Content uploaded by Jill Theresa Messing
Author content
All content in this area was uploaded by Jill Theresa Messing on Nov 04, 2015
Content may be subject to copyright.
The Oklahoma Lethality Assessment Study: A Quasi-Experimental Evaluation of the Lethality
Assessment Program
Author(s): Jill Theresa Messing, Jacquelyn Campbell, Daniel W. Webster, Sheryll Brown,
Beverly Patchell, and Janet Sullivan Wilson
Source:
Social Service Review,
Vol. 89, No. 3 (September 2015), pp. 499-530
Published by: The University of Chicago Press
Stable URL: http://www.jstor.org/stable/10.1086/683194 .
Accessed: 17/09/2015 14:08
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .
http://www.jstor.org/page/info/about/policies/terms.jsp
.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact support@jstor.org.
.
The University of Chicago Press is collaborating with JSTOR to digitize, preserve and extend access to Social
Service Review.
http://www.jstor.org
This content downloaded from 128.135.216.42 on Thu, 17 Sep 2015 14:08:51 PM
All use subject to JSTOR Terms and Conditions
The Oklahoma Lethality Assessment
Study: A Quasi-Experimental
Evaluation of the Lethality
Assessment Program
jill theresa messing
Arizona State University
jacquelyn campbell
Johns Hopkins University
daniel w. webster
Johns Hopkins Bloomberg School of Public Health
sheryll brown
Oklahoma State Health Department
beverly patchell
University of Utah
janet sullivan wilson
University of Oklahoma Health Sciences Center
abstract This quasi-experimental field trial examined the effectiveness of the
Lethality Assessment Program ðLAPÞ, a police–social service collaboration wherein
social service practitioners provide advocacy, safety planning, and referral for services
over the telephone during police-involved intimate partner violence ðIPVÞincidents
for women at high risk of homicide. We conducted structured telephone interviews
with survivors as soon as possible after the incident of violence and again approxi-
mately 7 months later. The majority of participants ð61.6 percentÞrecruited during the
intervention phase of the study talked to the hotline advocates, and propensity score–
matched analyses indicate that women who received the intervention reported using
significantly more protective strategies and were victimized by significantly less
physical violence than women in the comparison group. While additional research
Social Service Review (September 2015). © 2015 by The University of Chicago. All rights
reserved. 0037-7961/2015/8903-0003$10.00
499
This content downloaded from 128.135.216.42 on Thu, 17 Sep 2015 14:08:51 PM
All use subject to JSTOR Terms and Conditions
needs to be conducted, this study demonstrates that the LAP is a promising evidence-
informed intervention that holds the potential to increase survivors’safety and foster
decisions of self-care.
introduction
Intimate partner violence ðIPVÞis the systematic use of power and control
by one partner in an intimate relationship in order to subordinate another,
and it is often reinforced through physical or sexual violence ðStark 2007Þ.
Approximately 35 percent of women in the United States will be abused by a
current or former intimate partner in their lifetimes, and IPV is defined by
the Centers for Disease Control and Prevention as physical violence, sexual
violence, threats of violence, coercion, psychological aggression, or stalking
ðBlack et al. 2011Þ.Twenty-five percent of women will be victimized by
severe physical IPV ðincluding being hurt by having their hair pulled, being
hit with a fist or something hard, being kicked, being slammed against
something, being strangled or suffocated, being beaten, being burned on
purpose, or being abused with a knife or gunÞin their lifetimes ðBlack et al.
2011Þ. In 2010, IPVaccounted for 22 percent of all violent crimes committed
against women ðTruman 2011Þ. Although police are generally the first
responders in cases of IPV, their response focuses on accountability for the
offender and may not take into account the service needs of victim-
survivors.
The Lethality Assessment Program ðLAPÞ, which was developed by the
Maryland Network Against Domestic Violence, is a collaboration between po-
lice and social service providers and is intended to provide victim-survivors
with advocacy services at the scene of police-involved IPV incidents. Police
officers responding to the scene of an IPV incident use a risk assessment
called the Lethality Screen to identify victim-survivors who are at high
risk of homicide.Women who are determined by the screening to be at high
risk are offered the opportunity to speak on the telephone with an advo-
cate at a collaborating domestic violence agency. During the phone call, the
advocate provides the victim-survivor with immediate safety-planning as-
sistance and encourages her to come in for further services. Collaboration,
self-determination, and empowerment are the foundation of this intervention.
The purpose of this quasi-experimental field trial was to estimate the
effectiveness of the LAP. The study was conducted in seven police juris-
dictions in Oklahoma, a state where a high proportion of women are
victimized by IPV and intimate partner homicide ðBlack et al. 2011; Violence
500 |Social Service Review
This content downloaded from 128.135.216.42 on Thu, 17 Sep 2015 14:08:51 PM
All use subject to JSTOR Terms and Conditions
Policy Center 2013Þ. In particular, we asked the following research ques-
tions: Does the LAP ðiÞincrease the use of safety strategies and ðiiÞreduce
the frequency and severity of repeat IPV?
ipv risk assessment
IPV risk assessment instruments provide information about the likelihood
that an abuser will re-assault, severely re-assault, or kill his intimate partner
ðMessing and Thaller 2014Þ. Many victims of IPV, especially those who have
been severely abused, are acutely aware of the possibility of homicide
ðStuart and Campbell 1989; Langford 1996Þbut have difficulty assessing and
tend to underestimate their degree of risk ðHeckert and Gondolf 2000;
Sharps et al. 2001; Campbell 2004Þ. Victim-survivors of IPV should be
educated about their risk and potential risk factors ðCampbell 2004Þ, par-
ticularly because victim-survivors often take protective actions when they
recognize that the violence is escalating, and their concerns for their safety
may motivate them to leave their abusers or take other actions to increase
their safety ðGondolf and Fisher 1988; Fischer and Rose 1995; Campbell et al.
1998; Martin et al. 2000; Pape and Arias 2000; Short et al. 2000; Burke et al.
2004Þ. As police officers are often the first responders in cases of IPV, they
are in an ideal position to conduct risk assessment.
In order to predict which women are at a high risk of re-assault or
homicide, the LAP uses the Lethality Screen, an 11-item version of the
Danger Assessment ðDAÞthat is intended for use by field practitioners or
first responders. The DA is intended to predict femicide ðCampbell et al.
2003Þ, but it is also successful at predicting repeat assault ðMessing and
Thaller 2013Þ. The Lethality Screen has shown high levels of sensitivity for
predicting severe IPV ða woman’s partner forced sex, abused her with a
knife or gun, punched her, hit her with something that could hurt, strangled
her, beat her up, burned her, kicked her, tried to kill her, or nearly killed her;
93.18 percentÞand near-lethal IPV ðawoman’s partner tried to kill her or
nearly killed her; 92.86 percentÞ. It has shown a slightly lower sensitivity
ð86.96 percentÞwhen predicting any IPV ðMessing et al. 2015Þ.However,
specificity is low in all analyses ð21.3–21.95 percentÞ. The Lethality Screen
has good agreement with IPV survivors’perceptions of risk and with the DA
ðMessing et al. 2015Þ. The high sensitivity for predicting severe and near-
lethal IPV is ideal for use with the LAP, because, on balance, it seems better
to screen an IPV victim-survivor who may not be re-assaulted into a brief
Oklahoma Lethality Assessment Study | 501
This content downloaded from 128.135.216.42 on Thu, 17 Sep 2015 14:08:51 PM
All use subject to JSTOR Terms and Conditions
advocacy intervention than to decline to provide the intervention to a
woman who is at risk for future violence or homicide.
help-seeking among survivors of ipv
The Bureau of Justice Statistics indicates that approximately half of IPV
victim-survivorsreport that the police had been called due to IPV ðBachman
and Coker 1995; Rennison and Welchans 2000; Catalano et al. 2009Þ.Upto
92 percent ofwomen who are seeking services, such as help from the courts,
assistance with orders of protection, legal assistance, or domestic violence
services ðe.g., advocacy, shelterÞ, report that the police had been called due
to IPV ðBerk et al. 1984; Goodman et al. 2003; Goodkind, Sullivan, and Bybee
2004Þ. The proportion of women seeking help from the police significantly
increased from 1993 to 1998 ðRennison and Welchans 2000Þ, and when the
victim identifies IPVas a crime, domestic violence is reported to the police
at rates equal to the reporting of other crimes ðFelson et al. 2002Þ.
As the severity or frequency of abuse increases, so do calls to the police
ðJohnson 1990; Gondolf 1998; West, Kantor, and Jasinski 1998; Bonomi et al.
2006Þ. Fifty-six percent of victims of intimate partner homicide had called
the police in the year before they were killed, and 24.5 percent had an order
of protection ðCampbell et al. 2003Þ. Thus, the majority of victims of
femicide had contact with the police, yet the police response was unable to
save their lives. A call to the police does not always result in an arrest, and
research is mixed on the deterrent effects of arrest ðMaxwell, Garner, and
Fagan 2001; Campbell et al. 2003; Campbell et al. 2005; Felson, Ackerman,
and Gallagher 2005; Hirschel 2008; Cho and Wilke 2010Þ. Across studies,
recidivism rates after arrest range from 17 percent ðFelson et al. 2005Þto 49
percent ðHilton et al. 2008Þ.
Women access domestic violence services much less often than they seek
police help. Among samples of women not recruited from shelters or
domestic violence service agencies but generally recruited after coming into
contact with the police or seeking an order of protection, the percent of
victim-survivors accessing domestic violence services ranges from 4.8 per-
cent to 38 percent ðBrookoff et al. 1997; Gondolf 1998; Hutchison and
Hirschel 1998; Coker et al. 2000; Macy et al. 2005Þ, and the percent of
women accessing shelter services ranges from 3 percent to 8.9 percent
ðBrookoff et al. 1997; Gondolf 1998; Hutchison and Hirschel 1998; Wiist and
McFarlane 1998Þ. While research finds that accessing domestic violence
502 |Social Service Review
This content downloaded from 128.135.216.42 on Thu, 17 Sep 2015 14:08:51 PM
All use subject to JSTOR Terms and Conditions
services ðGondolf 1998; Coker et al. 2000; Henning and Klesges 2002; Macy
et al. 2005Þand shelters ðGondolf 1998; West, Kantor, and Jasinski 1998Þ
increases as the severity of physical violence increases, in one study exam-
ining homicide, only 4 percent of the women who were murdered by their
intimate partners had accessed domestic violence shelters or crisis services
in the previous year ðSharps et al. 2001Þ. Shelter services are shown to be
effective in reducing severe and moderate re-assault in one prospective
study ðMessing, O’Sullivan, et al. forthcomingÞ.
Women who lack awareness of available resources and who encounter
difficulty when accessing services are more likely to remain in abusive
relationships ðPatzel 2006Þ. A collaborative response, such as the LAP, that
is intended to bring together the criminal justice and social service systems
is therefore expected to reduce IPV ðSalazar et al. 2007Þ. Jurisdictions with a
coordinated community response find that their specialized police units
make more felony arrests of IPVoffenders ðBledsoe, Sar, and Barbee 2006Þ.
In addition, police–social service collaborations increase the likelihood that
victim-survivors will seekhelp from the criminal justice system in the future
ðDavis and Taylor 1997; Davis, Maxwell, and Taylor 2003; Hovell, Seid, and
Liles 2006; Stover 2012Þ. However, there is insufficient evidence showing
that these programs reduce victimization ðDavis and Taylor 1997; Davis et al.
2003; Garner and Maxwell 2008; Hovell et al. 2006; Stover 2012Þ.
Research suggests that low-cost, clear, simple assessments and referrals
such as teaching women safety strategies over the telephone can be effective
in helping women in abusive relationships enhance their safety skills ðMc-
Farlane et al. 2004; McFarlane et al. 2006Þ. Theoretically, the optimal time
for intervention may be shortly after an abusive episode, when women are
likely to believe that the violence will not cease and are more likely to reach
out for help ðCurnow 1997Þ. Data collected on the LAP from 100 agencies
across Maryland demonstrate that a portion of IPV victim-survivors who
are at high risk will seek services after the intervention occurs. In 2012,
Maryland jurisdictions conducted 12,108 risk assessments, determining
that 6,224 victims ð51 percentÞwere at high risk for homicide. Of the vic-
tims who were at high risk, 3,277 ð53 percentÞspoke on the phone to
an advocate. Of those victims who spoke on the phone to an advocate, 925
ð28 percentÞlater sought services, such as counseling, legal assistance, or
shelter from the service provider to whom they spoke ðMNADV 2013Þ.
Without a comparison group, however, it is unclear whether these findings
can be attributed to the LAP.
Oklahoma Lethality Assessment Study | 503
This content downloaded from 128.135.216.42 on Thu, 17 Sep 2015 14:08:51 PM
All use subject to JSTOR Terms and Conditions
This study, which analyzes the effectiveness of the LAP in seven police
jurisdictions in Oklahoma, is the first rigorous evaluation of the LAP. We
hypothesize that:
H1: The LAP increases the rates of measured emergency safety planning
and help-seeking both (a) immediately after the intervention and (b)at
follow-up.
H2: The LAP decreases the frequency and severity of repeat IPV among
women who had police contact due to IPV in participating jurisdictions
during the study’stimeframe.
method
study design
The study is a non–equivalent groups quasi-experimental field trial using a
historical comparison group ðsee also Messing, Campbell, and Wilson 2015Þ.
The comparison group was recruited into the study prior to the implemen-
tation of the LAP and received what was, at that time, the standard police
response, while the intervention and intent-to-treat groups were recruited
into the study after implementation of the LAP and received the LAP
response. Our aim was to assess the effect of a collaborative police–social
service intervention on victim-survivors who were at high risk of future
violence and homicide, in light of previous research suggesting that tele-
phone intervention and referral can lead to the adoption of safety strategies
ðMcFarlane et al. 2004, 2006Þ. Thus, our main analyses estimate treatment-
on-the-treated effects by comparing women who were at high risk for fu-
ture violence and homicide and received the standard police response ðhere-
after, the comparison groupÞto women who were at high risk for future
violence and homicide and were informed of their risk status by a police
officer and chose to speak to a hotline advocate at the scene of a police-
involved IPV visit ðhereafter, the intervention groupÞ. We additionally esti-
mate intent-to-treat effects by examining differences between women in the
comparison group and women who were at high risk for future violence and
homicide and were informed of their risk status by a police officer at the
scene of a police-involved IPV incident, regardless of whether or not they
chose to speak to the hotline advocate ðhereafter, the intent-to-treat group;
see fig. 1Þ. This research was approved by the institutional review boards of
504 |Social Service Review
This content downloaded from 128.135.216.42 on Thu, 17 Sep 2015 14:08:51 PM
All use subject to JSTOR Terms and Conditions
the University of Oklahoma Health Sciences Center, the Oklahoma State
Department of Health, Arizona State University, Johns Hopkins University,
the Cherokee Nation, and the National Institute of Justice.
study site
The seven participating police jurisdictions in Oklahoma were primarily
urban and had a total population of 1,150,000 people ðsee table 1Þ.There
FIGURE 1. Procedure and sample flow chart
table 1. Information about Participating Jurisdictions
Classification Population
Number of
Officers
Number of
Advocates
Urban 17,000 32 Unknown
Urban* 580,000 1,029 20
Urban*
,a
392,000 745 15
Suburban*
,a
99,000 130 15
Urban ðcollege townÞ* 46,000 71 Unknown
Rural ðIndian nationÞ
b
Unknown 14 Unknown
Urban ðcollege townÞ
b
16,000 30 Unknown
Note.—The population numbers for this table were obtained from the US Census Bureau, QuickFacts
Beta, April 1, 2010. The numbers of advocates were provided by the advocacy organizations. Classification
types having the same superscripted letters had the same advocacy organizations and therefore have the
same number of advocates.
* Referred participants during the intervention and comparison phases of the research study.
Oklahoma Lethality Assessment Study | 505
This content downloaded from 128.135.216.42 on Thu, 17 Sep 2015 14:08:51 PM
All use subject to JSTOR Terms and Conditions
was a range in jurisdiction size,with populations of 16,000 to 580,000, and
police departments ranging in size from 14 deputies to 1,029 sworn officers.
Where information is known, between 15 and 20 advocates per agency
assisted with the LAP intervention.
intervention training
The Maryland Network Against Domestic Violence ðMNADVÞprovided
their standard 1-day LAP train-the-trainer training to select police officers
and advocates in Oklahoma. Following this, MNADV provided each police
department with a PowerPoint presentation that trainers within the police
departments and social service agencies used to conduct mandatory train-
ings with additional officers and advocates. The LAP training was also
integrated into police academy training on IPV. In order to encourage
officers to carry out the LAP as intended, throughout the study period,
researchers ðsometimes in collaboration with advocatesÞprovided police
departments with brief electronic ðMediasite, PowerPointÞand in-person
roll-call refresher trainings. MNADV remained available throughout the
study period to provide support and technical assistance to police depart-
ments and advocacy organizations.
Prior to the LAP training, officers conducted their police investigation
and provided victim-survivors with a card that included information about
IPV and a toll-free hotline number.
1
Before implementation of the LAP,
officers did not conduct risk assessment during their visits, did not use risk
assessment to educate victim-survivors about the risk that their intimate
partners posed and risk factors for future violence and homicide, and did not
actively engage with victim-survivors to reach out to advocacy services and
implement safety strategies. Officers then received the LAP training on how
to assess risk using the Lethality Screen, score the risk assessment, and ini-
tiate a telephone call with a hotline advocate for high-risk victim-survivors.
Police officers were instructed to ask victim-survivors the 11 questions
included on the Lethality Screen outside of the presence and hearing of the
perpetrator ðsee fig. 2Þ.Officers were taught to then score the Lethality
Screen and classify victim-survivors as “high risk”or “not high risk”based
on the victim-survivor’s responses to the 11 questions. Officers were
1. Oklahoma has a discretionary arrest policy for domestic violence assault ðOK State
Statute §22–60.16Þ.
506 |Social Service Review
This content downloaded from 128.135.216.42 on Thu, 17 Sep 2015 14:08:51 PM
All use subject to JSTOR Terms and Conditions
additionally trained to use their own professional judgement; if an officer
believes that a victim-survivor is at high risk but she is not classified as such
based on the Lethality Screen, the officer is also able to make a determina-
tion of high risk based on his or her belief.
When a victim-survivor screens in as high risk, the officer is instructed
to tell her that she is at high risk for homicide and to ask her if she is will-
ing to speak on the telephone to a hotline advocate. If the victim-survivor
agrees to speak to an advocate, the officer then calls the hotline number and
FIGURE 2. The Lethality Screen
Oklahoma Lethality Assessment Study | 507
This content downloaded from 128.135.216.42 on Thu, 17 Sep 2015 14:08:51 PM
All use subject to JSTOR Terms and Conditions
places the victim-survivor on the telephone with the advocate. The advo-
cate is trained to keep the telephone conversation brief, express concern
for the victim-survivor’s well-being and the well-being of her children
if applicable, reinforce the officer’s message about the risk that she faces
due to IPV, conduct immediate safety planning, and encourage the victim-
survivor to come in for additional services. If a high-risk victim-survivor
chooses not to speak on the telephone to an advocate, the officer is trained
to call the advocacy hotline and obtain immediate safety planning advice
to share with the victim-survivor.While on the telephone, the officer may
ask the victim-survivor again if she would like to speak to the advocate. If
there are any immediate safety planning steps that the officer can then
take with the high-risk victim-survivor ðe.g., transport to shelterÞ,the
officer is trained to do so. If a victim-survivor is not classified as high risk,
the officer is trained to tell her that IPV can be dangerous, provide her with
information about risk factors for homicide, and refer her to local domestic
violence services.
comparison group
The comparison group was recruited into the study prior to the implemen-
tation of the LAP, between July 2009 and October 2010. During this time,
1,137 women were referred to researchers by police departments. Research-
ers were not able to contact 486 ð42.7 percentÞof the women who were
referred, and another 47 women ð4.1 percentÞwere not eligible to partici-
pate in the study ðe.g., they were under 18 or were notvictims of IPVÞ.Ofthe
604 eligible referrals researchers contacted, 440 ð72.8 percentÞparticipated
in a baseline interview. Seven duplicate participants were removed from the
comparison group ðn5433Þ. At the beginning of the baseline interview,
researchers conducted the Lethality Screen with comparison group partic-
ipants, referencing the victim-survivor’s current situation with her abusive
partner; 342 women ð79.0 percentÞwere deemed high risk, and they make
up the comparison group.
intervention and intent-to-treat groups
Recruitment for the intervention and intent-to-treat groups, conducted
after the implementation of the LAP, occurred from October 2010 through
February 2013. During this time, 2,022 women were referred to
508 |Social Service Review
This content downloaded from 128.135.216.42 on Thu, 17 Sep 2015 14:08:51 PM
All use subject to JSTOR Terms and Conditions
researchers by police departments. Researchers were not able to contact
1,041 ð51.5 percentÞof the women who were referred, and another 43
women ð2.1 percentÞwere not eligible to participate in the study ðe.g., they
were under age 18 or were not victims of IPVÞ. Of the 938 eligible referrals
the researchers contacted, 657 ð70 percentÞparticipated in a baseline inter-
view. Nine duplicate participants were removed. The intent-to-treat group
consists of 563 women who were classified as at high risk based on their
responses to the Lethality Screen conducted by a police officer at the scene
of the IPV incident ðn5538Þor based on officer belief ðn525Þ.The
intervention group consists of the 347 women ð61.6 percent of the intent-
to-treat groupÞwho both were classified as high risk by the officer and who
spoke with an advocate on the telephone at the scene of the IPV incident.
participant recruitment
At the scene of each IPVincident, after the police officer completed the visit,
he or she read an IRB-approved statement asking the victim-survivor if a
researcher could contact her.
2
If the victim-survivor was willing to have a
researcher contact her, the officer asked for and recorded one or two safe
telephone numbers and a safe time to call. Researchers were provided with
this information and called victim-survivors to explain the study and obtain
informed consent.
baseline and follow-up data collection
Structured telephone interviews lasting approximately 45 minutes were
conducted with study participants at two time points. Baseline interviews
were conducted as soon as possible after the referral, and the median time
between the police visit and the baseline interview was 15 days. Follow-up
interviews were completed with 212 participants in the comparison group
ð62 percent retention rateÞ, 315 participants in the intent-to-treat group
ð55.9 percent retention rateÞ, and 202 participants in the intervention
group ð58.2 percent retention rateÞ. Women who were employed full-time
or part-time were more likely to participate in follow-up interviews ðX
2
5
2. Police officers read IRB statements to women in the comparison group, women in the
intervention group,women in the intent-to-treat group, and women who were not ultimately
interviewed.
Oklahoma Lethality Assessment Study | 509
This content downloaded from 128.135.216.42 on Thu, 17 Sep 2015 14:08:51 PM
All use subject to JSTOR Terms and Conditions
5.923, p<.05Þ, as were participants who had higher levels of education
ðX
2
513.535, p<.05Þ. There was a significant difference in the mean time to
follow-up for the intervention and comparison groups ðt53.577; df 5
316.723; p<.0001Þ.Participantsinthecomparison group averaged
8.28 months ðSD 53.72Þto follow-up, and participants in the interven-
tion group averaged 7.25 months ðSD 51.89Þtofollow-up.Therewereno
significant differences in mean time to follow-up between the interven-
tion and intent-to-treat groups, with the intent-to-treat group averaging
7.2 2 m on th s ðSD 51.90Þto follow-up. The median time to the follow-up
interview for all participants was 7 months.
measurement
Demographic and Relationship Characteristics
At the baseline interview, participants were asked to report their educa-
tional achievements, their employment status, their ethnic background,
their immigration status, their age in years, their legal marital status, the
gender of their abusive partner, whether they had children living in their
households,whether they had children with their abusive partners,whether
they were currently pregnant, and whether they currently lived with their
abusive partners.
Protective Actions
Protective actions were assessed using an adapted version of McFarlane
and colleagues’ð2004Þsafety-promoting behavior checklist ðsee the ap-
pendixÞ. At baseline, participants were asked whether they had engaged in
any protective actions in the 6 months prior to the police visit or since the
police visit. At follow-up, participants were asked to report any protective
actions they had engaged in since they last spoke to the interviewer.
Intimate Partner Violence
Experiences of IPV were assessed at baseline ðin the past 6 monthsÞand
follow-up ðsince the last interviewÞusing an adapted version of the revised
Conflict Tactics Scale ðCTS-2; Straus et al. 1996Þ. In order to examine the
frequencyand severity of physical violence, items on the CTS-2 were scored
using the severity-times-frequency weighted score, as recommended by the
scale developer ðStraus 2004Þ. Scores from the baseline interview were
subtracted from scores on the follow-up interview in order to examine
510 |Social Service Review
This content downloaded from 128.135.216.42 on Thu, 17 Sep 2015 14:08:51 PM
All use subject to JSTOR Terms and Conditions
changes ðdecreases/increasesÞin violence severity and frequency between
baseline and follow-up.The CTS-2 has demonstrated median internal con-
sistency of .82 in studies examining female IPV victimization and reports
good construct validity ðStraus and Mickey 2012Þ. In this study, the CTS-2
frequency-by-severity scores demonstrated adequate internal consistency at
baseline ða5.77Þand excellent internal consistency at follow-up ða5.96Þ.
The Danger Assessment
Researchers administered the Danger Assessment ðDAÞat the baseline
interview, referencing victim-survivors’current situations with their abu-
sive partners. This risk assessment was administered in addition to the
Lethality Screen ðwhich was administered by police at the scene of the IPV
incident to women in the intervention and intent-to-treat groups and by
researchers at the time of the baseline interview to women in the compar-
ison groupÞ, as it has a more extensive history of validation, includes
additional risk factors, and has greater specificity ðCampbell et al. 2003;
Campbell, Webster, and Glass 2009; Messing and Thaller 2013Þ. Items on
the DA are weighted and summed to produce an overall score, with higher
numbers indicating greater risk.Weighted and summedscores can be placed
into four categories of risk: variable danger ð0–7Þ, increased danger ð8–13Þ,
severe danger ð14–17Þ,andextremedangerð18 or higherÞ. In this study, the
DA demonstrated adequate internal consistency at baseline ða5.76Þand at
follow-up ða5.76Þ.
analysis
Because the main analyses focus on treatment-on-the-treated effects, partic-
ipants in the comparison and intervention groups were compared on demo-
graphic and relationship characteristics using bivariate statistics. Where no
differences were found, these characteristics are reported for the interven-
tion and comparison groups combined. Where differences between groups
were found, these differences are reported and the characteristics of the
comparison and intervention groups are reported separately. As the inter-
vention group is a subset of the intent-to-treat group, the demographic and
relationship characteristics of these two groups are similar. Throughout, cases
with missing values on any variable were dropped from the analysis. Depend-
ing on the analysis, 1.2–6.1 percent of cases had missing data, accounting for
the slightly different sample sizes across analyses of similar data.
Oklahoma Lethality Assessment Study | 511
This content downloaded from 128.135.216.42 on Thu, 17 Sep 2015 14:08:51 PM
All use subject to JSTOR Terms and Conditions
Regression Controlling for Covariates
Logistic and linear regression models ðhereafter, regression modelsÞwere
used to estimate the effect of intervention group status ðvs. comparison
group statusÞon protective actions and violent victimization. Covariates
were included in the models to control for intervention group and compar-
ison group baseline differences in marital status, immigration status, DA
category, and time between the baseline and follow-up interviews.
Propensity Score Analysis
Given nonrandom assignment to intervention and comparison groups,
significant observed differences between the groups at baseline, and juris-
dictional differences in the implementation of the LAP intervention, pro-
pensity score matching was used to estimate the average treatment effect
on the treated participants ðRosenbaum and Rubin 1983Þ. By using propen-
sity scores, we adjust for observed selection bias, thus allowing an estima-
tion of causal inference in a study that is not a true experiment ðGuo and
Fraser 2015Þ. Because propensity scores are estimated ðnot knownÞ,we
used the “teffects psmatch”command in STATA 14 ðAbadie and Imbens
2012Þ. This command employs nearest-neighbor matching of cases with
replacement.When propensity scores are tied—that is, when there is more
than one match for an observation in the intervention group—the observa-
tion is matched with all tied observations. Three police jurisdictions were
dropped from the propensity score analysis because they did not refer
participants to both the comparison and intervention groups ðn58–11
depending on analysis; see table 1Þ. The standardized difference and
variance ratios were used to examine group balance before and after
propensity score matching, and balance was improved in the matched
data. Post–propensity score matching, hypotheses were tested using linear
or logistic regression.
Intent-to-Treat Analysis
Propensity score-matched analyses were also conducted using an intent-to-
treat framework, as 61.6 percent of participants who were assigned to the
LAP intervention ði.e., screened as high risk by a police officerÞwere
included in the intervention group ði.e., spoke to a hotline advocateÞ.
Intent-to-treat models are conservative. Because not everyone assigned to
receive the intervention receives it, an intent-to-treat analysis will consis-
tently provide a smaller estimate of the effect of the intervention than can be
512 |Social Service Review
This content downloaded from 128.135.216.42 on Thu, 17 Sep 2015 14:08:51 PM
All use subject to JSTOR Terms and Conditions
attributed to the intervention itself ðAngrist 2006Þ. In this study, a relatively
low proportion of participants who were assigned to the LAP intervention
received it, which would be expected to seriously attenuate any effect.
Intent-to-treat analysis relies on the assumption that the proportion of
people assigned to treatment and those actually treated will be similar
across all implementations of the intervention ðAngrist 2006Þ.However,
this proportion differed widely across jurisdictions in this study ðrange:
from 42.1 percent to 77.8 percentÞ. The intent-to-treat model cannot be
expected to provide an accurate average intervention effect across such a
wide range of compliance rates. Nevertheless, intent-to-treat analyses are
another strategy for taking selection into account in estimating intervention
effects, and thus they may still provide useful information.
results
description of the comparison
and intervention groups ðbaselineÞ
Participants in the comparison and intervention groups ranged in age from
18 to 79 years,with a mean age of 32.52 ðSD 59.94Þ. The largest racial/ethnic
group was white ð42.8 percentÞ, followed by African American ð29.4 per-
centÞ, Native American ð10.0 percentÞ,Latinað7. 9 pe r ce n tÞ, multiracial
ð7.5 p e rc en t Þ,andotherð2.2 percentÞ.Lessthanone-fifth of participants
ð16.8 percentÞreported that they currently lived with their partners, and
2.5 percent of participants reported that their partners were female. Sixty-
five percent of participants reported that they had children living in their
households, 45.7 percent of participants reported that they had children
in common with their partners, and 7.1 percent of participants reported that
they were currently pregnant at the time of the baseline interview. Approx-
imately half of participants reported that they had completed high school
or had a GED ð51.8 percentÞ, and less than half of participants were em-
ployed part-time or full-time ð40.64 percentÞ.
There were significant differences ðX
2
56.73, p<.05Þin marital status
between the comparison group and the intervention group, with nearly
twice as many participants in the comparison group reporting that they
were separated or divorced. In the comparison group, 58.3 percent of
participants reported that they were single, 22.8 percent reported that they
were married, and 18.9 percent reported that they were separated or
divorced. In the intervention group, 64.6 percent of participants reported
Oklahoma Lethality Assessment Study | 513
This content downloaded from 128.135.216.42 on Thu, 17 Sep 2015 14:08:51 PM
All use subject to JSTOR Terms and Conditions
that they were single, 24.3 percent reported that they were married, and 10.9
percent reported that they were separated or divorced.
Few participants overall were born outside of the United States, but
there were significant differences in immigration status ðX
2
54.7, p<.03Þ
between the intervention and comparison groups, with over twice as
many women born outside of the United States in the intervention group
ð5.6 percentÞthan in the comparison group ð2.4 percentÞ. There were
significant differences in DA category ðthe four categories being, in increas-
ing severity, variable danger, increased danger, severe danger, and extreme
dangerÞwith significantly more ðX
2
518.94, p<.0005Þintervention group
participants in the variable danger category ð12.7 percentÞthan comparison
group participants in the variable danger category ð3.8 percentÞ. A similar
proportion of comparison group participants ð51.8 percentÞand interven-
tion group participants ð50.4 percentÞfell into the extreme danger category,
demonstrating that any difference was primarily at the lower end of the
danger spectrum. DA category was significantly associated with immigra-
tion status ðX
2
510.07, p<.05Þ, with a greater proportion of immigrant
women at the lower end of the danger spectrum. In propensity score–
matched analyses, these variables are modeled using an interaction term,
which improves group balance but leads to small cell sizes and up to nine
cases being dropped in some analyses.
protective actions
We hypothesized that the LAP would increase women’s use of protec-
tive actions ðhypothesis 1Þ, both immediately after the police contact
ðhypothesis 1aÞand at the time of the follow-up interview ðhypothe-
sis 1bÞ. In order to examine the immediate safety strategies that the par-
ticipants engaged in ðhypothesis 1aÞ, we examined the protective actions
that participants took between the time of police contact and the baseline
interview. As shown in table 2, according to regression analyses, being in
the intervention group is associated with a significant increase in the
likelihood of having removed or hidden a partner’s weapons and having
sought formal services for domestic violence between the time when the
policerespondedtotheinitialoffenseandthetimeofthebaseline
interview. The results of the propensity score–matched analyses are
consistent with the regression results, although the parameter estimates
are attenuated. Being in the intent-to-treat group is associated with a
514 |Social Service Review
This content downloaded from 128.135.216.42 on Thu, 17 Sep 2015 14:08:51 PM
All use subject to JSTOR Terms and Conditions
smaller, but significant, increase in having sought formal services for do-
mestic violence and is not significantly associated with the likelihood of
having removed or hidden a partner’s weapons.
When examining the safety strategies that participants engaged in
between the baseline and follow-up interviews ðhypothesis 1b; see table 3Þ,
regression analyses and propensity score–matched analyses find that
women in the intervention group are more likely to obtain some form of
protection against an abusive partner, such as mace or pepper spray; apply
for and receive an order of protection; obtain medical care from a doctor or
nurse due to injuries or trauma sustained by IPV; and go someplace where
an abusive partner could not find or see them ðe.g., to stay with family or
friends; hereafter, hidingÞ. Additionally, in both regression and propensity
score–matched analyses, being in the intervention group is associated with
having a partner who went someplace where he could not find or see the
participant ðe.g., jail; hereafter, being detainedÞ. In the regression analyses
only, being in the intervention group is associated with having established a
code with family and friends to alert them of trouble and having engaged in
other safety strategies such as improving home security. Being in the intent-
to-treat group is associated with having obtained some form of protection
against an abusive partner, having hidden, and having an abusive partner
who was detained.
violent victimization
We hypothesized that the LAP would decrease the frequency and severity
of violent victimization at follow-up ðhypothesis 2Þ. As shown in table 4,
table 2. Association between Intervention Group Status and Protective Actions at Baseline
Protective Action/
Dependent Variable
Comparison
Group
N
ð%Þ
Intervention
Group
N
ð%Þ
Logistic
Regression
Analysis
ðN5681Þ;
Conditional
OR ½95% CI
Propensity
Score
Analysis
ðN5669Þ;
OR ½95% CI
Intent-to-
Treat
Analysis
ðN5884Þ;
OR ½95% CI
Removed or hid their
partner’s weapons 13 27 2.48* 1.04* 1.03
1
ð3.8%Þð7.8%Þ½1.14, 5.37½1.01, 1.08½.99, 1.07
Received services related
to domestic violence 75 106 1.79** 1.11** 1.08*
ð21.9%Þð30.5%Þ½1.25, 2.56½1.03, 1.20½1.01, 1.16
1
p<.10.
*p<.05.
** p<.01.
Oklahoma Lethality Assessment Study | 515
This content downloaded from 128.135.216.42 on Thu, 17 Sep 2015 14:08:51 PM
All use subject to JSTOR Terms and Conditions
table 3. Association between Intervention Group Status and Protect ive Act ions at Follo w-Up
Protective Action/
Dependent Variable
Comparison Group N
ð%Þ
Intervention Group N
ð%Þ
Logistic Regression
Analysis
ðN5409Þ;
Conditional OR
½95% CI
Propensity Score
Analysis
ðN5397Þ;
OR ½95% CI
Intent-to-Treat
Analysis
ðN5501Þ;
OR ½95% CI
Established a code with family and friends 84 97 1.63* 1.09 1.10
ð39.6%Þð48.0%Þ½1.07, 2.49½.95, 1.26½.97, 1.25
Obtained something to protect yourself 50 75 2.17*** 1.14* 1.14**
ð23.6%Þð37.1%Þ½1.37, 3.45½1.03, 1.27½1.04, 1.26
Engaged in other protective actions 80 90 1.54* 1.06 1.06
ð37.7%Þð44.6%Þ½1.01, 2.35½.92, 1.21½.94, 1.20
Applied for an order of protection 66 83 1.64* 1.13* 1.09
ð31.1%Þð41.1%Þ½1.07, 2.53½1.01, 1.26½.99, 1.21
Received an order of protection 50 69 1.59* 1.12* 1.07
ð23.6%Þð34.2%Þ½1.01, 2.51½1.01, 1.24½.98, 1.17
Obtained medical care due to violence 22 33 1.88* 1.08* 1.05
ð10.4%Þð16.3%Þ½1.02, 3.45½1.01, 1.16½.98, 1.11
Went somewhere partner could not find
you ðhidÞ72 82 1.61* 1.11* 1.14**
ð34.0%Þð40.6%Þ½1.04, 2.48½1.01, 1.23½1.04, 1.26
Partner went somewhere he could not
see you ðwas detainedÞ66 92 2.53*** 1.24*** 1.19**
ð31.1%Þð45.5%Þ½1.62, 3.95½1.12, 1.37½1.06, 1.33
*p<.05.
** p<.01.
*** p<.001.
This content downloaded from 128.135.216.42 on Thu, 17 Sep 2015 14:08:51 PM
All use subject to JSTOR Terms and Conditions
being in the intervention group is associated with less severe and frequent
IPV, with a subtracted CTS-2 frequency-by-severity score between 12.63
and 16.59 fewer points at follow-up. This difference is significant in the
regression, propensity score–matched, and intent-to-treat analyses. As an
example, 17 points on the severity-by-frequency scale could be translated
into a participant’s partner beating her up on two separate occasions,
punching her on an additional two occasions, and slapping her on yet
another occasion.
post hoc analyses
In order to examine whether the protective strategies that the intervention
group engaged in were associated with decreases in IPV, we conducted post
hoc one-tailed t-tests ðshown in fig. 3Þ. Among women in the intervention
group, removing or hiding their partner’s weapons, receiving formal ser-
vices for IPV, establishing a code with family and friends, applying for an
order of protection, obtaining medical care, and hiding are associated with
significantly less IPV at follow-up.
discussion
Our findings show that the Lethality Assessment Program ðLAPÞis associ-
ated with an increase in protective actions and a decrease in the frequency
and severity of violence among this sample of IPV survivors, particularly
among those who chose to speak to a hotline advocate. The majority of
protective actions engaged in by intervention group participants were asso-
ciated with decreases in the severity and frequencyof self-reported violence
at follow-up. This demonstrates a potential mechanism by which the LAP
table 4. Association between Intervention Group Status and the Frequency and Severity of
Violence at Follow-Up
Comparison
Group
Mean
ðSDÞ
Intervention
Group
Mean
ðSDÞ
Linear
Regression
Analysis
ðN5405Þ;
B½95% CI
Propensity
Score
Analysis
ðN5389Þ;
B½95% CI
Intent-to-
Treat
Analysis
ðN5497Þ;
B½95% CI
Subtracted frequency
by severity CTS-2 score 234.38 243.59 214.71 216.59 212.63
ð67.87Þð77.91Þ½228.60,2.81*½231.52, 21.65*½225.09, 2.18*
*p<.05.
Oklahoma Lethality Assessment Study | 517
This content downloaded from 128.135.216.42 on Thu, 17 Sep 2015 14:08:51 PM
All use subject to JSTOR Terms and Conditions
appears to reduce IPV risk. The LAP is a collaborative police–social service
intervention with an emerging evidence base that has the potential to
change the response to IPV and to prioritize social work goals of survivor
safety and empowerment within the context of criminal justice intervention.
Victim-survivors in the intervention and intent-to-treat groups experi-
enced less IPV at follow-up than victim-survivors in the comparison group,
but even women in the intervention and intent-to-treat groups continued to
experience violence after their contact with police officers conducting the
LAP. When women leave abusive relationships, the violence is likely to
continue ðand may even increaseÞwhenever the violent partner manages
to gain access to the survivor. Leaving is never a static event, and women
who have successfully disengaged from abusive partners have reported that
this process involves stages of reclaiming life, self, and safety ðWuest and
Meritt-Grey 2001; Messing, Mohr, and Durfee 2012Þ. As such, violence did
not cease in the short ðapproximately 7 monthsÞtime between the baseline
and follow-up interviews. It is important to note that IPV was defined, for
the purposes of this analysis, as actual acts of violence against an intimate
partner. Unlike definitions used by the Centers for Disease Control and
FIGURE 3. Subtracted CTS-2 frequency-by-severity score by protective actions taken/not
taken among women in the intervention group. *p<.05.
518 |Social Service Review
This content downloaded from 128.135.216.42 on Thu, 17 Sep 2015 14:08:51 PM
All use subject to JSTOR Terms and Conditions
Prevention ðe.g., Saltzman et al. 2002; Black et al. 2011Þ, threats of violence
were not included in our definition. Intimate partner sexual violence—a
common and devastating form of violence against women ðBagwell-Gray,
Messing, and Baldwin-White 2015Þ—was measured, but it was reported in-
frequently and therefore was not assessed separately.
Being in the intervention group ði.e., engaging in the risk assessment
process and speaking to the hotline advocateÞis associated with having
taken immediate protective actions and having engaged in a variety of
protective actions over the approximately 7 months between the baseline
and follow-up interviews. When examining protective actions, propensity
score–matched analyses find consistently smaller odds ratios than regres-
sion analyses, suggesting selection bias.That is, women who engaged in the
intervention and made the choice to speak with the hotline advocate
ðrepresented in this study by the intervention groupÞare different in mea-
sured, and likely in unmeasured ways, from the broader group of women
who came into contact with the police as part of an IPV incident
ðrepresented in this study by the comparison groupÞ. For example, women
in the intervention group may have been engaging in more protective
actions pre-intervention. Nevertheless, while the average treatment effect
on the treated is small in propensity score–matched analyses, it is statisti-
cally significant. Further, the findings over two time points demonstrate
some durability of the intervention, albeit over a relatively short time frame.
Thus, engaging in the risk assessment process and speaking to a hotline
advocate is associated with increases in a wide variety of protective actions
over multiple time periods.Women engaging in the LAP intervention sought
help from formal systems, including the social service system ðthrough the
use of formal domestic violence servicesÞand the civil court system
ðthrough applying for and receiving orders of protectionÞ. Partners of
women in the intervention and intent-to-treat groups were more likely to
be detained. This may indicate that women receiving the LAP were more
likely to engage the criminal justice system and that the system saw these
perpetrators as more dangerous. This may also partially explain the lower
levels of IPV observed among those participating in the LAP, although a
reduction in violence is not associated with having a partner who was
detained in post hoc intervention group analyses. Future research should
explicitly examine whether the LAP affects victim-survivors’engagement in
the criminal justice system and criminal justice system outcomes for their
abusive partners ðe.g., arrest, bond, conviction, and sentencingÞ.
Oklahoma Lethality Assessment Study | 519
This content downloaded from 128.135.216.42 on Thu, 17 Sep 2015 14:08:51 PM
All use subject to JSTOR Terms and Conditions
Being in the intervention group is also associated with having engaged in
several informal protective actions, demonstrating that the LAP provides a
variety of intervention strategies and that women may have chosen those
that were best suited to their situations. Women who engaged in the risk
assessment process and spoke with a hotline advocate were significantly
more likely to remove or hide their partners’weapons, although only a small
proportion of women used this protective strategy. Given that the majority
of intimate partner homicides of women are committed with a firearm in
Oklahoma and nationally and, further, given that Oklahoma state law does
not prohibit known domestic violence offenders from possessing guns ðstate
law only prohibits the purchase of firearms by those who have been arrested
for domestic violence or by those with an order of protection against themÞ,
this is an important safety strategy. Prohibiting the possession of firearms by
known domestic violence offenders reduces intimate partner homicides
ðVigdor and Mercy 2006; Zeoli and Webster 2010Þ. Future research should
examine why some protective strategies were used more by women in the
intervention group and whether this is a function of the safety-planning
strategies suggested by hotline advocates, a function of survivor choice
among myriad strategies offered, or a function of the police visit being more
effective when the LAP is used.
In addition to future experimental research conducted in additional
geographic locations, perhaps employing random assignment to control and
treatment conditions, mixed-methods research is important to understand
and contextualize the findings presented here.Qualitative and observational
research could play a role in understanding implementation fidelity and
provide insight into women’s experiences of the LAP. Important questions
remain about the consistency of police officers’implementation of the LAP
and whether officers implement the intervention differently based on fea-
tures of the situation, the offender, or the victim-survivor. The influence of
officer and advocate characteristics ðe.g., gender, empathy, and cooperationÞ
should be examined in future studies, as should police and advocate percep-
tions of the LAP. Survivors could provide information about their percep-
tions of the utility of the Lethality Screen and telephone call, as well as
their experiences of adopting ðor not adoptingÞsafety strategies post-LAP.
An exploration of how the intervention could be improved is also war-
ranted. The LAP provides an opportunity for victim-survivors to engage in
an advocacy intervention, but women choose whether or not to respond
to the questions on the Lethality Screen, speak to the hotline advocate,
520 |Social Service Review
This content downloaded from 128.135.216.42 on Thu, 17 Sep 2015 14:08:51 PM
All use subject to JSTOR Terms and Conditions
and engage in protective actions. The LAP is intended to encourage sur-
vivors to take action for self-care and should in no way be coercive or com-
promise women’s autonomy. Understanding women’s choices would assist
in the development of interventions that could reach a broader spectrum of
victim-survivors.
strengths and limitations
This study’sfindings should be understood in the context of its limitations.
The research was conducted with a sample of women in Oklahoma and is
not representative of all women. Further, women who chose to participate in
the research study may have been different from women who chose not to
participate. Having police officers refer women to researchers was novel,
and police were trained to emphasize safety, participant choice, and collab-
oration with social service agencies, but this referral process undoubtedly
introduced selection bias.While over 70 percent of participants who were
contacted by researchers agreed to participate, many more women who
provided contact information were unable to be reached. Baseline inter-
views were retrospective and were conducted a median of 15 days after the
police visit. Participating in the LAP may have affected participants’mem-
ories or awareness in ways that the usual police response does not, thus
creating differences in the responses of the intervention, intent-to-treat, and
comparison groups that do not reflect actual differences between these
conditions.
Attrition introduced another limitation into this research study. Across
all groups, approximately 40 percent of participants dropped out between
the baseline and follow-up interviews, primarily because interviewers were
unable to reach participants on follow-up. High attrition rates are consistent
with previous research on IPV survivors ðCampbell et al. 2005Þand may
have been exacerbated in this study given that the women had experienced
high levels of violence and were at heightened risk. Nevertheless, women
who did not complete the follow-up interviews may have had different
experiences of violence and safety at the time of the follow-up than the
women who were interviewed.
Women who went to shelters were generally not able to be reached due
to confidentiality constraints. Although we had only one participant who
reported that she had been to a shelter between the baseline and follow-up
interviews, data collected by one participating advocacy agency indicate
Oklahoma Lethality Assessment Study | 521
This content downloaded from 128.135.216.42 on Thu, 17 Sep 2015 14:08:51 PM
All use subject to JSTOR Terms and Conditions
that, in a single jurisdiction, 49 women who participated in the LAPentered
shelters. It is reasonable to expect that a small but important subset of the
intervention group who were never interviewed or were lost to follow-up
had gone into shelters.
When balancing the challenges of engaging in quasi-experimental field
research against the requirements of a tightly controlled, true experimental
design, we determined that random assignment of high-risk victims of IPV
would have presented insurmountable logistical and ethical issues ðsee
Messing, Campbell, and Wilson 2015Þ. Propensity score matching was used
to match the intervention and comparison groups on observed differences,
including important measures of risk. Propensity score matchingalso guards
against bias due to differences that are unobserved but related to measured
indicators. Nonetheless, given the use of a quasi-experimental design and
historical comparison group in this study, the most concerning threat to
internal validity is the threat of history effects, as an event may have occurred
ðe.g., a high-profile domestic homicide, the closing of a local shelterÞ
between recruitment of the comparison and intervention groups that
affected research outcomes. Throughout the study, therefore, we were par-
ticularly attentive to changes in participating communities by engaging with
local police departments and domestic violence services, developing advi-
sory committees, and building relationships with state policy makers.We did
not note any events that may have affected research outcomes, yet there
were observed differences between intervention and comparison groups.
Pragmatic trials, or intervention tests in the real world, are difficult to
implement.We enlisted the cooperation of police officers and advocates in a
number of police jurisdictions in Oklahoma, including the two largest
population centers in the state. Although implementation fidelity may have
been compromised because practitioners implemented the LAP, training
and implementation of the LAP was consistent with implementation in police
departments throughout the United States. Few true or quasi-experimental
field trials of this size and scope have been conducted to examine innovative
police practices or police–social work collaborations. This study examined a
relatively well-established intervention with training and technical assis-
tance available for police departments and social service organizations in-
terested in implementing it. The sample included a substantial number of
Native American women, who are at higher risk for experiencing IPV and
intimate partner homicide ðBlack et al. 2011Þ.We also used anoutcome
522 |Social Service Review
This content downloaded from 128.135.216.42 on Thu, 17 Sep 2015 14:08:51 PM
All use subject to JSTOR Terms and Conditions
measure for repeat violence that took into account the frequency and sever-
ity of IPV, not just the presence or absence of different forms of violence.
This allowed us to offer a more nuanced picture of the changes associated
with intervention and intent-to-treat group status.
conclusion
Social workers have an ethical obligation to determine the best intervention
for each client, taking into account the best available research evidence,
practitioner knowledge, and client self-determination ðGambrill 2006Þ.The
preponderance of evidence, albeit in a quasi-experimental design with some
important limitations as discussed above, is that the LAP is associated with
increased use of safety strategies and reduced violent victimization. Social
workers collaborating with law enforcement agencies have a unique oppor-
tunity to provide survivors who interact with the police holistic interven-
tions that address a range of domains. This may include, for example,
assisting women in creating safety plans that respect their choice to sever
or remain in their relationships or providing services that respect survivors’
self-determination and take into account the many ways in which women
are connected to their intimate partners ðDavies 2009Þ. The primary role of
an advocate in the LAP intervention is to provide women with safety tips
and information on available resources; survivors may or may not choose to
make use of this information.Our findings suggest that receiving this type of
consultation enhances the chances that women will pursue formal and
informal protective strategies, thusdecreasing the IPV that they experience.
Appendix
Protective Actions
Participants could answer “yes”or “no”as to whether they had taken any of
the following actions ðMcFarlane et al. 2004Þ:
1. Hidden money, an extra set of house keys, car keys, or another
belonging or object that may help you to flee your relationship
2. Established a code with family or friends ðto let them know when you
are in troubleÞ
3. Asked neighbors to call the police if violence begins
Oklahoma Lethality Assessment Study | 523
This content downloaded from 128.135.216.42 on Thu, 17 Sep 2015 14:08:51 PM
All use subject to JSTOR Terms and Conditions
4. Removed or hidden their partner’sweapons
5. Made available paperwork such as ½their own and/or their children’s
Social Security numbers, rent and utility receipts, birth certificates,
bank account numbers, driver’s license or identification, insurance
policies or numbers ½to facilitate fleeing the relationship
6. Hid valuable jewelry
7. Hid extra money
8. Hid a bag with extra clothing
We added the following dichotomous ðyes/noÞquestions:
9. Have you applied for an order of protection/restraining order against
your partner?
10. Have you received an order of protection/restraining order against
your partner?
11. Have you received services related to domestic violence in this
relationship?
12. Have you gone someplace where your partner couldn’tfind you or
see you?
13. Has your partner been someplace where he couldn’tfind you or see
you?
14. Has there been a period of time when you didn’tseeyourpartnerfor
a while because one or both of you chose not to?
15. Have you been treated bya doctor or nurse for injuries or trauma that
your partner caused in this relationship?
16. Have you obtained something to protect yourself against your part-
ner, such as mace, pepper spray, or a weapon?
note
Jill Theresa Messing, PhD, MSW, is an associate professor in the School of Social Work at
Arizona State University. Her interest areas are intimate partner violence, femicide, risk
assessment, innovative interventions to combat intimate partner violence, andevidence-based
practice. Her work has been funded by the National Institute of Justice, the National Science
Foundation, and the National Institutes of Health.
Jacquelyn Campbell, PhD, RN, FAAN, is a professor at Johns Hopkins University’sSchoolof
Nursing. She has been the principal investigator of 12 National Institute of Justice, National
Institutes of Health, US Department of Defense, and Centers for Disease Control and
Prevention-funded research studies of intimate partner violence and homicide and has
authored more than 220 articles and seven books on the subject.
524 |Social Service Review
This content downloaded from 128.135.216.42 on Thu, 17 Sep 2015 14:08:51 PM
All use subject to JSTOR Terms and Conditions
Daniel W. Webster, ScD, MPH, is a professor of health policy and management at the Johns
Hopkins Bloomberg School of Public Health. His research interests focus on evaluating strat-
egies to reduce violence and substance abuse, especially lethal outcomes.
Sheryll Brown, MPH, is the director of the Injury Prevention Service of the Oklahoma State
Department of Health. As the director of a state injury program, her research interests are
related to all areas of injury and violence prevention, evaluation of evidence-based interven-
tions, and translating research to public health practice.
Beverly Patchell, PhD, APRN, PMH-BC, is an assistant professor at the University of Utah
College of Nursing. She has an extensive background in working with Native American
communities. Her community-based research has addressed substance abuse, interpersonal
violence, diabetes, and other chronic illnesses affecting Native American communities.
Janet Sullivan Wilson, PhD, RN, is an associate professor in the College of Nursing at the
University of Oklahoma Health Sciences Center Graduate College. Her research focuses on
evaluating family violence prevention strategies in the community.
This research was funded by the National Institute of Justice ðno. 2008-WG-BX-0002Þ. We wish to
sincerely thank all of the women who participated in this research study for speaking to our interview-
ers and sharing their stories. We also want to thank all of the police departments, domestic violence
advocacy organizations, and tribal entities that contributed to the success of this research project.
references
Abadie, Alberto, and Guido W. Imbens. 2012. “Matching on the Estimated Propensity Score.”
Working paper, National Bureau of Economic Research, Cambridge, MA. http://www.hks
.harvard.edu/fs/aabadie/pscore.pdf.
Angrist, Joshua D. 2006. “Instrumental Variables Methods in Experimental Criminological
Research: What, Why, and How.”Journal of Experimental Criminology 2:23–44.
Bachman, Ronet, and Ann L. Coker. 1995. “Police Involvement in Domestic Violence: The
Interactive Effects of Victim Injury, Offender’s History of Violence, and Race.”Violence and
Victims 10 ð2Þ:91–106.
Bagwell-Gray, Meredith, Jill Theresa Messing, and Adrienne Baldwin-White. 2015. “Intimate
Partner Sexual Violence: A Review of Terms, Definitions, and Prevalence.”Trauma,Violence,
and Abuse 16 ð3Þ:316–35.
Berk, Richard A., Sarah Fenstermaker Berk, Phyllis J. Newton, and Donileen R. Loseke. 1984.
“Cops on Call: Summoning the Police to the Scene of Spousal Violence.”Law and Society
Review 18 ð3Þ:479–98.
Black, Michele C., Kathleen C. Basile, Matthew J. Breiding, Sharon G. Smith, Mikel L.Walters,
Melissa T. Merrick, Jieru Chen, and Mark R. Stevens. 2011. “National Intimate Partner and
Sexual Violence Survey: 2010 Summary Report.”Report, Centers for Disease Control and
Prevention, Atlanta.
Oklahoma Lethality Assessment Study | 525
This content downloaded from 128.135.216.42 on Thu, 17 Sep 2015 14:08:51 PM
All use subject to JSTOR Terms and Conditions
Bledsoe, Linda, Bibhuti K. Sar, and Anita P. Barbee. 2006. “Impact of Coordinated Response to
Domestic Violence on Offender Accountability.”Journal of Aggression, Maltreatment, and
Trauma 13 ð1Þ:111–31.
Bonomi, Amy E., Victoria L. Holt, Diane P. Martin, and Robert S.Thompson. 2006.“Severity of
Intimate Partner Violence and Occurrence and Frequency of Police Calls.”Journal of
Interpersonal Violence 21 ð10Þ:1354–64.
Brookoff, Daniel, Kimberly K. O’Brien, Charles S. Cook, Terry D. Thompson, and Charles
Williams. 1997. “Characteristics of Participants in Domestic Violence.”JAMA: The Journal
of the American Medical Association 277 ð17Þ:1369–73.
Burke, Jessica G., J. Denison, Andrea Carlson Gielen, Karen A. McDonnell, and Patricia
O’Campo. 2004. “What Helps Women EndAbuse in Intimate Relationships? An Application
of the Transtheoretical Model.”American Journal of Health Behavior 28:122–33.
Campbell, Jacquelyn C. 2004. “Helping Women Understand Their Risk in Situationsof Intimate
Partner Violence.”Journal of Interpersonal Violence 19 ð12Þ: 1464–77.
Campbell, Jacquelyn C., Chris O’Sullivan, Janice Roehl, and Daniel W. Webster. 2005. “Inti-
mate Partner Violence Risk Assessment Validation Study: The RAVE Study.”Final report to
the National Institute of Justice, NCJ, 209731–32. National Institute of Justice, Washington,
DC.
Campbell, Jacquelyn C., Linda Rose, Joan Kub, and Daphne Nedd. 1998. “Voices of Strength and
Resistance: A Contextual and Longitudinal Analysis of Women’s Responses to Battering.”
Journal of Interpersonal Violence 13 ð6Þ:743–62.
Campbell, Jacquelyn C., Daniel W. Webster, and Nancy Glass. 2009. “The Danger Assessment
Validation of a Lethality Risk Assessment Instrument for Intimate Partner Femicide.”
Journal of Interpersonal Violence 24 ð4Þ:653–74.
Campbell, Jacquelyn C., Daniel Webster, Jane Koziol-McLain, Carolyn Block, Doris Campbell,
Mary Ann Curry, Faye Gary, Nancy Glass, Judith McFarlane, Carolyn Sachs, Phyllis Sharps,
Yvonne Ulrich,Susan A.Wilt, Jennifer Manganello, Janet Schollenberger, Victoria Frye, and
Kathryn Laughon. 2003.“Risk Factors for Femicide in Abusive Relationships: Results from a
Multi-site Case Control Study.”American Journal of Public Health 93 ð7Þ:1089–97.
Catalano, Shannan, Erica Smith, Howard Snyder, and Michael Rand. 2009. “Female Victims of
Violence.”US Department of Justice Publications and Materials Paper 7, US Department of
Justice, Washington, DC.
Cho, Hyunkag, and Dina J. Wilke. 2010. “Gender Differences in the Nature of the Intimate
Partner Violence and Effects of Perpetrator Arrest on Revictimization.”Journal of Family
Violence 25 ð4Þ:393–400.
Coker, Ann L., Christina Derrick, Julia L. Lumpkin, Timothy E. Aldrich, and Robert Oldendick.
2000. “Help-Seeking for Intimate Partner Violence and Forced Sex in South Carolina.”
American Journal of Preventive Medicine 19 ð4Þ:316–20.
Curnow, Sally A. 1997. “The Open Window Phase: Helpseeking and Reality Behaviors by
Battered Women.”Applied Nursing Research 10 ð3Þ:128–35.
Davies, Jill. 2009. “Advocacy Beyond Leaving: Helping Battered Women in Contact with
Current or Former Partners: A Guide for Domestic Violence Advocates.”Report, Family
Violence Prevention Fund, San Francisco, CA. http://www.futureswithoutviolence.org
/userfiles/file/Children_and_Families/Advocates%20Guideð1Þ.pdf.
526 |Social Service Review
This content downloaded from 128.135.216.42 on Thu, 17 Sep 2015 14:08:51 PM
All use subject to JSTOR Terms and Conditions
Davis, Robert C., Christopher Maxwell, and Bruce Taylor. 2003. “Preventing Repeat Incidents
of Family Violence: Analysis of Data from Three Field Experiments.”Journal of Experimen-
tal Criminology 2:183–210. doi:10.1007/s11292–006–9002–0.
Davis, Robert C., and Bruce Taylor. 1997. “A Proactive Response to Family Violence: The Results
of a Randomized Experiment.”Criminology 35 ð2Þ:307–33.
Felson, Richard B., Jeffrey M. Ackerman, and Catherine A. Gallagher. 2005. “Police Interven-
tion and the Repeat of Domestic Assault.”Criminology 43 ð3Þ:563–88.
Felson, Richard B., Steven F. Messner, Anthony W. Hoskin, and Glenn Deane. 2002.“Reasons for
Reporting and Not Reporting Domestic Violence to the Police.”Criminology 40 ð3Þ:617–48.
Fischer, Karla, and Mary Rose. 1995. “When ‘Enough Is Enough’: Battered Women’s Decision
Making around Court Orders of Protection.”Crime and Delinquency 41 ð4Þ:414–29.
Gambrill, Eileen. 2006. “Evidence-Based Practice and Policy: Choices Ahead.”Research on
Social Work Practice 16 ð3Þ:338–57.
Garner, Joel H., and Christopher D. Maxwell. 2008. “Coordinated Community Responses
to Intimate Partner Violence in the 20th and 21st Centuries.”Criminology and Public Policy
7ð4Þ:525–35.
Gondolf, Edward W. 1998. Assessing Woman Battering in Mental Health Services. Thousand
Oaks, CA: Sage.
Gondolf, Edward W., and Ellen R. Fisher. 1988. Battered Women as Survivors: An Alternative to
Treating Learned Helplessness. Lexington, MA: Lexington Books.
Goodkind, Jessica R., Cris M. Sullivan, and Deborah I. Bybee. 2004. “A Contextual Analysis of
Battered Women’s Safety Planning.”Violence Against Women 10 ð5Þ:514–33.
Goodman, Lisa, Mary Ann Dutton, Kevin Weinfurt, and Sarah Cook. 2003. “The Intimate
Partner Violence Strategies Index Development and Application.”Violence Against Women
9ð2Þ:163–86.
Guo, Shenyang, and Mark W. Fraser. 2015. Propensity Score Analysis: Statistical Methods and
Applications. 2nd ed. Los Angeles: Sage.
Heckert, Alex, and Edward W. Gondolf. 2000. “Assessing Assault Self-Reports by Batterer
Program Participants and Their Partners.”Journal of Family Violence 15 ð2Þ:181–97.
Henning, Kris R., and Lisa Klesges. 2002. “Utilization of Counseling and Supportive Services
by Female Victims of Domestic Abuse.”Violence and Victims 17 ð5Þ:623–36.
Hilton, N. Zoe, Grant T. Harris, Marnie E. Rice, Ruth E. Houghton, and Angela W. Eke. 2008.
“An Indepth Actuarial Assessment for Wife Assault Recidivism: The Domestic Violence
Risk Appraisal Guide.”Law and Human Behavior 32 ð2Þ:150–63.
Hirschel, David. 2008. “Domestic Violence Cases: What Research Shows about Arrest and
Dual Arrest Rates.”Report, National Institute of Justice, Washington, DC. http://www.ojp
.usdoj.gov/nij/publications/dv-dual-arrest-222679/welcome.htm.
Hovell, Melbourne Frank, Arlene G. Seid, and Sandy Liles. 2006. “Evaluation of a Police and
Social Services Domestic Violence Program: Empirical Evidence Needed to Inform Public
Health Policies.”Violence Against Women 12 ð2Þ:137–59. doi:10.1177/1077801205277723.
Hutchison, Ira W., and J. David Hirschel. 1998. “Abused Women: Help-Seeking Strategies and
Police Utilization.”Violence Against Women 4ð4Þ:436–56.
Johnson, Ida M. 1990. “A Loglinear Analysis of Abused Wives’Decisions to Call the Police in
Domestic-Violence Disputes.”Journal of Criminal Justice 18 ð2Þ:147–59.
Oklahoma Lethality Assessment Study | 527
This content downloaded from 128.135.216.42 on Thu, 17 Sep 2015 14:08:51 PM
All use subject to JSTOR Terms and Conditions
Langford, David R. 1996. “Predicting Unpredictability: A Model of Women’s Processes of Pre-
dicting Battering Men’s Violence.”ResearchandTheoryforNursingPractice10 ð4Þ:371–85.
Macy, Rebecca J., Paula S. Nurius, Mary A. Kernic, and Victoria L. Holt. 2005. “Battered Wom-
en’s Profiles Associated with Service Help-Seeking Efforts: Illuminating Opportunities
for Intervention.”Social Work Research 29 ð3Þ:137–50.
Martin, Andrea J., Kathy R. Berenson, Sascha Griffing, Robert E. Sage, Lorraine Madry,
Lewis E. Bingham, and Beny J. Primm. 2000. “The Process of Leaving an Abusive Rela-
tionship: The Role of Risk Assessments and Decision-Certainty.”Journal of Family Violence
15 ð2Þ:109–22.
Maryland Network Against Domestic Violence ðMNADVÞ.2013.“Lethality Assessment Pro-
gram: Maryland Model ðLAPÞ—Maryland Annual Report Summary 2012.”MNADV, Lan-
ham, MD.
Maxwell, Christopher D., Joel H. Garner, and Jeffrey A. Fagan. 2001. “The Effects of Arrest on
Intimate Partner Violence: New Evidence from the Spouse Assault Replication Program.”
National Institute of Justice Research in Brief, US Department of Justice, Washington, DC.
McFarlane, Judith M., Janet Y. Groff, Jennifer A. O’Brien, and Kathy Watson. 2006.“Secondary
Prevention of Intimate Partner Violence: A Randomized Controlled Trial.”Nursing Research
55 ð1Þ:52–61.
McFarlane, Judith, Ann Malecha, Julia Gist, Kathy Watson, Elizabeth Batten, Iva Hall, and
Sheila Smith. 2004. “Increasing the Safety-Promoting Behaviors of Abused Women: In This
Study, a Telephone Intervention for Victims of Intimate-Partner Violence Showed Efficacy
for 18 Months.”American Journal of Nursing 104 ð3Þ:40–50.
Messing, Jill Theresa, Jacquelyn Campbell, and Janet Sullivan Wilson. 2015.“Research Designs
in the Real World: Testing the Effectiveness of an IPV Intervention.”National Institute
of Justice Journal, no. 275, NCJ 248781. http://www.nij.gov/journals/275/pages/research
-designs.aspx.
Messing, Jill Theresa, Jacquelyn Campbell, Janet Sullivan Wilson, Sheryll Brown, and Beverly
Patchell. 20 15. “The Lethality Screen: The Predictive Validity of an Intimate Partner Vio-
lence Risk Assessment for Use by First Responders.”Journal of Interpersonal Violence.
doi:10.1177/0886260515585540.
Messing, Jill Theresa, Rebecca Mohr, and Alesha Durfee. 2012. “Wom en’s Experiences of Grief
upon Leaving an Abusive Relationship.”Child and Family Social Work. Accessed online.
doi:10.1111/cfs.12051.
Messing, Jill Theresa, C. O’Sullivan, C. E. Cavanaugh, and Jacquelyn Campbell. Forthcoming.
“Are Abused Women’s Protective Actions Associated with Reduced Threats, Stalking, and
Violence Perpetrated by Their Male Intimate Partners?”Violence Against Women.
Messing, Jill Theresa, and Jonel Thaller. 2014. “Intimate Partner Violence Risk Assessment: A
Primer for Social Workers.”British Journal of Social Work. Accessed online. doi:10.1093
/bjsw/bcu012.
———.2013.“The Average Predictive Validity of Intimate Partner Violence Risk Assessment
Instruments.”Journal of Interpersonal Violence 28 ð7Þ: 1537–58.
Pape, Karen T., and Ileana Arias. 2000. “The Role of Perceptions and Attributions in Battered
Wom en’s Intentions to Permanently End Their Violent Relationships.”Cognitive Therapy
and Research 24 ð2Þ:201–14.
528 |Social Service Review
This content downloaded from 128.135.216.42 on Thu, 17 Sep 2015 14:08:51 PM
All use subject to JSTOR Terms and Conditions
Patzel, Brenda. 2006. “What Blocked Heterosexual Women and Lesbians in Leaving Their
Abusive Relationships.”Journal of the American Psychiatric Nurses Association 12 ð4Þ:208–
15.
Rennison, Callie M., and Sarah Welchans. 2000. “Intimate Partner Violence.”Bureau of Justice
Statistics Special Report, US Department of Justice, Washington, DC. http://www.popcenter
.org/problems/domestic_violence/ PDFs/ Rennison%26Welchans_2000.pdf.
Rosenbaum, Paul R., and Donald B. Rubin. 1983. “The Central Role of Propensity Score in
Observational Studies and Causal Effects.”Biometrika 70 ð1Þ:41–55.
Salazar, Laura F., James G. Emshoff, Charlene K. Baker, and Terrence Crowley. 2007. “Exam-
ining the Behavior of a System: An Outcome Evaluation of a Coordinated Community
Response to Domestic Violence.”Journal of Family Violence 22 ð7Þ:631–41.
Saltzman, Linda E., Janet L. Fanslow, Pamela M. McMahon, and Gene A. Shelley. 2002.
“Intimate Partner Violence Surveillance: Uniform Definitions and Recommended Data
Elements, Version 1.0.”Set of recommendations, National Center for Injury Prevention and
Control, Centers for Disease Control and Prevention, Atlanta.
Sharps, Phyllis W., Jacquelyn C. Campbell, Doris Campbell, Faye Gary, and Daniel Webster
2001. “The Role of Alcohol Use in Intimate Partner Femicide.”American Journal on Addic-
tions 10 ð2Þ:122–35.
Short, Lynn M., Pamela M. McMahon, Doryn Davis Chervin, Gene A. Shelley, Nicole Lezin,
Kira Sue Sloop, and Nicola Dawkins. 2000. “Survivors’Identification of Protective Factors
and Early Warning Signs for Intimate Partner Violence.”Violence Against Women 6ð3Þ:
272–85.
Stark, Evan. 2007. Coercive Control: The Entrapment of Women in Personal Life.OxfordNew
York: Oxford University Press.
Stover, Carla Smith. 2012. “Police-Advocacy Partnerships in Response to Domestic Violence.”
Journal of Police Crisis Negotiations 12 ð2Þ:183–98.
Straus, Murray A. 2004. “Cross-Cultural Reliability and Validity of the Revised Conflict Tactics
Scales: A Study of University Student Dating Couples in 17 Nations.”Cross-Cultural Research
38 ð4Þ:407–32.
Straus, Murray A., Sherry L. Hamby, Sue Boney-McCoy, and David B. Sugarman. 1996. “The
Revised Conflict Tactics Scales ðCTS2ÞDevelopment and Preliminary Psychometric Data.”
Journal of Family Issues 17 ð3Þ:283–316.
Straus, Murray A., and Ethel L. Mickey. 2012. “Reliability, Validity, and Prevalence of Partner
Violence Measured by theConflict Tactics Scale in Male-Dominant Nations.”Aggression and
Violent Behavior 17 ð5Þ:463–74.
Stuart, Ellen Patrice, and Jacquelyn C. Campbell. 1989. “Assessment of Patterns of Dangerous-
ness with Battered Women.”Issues in Mental Health Nursing 10 ð3–4Þ:245–60.
Truman, Jennifer L. 2011. “National Crime Victimization Survey: Criminal Victimization.”
Report, Bureau of Justice Statistics, Washington, DC.
Vigdor, Elizabeth Richardson, and James A. Mercy. 2006. “Do Laws Restricting Access to
Firearms by Domestic Violence Offenders Prevent Intimate Partner Homicide?”Evaluation
Review 30 ð3Þ:313–46.
ViolencePolicy Center. 2013.“When Men Murder Women: An Analysis of 2011 Homicide Data.”
Report, Violence Policy Center, Washington, DC.
Oklahoma Lethality Assessment Study | 529
This content downloaded from 128.135.216.42 on Thu, 17 Sep 2015 14:08:51 PM
All use subject to JSTOR Terms and Conditions
West, Carolyn M., Glenda Kaufman Kantor, and Jana L. Jasinski. 1998. “Sociodemographic
Predictors and Cultural Barriers to Help-Seeking Behavior by Latina and Anglo American
Battered Women.”Violence and Victims 13 ð4Þ:361–75.
Wiist, William H., and Judith McFarlane. 1998. “Utilization of Police by Abused Pregnant
Hispanic Women.”Violence Against Women 4ð6Þ:677–93.
Wuest, Judith, and Marilyn Merritt-Gray. 2001. “Beyond Survival: Reclaiming Self after Leav-
ing an Abusive Male Partner.”Canadian Journal of Nursing Research 32 ð4Þ:79–94.
Zeoli, April M., and Daniel W. Webster. 2010. “Effects of Domestic Violence Policies, Alcohol
Taxes, and Police Staffing Levels on Intimate Partner Homicide in Large US Cities.”Journal
of Injury Prevention 16 ð2Þ:90–95.
530 |Social Service Review
This content downloaded from 128.135.216.42 on Thu, 17 Sep 2015 14:08:51 PM
All use subject to JSTOR Terms and Conditions
