaining approval from an on-call judge), the perpetrator is 2.16 times more likely to be prosecuted. Previous research has found that victims with restraining orders are more likely to have their cases prosecuted (Kingsnorth & Macintosh, 2007; Martin, 1994). In addition, previous criminal history (not examined here) has been consistently tied to increased prosecution and may act in much the same way (Davis et al., 2003; Henning & Feder, 2005; Hirschel & Hutchinson, 2001; Kingsnorth & Macintosh, 2007; Kingsnorth et al., 2002; Martin, 1994; Rauma, 1984; Schmidt & Steury, 1989).

Limitations

This research is limited in several ways. Because of the method of gathering data, through searching police case files, consistent information could not be collected for several variables. Most notably, these are the criminal history of the offender (except in cases where the victim has a restraining order or where the perpetrator committed more than one domestic violence crime within the year 2005), employ-ment, and socioeconomic status of perpetrators and victims. Because officers were not generally present at the time of the crime, statements made by victims, wit-nesses, and perpetrators about the violent incident are assumed to be true. When discrepancies occurred between statements, the officer’s interpretation of events was taken to be the truth. The decision to proceed with prosecution is a dichoto-mous variable that does not reflect all of the complexities of the criminal justice process (cf. Garner & Maxwell, 2009). Because this decision was documented in police case files, it is unknown what happened to the victim or the perpetrator after the decision to prosecute was made. It is also not known what further

at ARIZONA STATE UNIV on October 22, 2012cad.sagepub.comDownloaded from

Page 17: Crime & Delinquency Evidence-Based © 2010 SAGE ... · Intimate Partner Violence in the Post-Crawford Era: A Single-City Study of the Factors Leading to Prosecution Jill Theresa Messing1

Messing 17

information the prosecutor had regarding the case that was not included in the police case file or what the stated rationale for the decision to or not to prosecute may have been. Finally, a larger, more diverse sample from a larger number of jurisdictions would have been more useful. As it stands, this research is able to provide a descriptive analysis of prosecutorial decision making post-Crawford in a single, northern California city.

ConclusionsThis research is the first to examine an evidence-based prosecution strategy post-Crawford. Since this policy shift, injury documentation variables take on greater importance than violence and/or injury variables in supporting prosecutions of domestic violence perpetrators. Photographs of injuries and medical reports pro-vide irrefutable evidence that injury has occurred and, even without victim testi-mony, may be submitted as evidence of a crime. This strategy will not capture violent acts that do not leave immediate and/or visible injury but may be highly lethal, such as strangulation. It is not necessarily the case that that victims with the most severe injuries received medical treatment, nor is it necessarily true that the most severe injuries are visible and can be photographed. As demonstrated in Table 2, a small number of victims (31.7% of high violence, 8.8% of low-moderate violence) receive medical treatment for their injuries. Although offi-cers did not classify all injuries receiving medical treatment as severe, those with injuries classified as severe received medical treatment in 78% of high violence cases and 50% of low-moderate violence cases. When injuries are classified as severe, photos are taken in 62.5% of high violence cases and 45.9% of low-moderate violence cases. It may also be true that officers take photographs when they believe that an especially egregious crime has occurred, although further research must be conducted to examine when photographs are taken. Although variables that document injury are meaningful and combine to have a large effect on prosecution, this evidence-based strategy will not capture all victims who have experienced severe and/or life threatening violence and this strategy should be used cautiously.

Even prior to Crawford, it had been suggested that prosecutors should develop evidence-based strategies to successfully prosecute domestic violence cases with-out victim cooperation (Belknap & Graham, 2000). Because it is often difficult to gain victim cooperation and follow-through (Davis, Smith, & Nickels, 1997), evidence-based strategies are in part an attempt to render victim cooperation unnecessary through the thorough gathering and presentation of evidence (Gerwirtz et al., 2006). Nonetheless, victim cooperation remains important in evidence-based prosecutorial strategies post-Crawford. Within this sample, both a victim’s state-ment at the scene of the crime that she wanted her abuser arrested as well as a later

at ARIZONA STATE UNIV on October 22, 2012cad.sagepub.comDownloaded from

Page 18: Crime & Delinquency Evidence-Based © 2010 SAGE ... · Intimate Partner Violence in the Post-Crawford Era: A Single-City Study of the Factors Leading to Prosecution Jill Theresa Messing1

18 Crime & Delinquency XX(X)

statement on follow-up that she would assist with prosecution increased the likeli-hood of prosecution. In the case of a trial, a victim who is able to articulate what occurred during the abusive incident may be able to make the abuse more salient to the jury, who, in turn, may be more likely to convict the batterer.

Prosecutors have the ability to compel a victim to participate through sub-poena, arrest, and jail time; in the post-Crawford era, where victim statements to police regarding the abusive incident can only be used when the victim is avail-able for cross-examination, coercive practices forcing the victim to testify may appear to be a viable option. On the other hand, there is an increasing understand-ing that the victim may be the best judge of her situation and that coercion is nei-ther productive nor empowering for the victim (Mills, 2003; Ford, 2003). Previous research has found that victim advocacy during the court process—including pro-viding information about the process, referrals to social service agencies, and escort and support in pretrial meetings and in court—is the most important predic-tor of victim cooperation (Dawson & Dinovitzer, 2001). In addition to victim ser-vices, the use of specialized investigators, trained to understand the dynamics of IPV, collect evidence, and assist in interviewing victims, may assist in the develop-ment of successful evidence-based prosecutions, whether or not the victim chooses to cooperate (Henning & Feder, 2005).

Within police case files, there is not typically information on relationship his-tory, past violence, or the risk that a perpetrator poses his victim. The criminal justice system operates within an incident framework; in general, rather than cre-ating a holistic picture of the relationship, violence, and risk, the function of the police and prosecutor is to focus on a criminal incident that occurred at a particu-lar time and place. Henning and Feder (2005) speculated that the high rate of suc-cessful prosecutions in their sample was, in part, the result of a specialized domestic violence court and prosecutions unit, trained to understand the ongoing nature and dynamics of intimate partner violence. Variables such as the presence of a restraining order may act as a proxy for past violence. Although problematic, this provides information that there is a significant history of violence and that the victim is interested in keeping the abuser at some distance. Violation of the restrain-ing order is also a crime, and this, rather than the additional violence implied, may explain the observed relationship. This is similar to the presence of threats to kill; threats to kill increase the likelihood of prosecution, indicate a pattern of abusive behavior, and are an additional crime. Threats to kill are also a risk factor for domestic homicide (Campbell, 2001; Campbell et al., 2003; Hilton et al., 2004; Kropp & Hart, 2000) and should be taken seriously by prosecutors.

Of the remaining significant variables, Latino/Latina ethnicity is a personal (offender/victim) characteristic that leads to an increase in the decision to pros-ecute. One may hypothesize that this is actually an effect due to other personal

at ARIZONA STATE UNIV on October 22, 2012cad.sagepub.comDownloaded from

Page 19: Crime & Delinquency Evidence-Based © 2010 SAGE ... · Intimate Partner Violence in the Post-Crawford Era: A Single-City Study of the Factors Leading to Prosecution Jill Theresa Messing1

Messing 19

characteristics (cf. Henning & Feder, 2005), such as language barriers, although the data do not provide information on immigration status or primary language spo-ken in these cases. With some exceptions (Kingsnorth et al., 2001; Kingsnorth & Macintosh, 2007), previous research has not made comparisons between African American and Latino perpetrators or examined perpetrator and victim race/ethnicity in concert. Researchers should examine more diverse samples, take into account the race/ethnicity of the victim and the offender, and investigate why apparent biases occur. Similarly difficult to explain is the relationship between cohabitation and the decision to prosecute. The prosecutor may believe that victims who live with their abusers are more vulnerable to future abuse, although researchers should examine the rationale behind these prosecutorial decisions.

Finally, in this sample, it is likely that if the perpetrator flees the scene of the crime, his case will not be followed up by the district attorney. Although an arrest report will be written by the responding officer, it is further likely that the perpetra-tor will never be taken into police custody. In this research, and in the one other study that has examined the relationship of arrest at the scene to prosecutorial deci-sion making, this is the single most predictive variable in the model (Worrall et al., 2006). This relationship warrants further study. Previous research has demonstrated that because of mandatory arrest policies, prosecutors and detectives must spend time determining not to charge a large proportion of domestic violence cases; there-fore, they have less time to track down offenders who have fled, prepare cases, and follow leads on more dangerous perpetrators (Davis et al., 2003). It seems, then, that if a perpetrator is able to flee the scene of the crime, he is also able to beat his inti-mate partner or ex-intimate partner with (relative) impunity. This is true even though the police know who committed the crime and, in most cases, where this perpetrator lives. The creation of innovative policing strategies to identify and target the most dangerous offenders for additional follow-up when they have fled the scene may be effective in protecting the safety of victims in cases of IPV.

This research provides a preliminary examination of the factors associated with prosecution, post-Crawford, using an evidence-based strategy and compares these with the factors that have been found to increase the likelihood of prosecu-tion prior to the Crawford decision. An important limitation of this research is reliance on information from a single metropolitan city in northern California. The prosecution strategies evidenced here may be different than those in other jurisdictions; therefore, further research should focus on replicating these find-ings in different locations, with different populations. In addition, research should focus on gaining a better understanding of the rationale behind prosecutorial deci-sions. The nature of intimate partner violence is such that discrete violent incidents do not necessarily reflect the underlying patterns of power and control that play out over the course of a violent relationship. The criminal justice system, on the other

at ARIZONA STATE UNIV on October 22, 2012cad.sagepub.comDownloaded from

Page 20: Crime & Delinquency Evidence-Based © 2010 SAGE ... · Intimate Partner Violence in the Post-Crawford Era: A Single-City Study of the Factors Leading to Prosecution Jill Theresa Messing1

20 Crime & Delinquency XX(X)

hand, is equipped to deal with crimes or discrete incidents of violence. As policy and case law are evolving to deal with the unique situation of intimate partner violence, it is important to explore and evaluate innovative and collaborative strategies—including targeted police follow-up, specially trained investigators for evidence collection and victim interviews, advocacy services to assist victims through the criminal justice process, trained and dedicated prosecutors with limited caseloads, and judges who understand the dynamics of intimate partner violence—to combat violence within the home.

Declaration of Conflicting Interests

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

Funding

The author received no financial support for the research and/or authorship of this article.

References

Albonetti, C. (1986). Criminality, prosecutorial screening, and uncertainty: Toward a theory of discretionary decision making in felony case processings. Criminology, 24, 623-644.

Albonetti, C. (1987). Prosecutorial discretion: The effects of uncertainty. Law and Society Review, 21, 291-313.

Belknap, J., Graham, D. L. R., Hartman, J., Lippen, V., Allen, P. G., & Sutherland, J. (2000). Factors related to domestic violence court dispositions in a large urban area: The role of victim/witness reluctance and other variables, final report. http://www.ncjrs.gov/pdffiles1/nij/grants/184232.pdf., DC: National Institute of Justice.

Bernat, F. (2001). Gender and the law. In C. Renzetti & L. Goodstein (Eds.), Women, crime, and criminal justice: Original feminist readings (pp. 212-221). Los Angeles: Roxbury.

Black, M. C., & Breiding, M. J. (2008). Adverse health conditions and health risk behaviors associated with intimate partner violence—United States, 2005. CDC Morbidity and Mortality Weekly Report, 57, 113-117.

Brush, L.D. (1990). Violent acts and injurious outcomes in married couples: Method-ological issues in the national survey of families and households. Gender & Society, 4, 56-67.

Buzawa, E. S., & Buzawa, C. G. (2003). Domestic violence: The criminal justice response. Thousand Oaks, CA: Sage.

Campbell, J. C. (2001). Safety planning based on lethality assessment for partners of batterers in intervention programs. Journal of Aggression, Maltreatment & Trauma, 2, 129-143.

at ARIZONA STATE UNIV on October 22, 2012cad.sagepub.comDownloaded from

Page 21: Crime & Delinquency Evidence-Based © 2010 SAGE ... · Intimate Partner Violence in the Post-Crawford Era: A Single-City Study of the Factors Leading to Prosecution Jill Theresa Messing1

Messing 21

Campbell, J. C., Webster, D., Koziol-McLain, J., Block, C., Campbell, D., Curry, M. A., et al. (2003). Risk factors for femicide in abusive relationships: Results from a multi-site case control study. American Journal of Public Health, 93, 1089-1098.

Centers for Disease Control and Prevention. (2005). Adverse health conditions and health risk behaviors associated with intimate partner violence. Washington, DC: CDC.

Cantos, A., Neidig, P. H., & O’Leary, K. D. (1994). Injuries of women and men in treatment programs for domestic violence. Journal of Family Violence, 9, 113-124.

Chesney-Lind, M. (1987). Female offenders: Paternalism reexamined. In L. L. Crites & W. L. Heperle (Eds.), Women, the courts and inequality (pp. 114-139). Williamsburg, VA: National Center for State Courts.

Cramer, E. P. (1999). Variables that predict verdicts in domestic violence cases. Journal of Interpersonal Violence, 14, 1137-1150.

Crawford v. Washington, 541 U.S. 36 (2004).Davis, R. C., Smith, B. E., & Nickels, L. (1997). Prosecuting domestic violence cases

with reluctant victims: Assessing two novel approaches in Milwaukee. Washington, DC: National Institute of Justice.

Davis, R. C., Smith, B. E., & Taylor, B. (2003). Increasing the proportion of domestic violence arrests that are prosecuted: A natural experiment in Milwaukee. Criminol-ogy & Public Policy, 2, 263-282.

Davis v. Washington, 547 U.S. 813 (2006).Dawson, M., & Dinovitzer, R. (2001). Victim cooperation and the prosecution of

domestic violence in a specialized court. Justice Quarterly, 18, 593-622.Dobash, R. P., Dobash, R. E., Wilson, M., & Daly, M. (1992). The myth of sexual

symmetry in marital violence. Social Problems, 39, 71-91.Dugan, L., Nagin, D., & Rosenfeld, R. (2001). Exposure reduction or backlash? The

effect of domestic violence resources on intimate partner homicide, final report. Washington, DC: National Institute of Justice.

Ford, D. A. (2003). Coercing victim participation in domestic violence prosecutions. Journal of Interpersonal Violence, 18, 669-684.

Ford, D. A., & Regoli, M. J. (1993). The criminal prosecution of wife assaulters: Process, problems, and effects. In N. Z. Hilton (Ed.), Legal responses to wife assault: Current trends and evaluation (pp. 127-164). Newbury Park, CA: Sage.

Garner, J. H., & Maxwell, C. D. (2009). Prosecution and conviction rates for intimate partner violence. Criminal Justice Review, 34, 44-79.

Gerwirtz, A., Weidner, R. R., Miller, H., & Zehm, K. (2006). Domestic violence cases involving children: Effects of an evidence-based prosecution approach. Violence & Victims, 21, 213-229.

Henning, K., & Feder, L. (2005). Criminal prosecution of domestic violence offenses: An investigation of factors predictive of court outcomes. Criminal Justice and Behavior, 32, 612-642.

at ARIZONA STATE UNIV on October 22, 2012cad.sagepub.comDownloaded from

Page 22: Crime & Delinquency Evidence-Based © 2010 SAGE ... · Intimate Partner Violence in the Post-Crawford Era: A Single-City Study of the Factors Leading to Prosecution Jill Theresa Messing1

22 Crime & Delinquency XX(X)

Hilton, N. Z., Harris, G. T., Rice, M. E., Lang, C., Cormier, C. A., & Lines, K. J. (2004). A brief actuarial assessment for the prediction of wife assault recidivism: The Ontario domestic assault risk assessment. Psychological Assessment, 16, 267-275.

Hirschel, D., & Hutchinson, I. W. (2001). The relative effects of offense, offender, and victim variables on the decision to prosecute domestic violence cases. Violence Against Women, 7, 46-59.

Kingsnorth, R., & Macintosh, R. (2007). Intimate partner violence: The role of suspect gender in decision making. Justice Quarterly, 24, 460-495.

Kingsnorth, R., Macintosh, R., Berdhal, T., Blades, C., & Rossi, S. (2001). Domestic violence: The role of interracial/ethnic dyads in criminal court processing. Journal of Contemporary Criminal Justice, 17, 123-141.

Kingsnorth, R., Macintosh, R., & Sutherland, S. (2002). Criminal charge or probation violation? Prosecutorial discretion and the implications for research in criminal court processing. Criminology, 40, 553-578.

Kropp, P. R., & Hart, S. D. (2000). The spousal assault risk assessment guide (SARA): Reliability and validity in adult male offenders. Law and Human Behavior, 24, 101-118.

Kurz, D. (1989). Social science: perspectives on wife abuse. Gender & Society, 3, 489-505.Martin, M. (1994). Mandatory arrest for domestic violence: The courts’ response.

Criminal Justice Review, 19, 212-227.Melton, H. C., & Belknap, J. (2003). He hits, she hits: Assessing the gender differences

and similarities in officially reported intimate partner violence. Criminal Justice and Behavior, 30, 328-348.

Messing, J. T., & Heeren, J. W. (2009). Gendered justice: Domestic homicide and the death penalty. Feminist Criminology, 4, 170-188.

Mills, L. G. (2003). Insult to injury: Rethinking our responses to intimate abuse. Princeton, NJ: Princeton University Press.

National Center for Injury Prevention and Control (NCIPC). (2003). Costs of intimate partner violence against women in the United States. Atlanta (GA): Centers for Disease Control and Prevention.

Peterson, R. P., & Dixon, J. (2005). Court oversight and conviction under mandatory and nonmandatory domestic violence case filing policies. Criminology and Public Policy, 4, 535-558.

Pleck, E. (1987). Domestic tyranny: The making of social policy against family violence from colonial times to the present. New York: Oxford University Press.

Rauma, D. (1984). Going for the gold: Prosecutorial decision making in cases of wife assault. Social Science Research, 13, 321-351.

Rebovich, D. J. (1996). Prosecution responses to domestic violence: Results of a sur-vey of large jurisdictions. In E. S. Buzawa & C. G. Buzawa (Eds.), Do arrests and restraining orders work? (pp. 176-191). Thousand Oaks, CA: Sage.

at ARIZONA STATE UNIV on October 22, 2012cad.sagepub.comDownloaded from

Page 23: Crime & Delinquency Evidence-Based © 2010 SAGE ... · Intimate Partner Violence in the Post-Crawford Era: A Single-City Study of the Factors Leading to Prosecution Jill Theresa Messing1

Messing 23

Schmidt, J., & Steury, E. H. (1989). Prosecutorial discretion in filing charges in domestic violence cases. Criminology, 27, 487-510.

Simon, R. J., & Landis, J. (1991). The crimes women commit, the punishments they receive. Lexington, MA: Lexington Books.

Smith, E., Durose, M. R., & Langan, P. A. (2008). Bureau of Justice Statistics special report: State court processing of domestic violence cases. Washington, DC: U.S. Department of Justice, Office of Justice Programs.

Tjaden, P., & Thoennes, N. (2000). Full report of the prevalence, incidence, and con-sequences of violence against women: Findings from the National Violence Against Women Survey. Washington, DC: National Institute of Justice.

Worrall, J. L., Ross, J. W., & McCord, E. S. (2006). Modeling prosecutors’ charging decisions in domestic violence cases. Crime & Delinquency, 52, 472-503.

Bio

Jill Theresa Messing earned her MSW and PhD in social welfare at the University of California, Berkeley, and went on to complete a postdoctoral fellowship in inter-disciplinary violence research at the Johns Hopkins University (T32-MH20014). She is an assistant professor in the School of Social Work at Arizona State University. Dr. Messing specializes in intervention research; her interest areas are intimate partner violence, domestic homicide/femicide, risk assessment, and criminal justice–social service collaborations.

at ARIZONA STATE UNIV on October 22, 2012cad.sagepub.comDownloaded from