ArticlePDF Available

Domestic Violence MARACs (Multi-Agency Risk Assessment Conferences) for Very High-Risk Victims in Cardiff, Wales: A Process and Outcome Evaluation

Domestic Violence MARACs
(Multi-Agency Risk Assessment Conferences)
for Very High-Risk Victims
in Cardiff, Wales:
A Process and Outcome Evaluation
June, 2004
Dr Amanda L Robinson
Lecturer in Criminology and Criminal Justice
School of Social Sciences
Cardiff University
Glamorgan Building
King Edward VII Avenue
Cardiff CF10 3WT
(02920) 875401
Table of Contents
Executive Summary........................................................................................................................3
List of Tables .................................................................................................................................4
List of Charts .................................................................................................................................4
Acknowledgments ..........................................................................................................................4
Background to Multi-Agency Partnerships in Cardiff..........................................................................5
Women’s Safety Unit...................................................................................................................6
South Wales Police......................................................................................................................6
Domestic Violence Unit ............................................................................................................6
Police Watch ...........................................................................................................................7
Risk Assessment......................................................................................................................7
Fast-track DV Court.....................................................................................................................8
Literature on Risk Assessment in DV Cases ......................................................................................8
Correlates of Severe Harm/Mortality.............................................................................................9
Clinical Judgment vs. Victim Intuition .........................................................................................11
Methodology ................................................................................................................................11
Process Evaluation ....................................................................................................................12
Observations .........................................................................................................................12
Outcome Evaluation ..................................................................................................................12
Police Data............................................................................................................................13
Follow-up Interviews with Victims...........................................................................................13
Analytic Strategy ...................................................................................................................13
Findings from the Process Evaluation.............................................................................................13
Workload Issues .......................................................................................................................13
What’s Done?........................................................................................................................13
Who’s There? ........................................................................................................................14
Perceptions of the Work.........................................................................................................14
Information-Sharing between Agencies ......................................................................................15
Contributing to Victims’ Safety...................................................................................................17
Raising Awareness about Children..............................................................................................18
Limits of the MARACs ................................................................................................................18
Victim Cooperation ................................................................................................................18
Resources .............................................................................................................................19
Findings from the Outcome Evaluation ..........................................................................................20
Characteristics of Very High Risk Victims ....................................................................................20
Demographic Information ......................................................................................................20
History of Domestic Violence ..................................................................................................21
Risk Factors and Their Prevalence ..........................................................................................22
Post-MARAC Violence and Abuse................................................................................................24
Police Data............................................................................................................................24
Other Data ............................................................................................................................25
Follow-up Interviews .............................................................................................................26
Agencies’ Actions ......................................................................................................................27
Conclusion ...................................................................................................................................29
Appendix A ..................................................................................................................................32
Appendix B..................................................................................................................................33
Appendix C ..................................................................................................................................34
Evaluation of the Cardiff MARACs
Executive Summary
§ The first MARAC was held in April 2003 and was attended by members of 16 agencies,
including police, probation, local authority, health, housing, refuge and the Women’s Safety
Unit. The goal of these meetings is to provide a forum for sharing information and taking
actions that will reduce future harm to very high-risk victims and their children. The MARACs
were held monthly and included about 20-30 very high-risk victims.
§ Very high-risk victims are identified from the South Wales Police (SWP) Victim Initial Risk
Indicator Form that responding officers complete at the scene of domestic violence incidents.
Scores on 15 yes/no questions enable police to assess victims’ levels of risk.
§ The evaluation of the MARACs lasted 6-months and included a process evaluation that
involved site visits and key informant interviews, and an outcome evaluation that incorporated
police data and victim interviews.
§ Results from the process evaluation showed that the respondents viewed the MARACs as
invaluable, and that the added work they undertook (which was substantial) did ‘pay off’
because agencies could assist victims more efficiently.
§ While all agencies have an important role to play in the MARACs, the information provided by
and the actions undertaken by the South Wales Police and the Women’s Safety Unit appear
particularly significant.
§ The respondents made it clear that MARACs facilitate the accomplishment of many key
objectives including information-sharing between agencies, contributing to victims’ safety,
identifying key contacts within agencies, and raising awareness about the impact of domestic
violence on children.
§ Resource/time limitations and lack of cooperation from victims were viewed as the main
barriers reducing the effectiveness of the MARACs.
§ The average victim scored ‘yes’ on about 6 of the 15 risk factors. The most frequently
reported risk factors were the partner/ex has a criminal record (84%), relationship separation
(77%), perpetrator has aggravating problems (alcohol, drugs, and/or mental health) (71%),
and perpetrator is jealous or controlling (62%).
§ Analysis of the risk factors revealed that ‘perpetrator is jealous or controlling’ is a particularly
important risk factor, as its presence makes 11 of the 14 other risk factors significantly more
likely to occur.
§ Whether MARACs were successful at improving the safety of victims was measured by the
number of police domestic violence complaints post-MARAC, the number of police call-outs for
domestic violence post-MARAC, and telephone interviews with victims. All three indicators
revealed that the majority of victims (about 6 in 10) had not been revictimized since the
MARAC. These are very positive results that reveal the benefits of taking a multi-agency
approach to helping women (and their children) who are experiencing domestic violence.
§ Four of the 15 risk factors were significantly correlated with post-MARAC domestic violence
complaints and additional police call-outs for domestic violence: perpetrator has aggravating
problems; perpetrator is jealous or controlling; relationship separation; and, abuse is
becoming worse or more frequent. At this stage, however, the risk factors should be viewed
as valuable information-gathering tools rather than for their formal predictive ability.
List of Tables
Table 1 Overview of Risk Factors
Table 2 Background Characteristics of Victims and Offenders
Table 3 Prevalence of Risk Factors Among Very High-Risk Victims
Table 4 Correlation Matrix of Risk Factors
List of Charts
Chart 1 Previous Police Domestic Violence Complaints
Chart 2 Percentage with Post-MARAC Police Domestic Violence Complaints
Chart 3 Percentage with Post-MARAC Police Domestic Violence Call-Outs
Chart 4 Percentage with Additional MARACs
Chart 5 Percentage with Other Evidence of Abuse or Problems
Chart 6 Number of Actions Taken by Each Agency
Many people made this project possible. First, thanks to Safer Cardiff for funding this research, and
to Jan Pickles and Steve Bartley for being aware that the research was needed in the first place, and
trusting me to carry it out.
Jasmin Tregidga made an important and dedicated contribution as principal field researcher, and her
insights were very much appreciated (as was her camaraderie). Thanks to Sue Lewis and Jo Silver, of
the Domestic Violence Unit, for letting us troll through your well-organized files. For the boring task
of collecting police call-out data, thanks go to Mike Brinkworth (and DCI Gaughn who approved the
task). Claire and Emma at the WSU provided us with a glimpse of ‘victims perspectives’ by
conducting the telephone interviews. These are always key.
Members of MARACs deserve thanks for sharing their experiences and opinions in a frank and honest
way. Your commitment and hard work is awe inspiring.
Evaluation of the Cardiff MARACs
In December 2002 the South Wales Police (SWP) piloted a Victim Initial Risk Indicator Form for
responding officers to complete at the scene of domestic violence incidents, as a third page to the
standard 2-page domestic violence complaint form (see Appendix A). The risk instrument, developed
jointly with the Domestic Violence Prevention Service of the NSPCC, emerged from a review of 47
domestic homicides, relevant research, and communication with other community and criminal justice
agencies. The aim of the risk indicator form is to identify serious cases of domestic violence that can
be addressed through Multi-Agency Risk Assessment Conferences (MARAC). Written protocols for the
MARACs are contained in Appendix B. They outline the purpose of the MARACs and information-
sharing procedures to be followed by attending agencies.
The MARAC process is part of the Multi-Agency Public Protection Arrangements (MAPPA) that were
introduced in the South Wales area for the management of violent and sex offenders arising from Sec
67 of the Criminal Justice and Court Services Act 1999. The first MARAC was held on 1 April 2003 and
was attended by members of 16 agencies, including police, probation, local authority, health, housing,
refuge, the NSPCC and the Women’s Safety Unit. It is expected that these meetings will occur at least
monthly and provide a forum for sharing information and taking actions that will reduce future harm
to high-risk victims and their children. At the first meeting, the circumstances of individual victims (all
women) were discussed and plans were created to help promote their safety. Representatives from
various agencies contributed information, and often this process revealed discrepancies in the
information held across agencies. For example, the SWP might have knowledge of one woman as a
repeat, high-risk victim, but South Wales Probation might not have any knowledge of her partner
being a serious domestic abuser, or he might not be on their computer system at all. Only in a multi-
agency framework can these loopholes be identified and closed.
According to the scoping exercise report,1 the original intention was that all high or very-high risk
victims would be included in the MARACs held each month. However, since approximately 20-30
victims are identified as very high-risk each month, resource limitations dictate that MARACs are held
for very high-risk victims only.2 However, it should be noted that as the MARAC process has evolved,
representatives from all agencies know that they can add a victim (whom they have concerns about)
that either did not come to police attention at all, or that did not score as very high-risk on the risk
indicator form. In this way all agencies take responsibility for identifying risk, although the police play
the leading role.
Background to Multi-Agency Partnerships in Cardiff
Over the past few years, Cardiff has been the site of much innovation in terms of providing better
service to domestic violence victims through partnerships between criminal justice agencies and the
voluntary support sector. MARACs are the newest addition to this long (and increasingly
distinguished) line of progressive programmes. These programmes are discussed in the next sections
in order to provide the reader with an understanding of what a ‘multi-agency’ approach means in
Cardiff, and should illustrate how the trusting and productive working relationships between agencies
are the vital ingredient for reducing domestic violence in the community.3
1 Authored by Sgt. Steve Bartley (SWP) and Shaun Kelly (NSPCC Children’s Services Manager), 2nd May 2003.
2 As of June 2004 MARACs are held fortnightly and about 15 victims are discussed.
3 See also Robinson AL and Pickles J (2003). Gweithio Gyva Ein Gilydd: Y Profiad Cymreig (Working
Together: The Welsh Experience) in Safe: The Domestic Abuse Quarterly.
Women’s Safety Unit
The WSU provided service to its first client on 10 December 2001, and as of 1st March 2004, 2169
women and their 3437 children have been referred to the WSU.4 The WSU provides a central point of
access for women and their children experiencing domestic violence or known-perpetrator rape in the
Cardiff area. While the overriding aim of the WSU is to help victims gain safety, the WSU team also
provides advice, advocacy, specialist counselling services, legal services, housing services, refuge
provision, target hardening and collects evidence. In this way, victims are provided with an effective,
immediate and consistent range of support services at one referral point. Through the provision of
these services, the WSU hopes to restore women’s faith in the criminal justice system in order to
improve reporting rates for DV and KPR and to reduce the level of attrition of these types of cases.
The aim of the WSU is to facilitate inter-agency cooperation to provide victims with a seamless
response to their cases. The WSU has also developed protocols with the South Wales Police and the
Crown Prosecution Service in Cardiff to provide more effective and sensitive treatment of victims, and
has engaged in a media campaign to draw attention to the prevalence of these crimes. Additionally,
the WSU provides multi-agency training to increase professionals’ understanding and awareness in
identifying risk and providing protection for women and children from DV and KPR.
People from many agencies working together conceptualised how a one-stop-shop for women
experiencing domestic violence in Cardiff should take shape. The WSU was therefore the baby of
people who had invested time and effort from the beginning. Additionally, the day-to-day working
relationships of people from the police, the CPS, the WSU, refuge and others are imbued with trust,
understanding and cooperation. Partly because many of the key players are professionals with
extensive knowledge of and many years experience within the criminal justice system, relationships go
back further than the creation of the WSU. The vision of the WSU is to bring about sustainable
change in the multi-agency arena and to help create a culture in Cardiff where domestic violence and
non-consensual sex are not accepted. It is believed that only through multi-agency partnerships can
this be accomplished.
South Wales Police
Officers are required to submit domestic violence complaints (including the risk indicator form) to the
Domestic Violence Unit before the end of their tour of duty. The definition of domestic violence
adopted by the SWP is listed on the top of the form: “any incident of violence or aggression, wherever
and whenever it occurs. The violence may include physical, sexual, emotional, or financial abuse of
an individual by a family member, partner, or ex-partner in an existing or previous relationship,
regardless of gender, culture, or sexual orientation.” These forms are not computerised.
The South Wales Police domestic violence policy states that the force is “committed to taking positive
action in all cases of domestic violence.In other words, arrest is the preferred response that officers
should take at the scene of a domestic violence incident.
Cardiff is one of one of seven BCUs (Basic Command Units) in the South Wales Police. Cardiff BCU
polices a primarily urban population of approximately 305,000. This is larger than any other BCU in
South Wales. Each month approximately 250-300 domestic violence complaints are received by police
in Cardiff.
Domestic Violence Unit
The role of the Domestic Violence Unit (DVU) of the South Wales Police is to receive domestic violence
reports from officers attending initial incidents of domestic violence, provide advice and support to
victims of domestic violence and refer them to other agencies as appropriate. Additionally, the DVU is
required: to liaise with the Crown Prosecution Service on domestic violence issues; to develop working
relationships with divisional officers; to keep statistics relating to domestic violence; to develop good
working relationships with Women’s Aid and other agencies who are able to assist (such as the WSU
4 The final evaluation report of their first year of operation may be downloaded from the author’s website at The second year report will be available in Autumn 2004.
once it was launched); and to provide information to local intelligence officers on families, particularly
where those families exhibit violence towards police officers who attend at their homes.
Police Watch
The Police Watch programme was implemented throughout Cardiff in April 2002. This police
intervention was specifically designed to reduce the level of repeat domestic violence victimisation by
providing enhanced levels of police intervention. Level 1 occurs when officers attend a domestic
violence incident when violence is used or threatened. They are to handle the incident in accordance
with force policy, submit a domestic violence report and advise both victim and offender of the Police
Watch programme. Police Watch provides enhanced police presence to both the victim and the
perpetrator via high-visibility police patrols, which are to occur twice per week in the vicinity of the
incident. Level 2 initiates at the second reported incident. Officers are supposed to reiterate the
policy at the scene. A second information letter will be sent to the victim, and a second official
warning letter will be sent to the perpetrator. The Sector Inspector arranges for a Community
Constable to visit the victim, who will encourage and help the victim complete a risk assessment form.
As part of the follow-up, the Constable is supposed to inform the victim of Police Watch, Cocoon
Watch5, and options for target hardening. Level 3 occurs after the third incident is reported. Police
Watch is increased and a Domestic Violence Officer (DVO) from the DVU will make contact with the
victim and attempt to arrange a visit. A referral will be made to the Crime Prevention Officer for
further target-hardening advice, including panic alarms, mobile phones, etc. The DVO will arrange a
multi-agency meeting to facilitate improved communication and information sharing within and
between agencies. This has now been formalized as the domestic violence MARACs.
To date the Police Watch programme has not been formally evaluated; however anecdotal evidence
suggests that it improves victims satisfaction with the police response. Additionally, Cardiff has had
continually decreasing levels of repeat victimisation over the past 2-3 years.
Risk Assessment
The SWP has been particularly innovative in the area of risk assessment in cases of domestic violence.
As mentioned previously, their risk assessment form was developed from a review of the past 47
domestic homicides investigated by the SWP, relevant research, and multi-agency partnerships. In
addition, survivors of domestic violence reviewed and approved the form. The form contains 15
yes/no questions, and an area for officer’s to note any additional information about the victim, such as
whether the victim is particularly isolated from family and/or friends, how frightened the victim is, and
what the victim believes the perpetrator will do (refer again to Appendix A). The questions were
designed with an understanding of the research on correlates of severe harm/mortality in cases of
domestic violence (see page 9).
Force policy dictates that responding officers will complete the 3-page domestic violence complaint
(also known as an FSU9) before the end of their shift. The risk indicators are designed to be asked to
victims at the scene; however it is acknowledged that some officers will instead fill out the forms
afterwards based on their understanding of the situation.
All responding officers have been trained on using the risk assessment form, and the importance of
gathering this type of information from victims, via a 20 minute training video. Over 100 copies of
this video, which link domestic violence with child protection and homicide, were made in May 2003.6
Furthermore, the DVOs have received more in-depth training on domestic violence generally which
they then pass along to responding officers in a more informal way.
5 Cocoon Watch is a police initiative that requires the help and support of neighbours, family members and
relevant agencies in protecting the victim by contacting the police immediately if further incidents of domestic
violence occur. The intervention is only implemented with the informed consent of the victim. The perpetrator
is made aware that a Cocoon Watch has been implemented.
6 Created by the SWP (DCI Public Protection Steve James and DI Steve Bartley) with Jan Pickles from the
Women’s Safety Unit.
Scoring of the risk indicator form allows victims to be categorized as ‘standard,’ ‘medium,’ ‘high’ and
‘very high’ risk. Seven or more positive responses to the risk indicator form means that DVOs will
classify the victim as very high-risk and the case will be presented at the next MARAC. However, an
officer’s observations about the victim’s level of fear, her isolation, and/or the potential for future
violence can trigger a MARAC regardless of the victim’s score. Additionally, agencies can request that
a victim is discussed at a MARAC where there has been no police involvement, although this is rare.
Fast-track DV Court
Staff of the CPS, Jan Pickles (who would eventually manage the WSU) and Martin Waygood (Clerk of
the Justice) started meeting in July 2001 to discuss how to increase the efficiency and effectiveness
with which cases of domestic violence are prosecuted in Cardiff. This group looked at the specialist
domestic violence court in Leeds as an example of good practice that could be adapted in Cardiff.
The focus of the meetings was on structuring a fast-track process whereby domestic violence cases
were flagged, prioritised and streamlined through the criminal justice system. The two primary goals
were to shorten the process time and to increase women’s safety by keeping them informed about
their cases. In October 2001 a joint memorandum of understanding was agreed between the CPS
and the WSU. The new system was adopted and in place beginning January 2002.
The Cardiff Fast-Track System was one of five sites in England and Wales that was independently
evaluated by the author and others as part of the CPS-funded Domestic Violence Project. The report
was launched in March 2004 and includes detailed qualitative and quantitative information about the
five sites, and can be found on the CPS website.7 It should also be noted that special protocols for
handling domestic violence cases have been agreed and in place in the Cardiff Crown Court since
January 2003.
In conclusion, it can be seen that multi-agency work and trusting partnerships between criminal
justice and voluntary sector agencies is not only pervasive, but common practice in Cardiff.
Literature on Risk Assessment in DV Cases
‘Risk assessment’ in cases of domestic violence victims means trying to identify those victims who are
most at risk of experiencing violence in the future. Accurate risk assessments serve several important
objectives. First, risk assessment tools (such as that used by the South Wales Police) can provide a
structured way for responding officers to gather detailed and relevant information from victims. This
information, particularly when shared with other agencies, can help provide better service to victims
because their specific needs are identified. For example, some victims will require more intensive
advocacy or support to proceed with a case or leave their partner than will others. It also provides an
enhanced ‘paper-trail’ of evidence should the victim chose not to go forward with the case, helping
prosecutors make more informed decisions about whether to proceed with cases when victims retract,
for example in the ‘public interest.’
A second related reason is that risk assessment can help save scarce criminal justice resources, by
helping identify those victims in particularly dire situations that will require more intensive assistance
from police or other agencies. It is hoped that expending more resources or effort ‘up-front’ for these
victims will pay off in terms of preventing future incidents, as it is well known that incidents of
domestic violence tend to escalate in severity over time. It is for this reason that some view risk
assessment in cases of domestic violence as ‘homicide prevention.’
Third, when embedded within multi-agency frameworks risk assessment helps more agencies become
aware of the most dangerous offenders, helping to keep their workers safe. Typically, police and
maybe probation would be the only agencies privy to this information. By sharing risk assessment
information it is possible to keep health visitors and others aware of households where they could be
at higher risk of harm, such as those where weapons are present.
7 The web address is
The next section reviews the existing research on factors that are frequently shown to lead to future
violence, more severe harm, and mortality. This provides the necessary background information with
which to judge the relative importance of those questions included on the risk assessment form used
by the SWP. It also reinforces that this was not a ‘blind’ exercise taken by the police, but that the risk
assessment tool was developed with the understanding of existing research. The section following
discusses the debate over the accuracy of clinical judgment versus victim intuition, as it provides the
context surrounding many of the tough decisions made by criminal justice officials for example
when do victims know what’s best for them and their children, and when are they minimizing
potentially fatal abuse?
Correlates of Severe Harm/Mortality
Goodman et al (2000) explored the utility of the Danger Assessment Scale8 (DAS) in predicting repeat
abuse among a sample of battered women seeking help from the criminal justice system. This scale
is similar to that used by the SWP, as it is a simple 15-item checklist designed to elicit information
from the victim to identify those that are high risk for repeat violence or homicidal violence. The DAS
gathers information about escalation of violence, psychological abuse, threats, use of weapons, drugs
and alcohol. After 92 women in Washington DC completed the DAS as part of a court intake program,
Goodman et al (2000) were able to re-contact 49 of them 12 weeks later to assess whether they had
experienced any additional threats or violence. Results indicated that women’s scores on the DAS
significantly predicted repeat abuse. In addition, they found that the DAS was a much better
predictor of repeat abuse than the widely known Conflict Tactics Scale. Although the authors
recognize the limitations of their study, including a small sample size, 53% re-contact rate, and short
follow-up of 3-months, it does provide empirical support for the use of risk assessment checklists for
identifying victims who are at greater risk of additional violence in the short-term. However, no
attempt is made at distinguishing whether some risk factors might be better predictors of repeat
violence than others. Is it appropriate to give them all equal weighting?
Bennett et al (2000) go some way towards answering this question with their study. In particular
they wanted to assess the importance of psychological abuse as a risk indicator. Two main
components of psychological abuse are identified: dominance-isolation (e.g., demands for
subservience, isolation from resources) and emotional/verbal abuse (e.g., degrading the victim,
attacking her verbally). In their review of the literature they note that psychological abuse has been
correlated with more long-term, severe physical abuse. The theoretical rationale for this is based on
the Duluth model where perpetrators are seen to use both psychological abuse and physical violence
to maintain control over their partners. Quantitative analyses showed that psychological abuse,
especially dominance, was a strong predictor of repeat violence, even when other standard risk
factors9 were also included in the model (such as past violence, criminal record, drug abuse, etc.).
These findings support the notion that police and others working with victims of domestic violence
need to pay attention to the psychological as well as the physical abuse experienced by victims.
Threatening behaviour is another form of psychological abuse that perpetrators of domestic violence
may use to keep their partners living in fear. Perpetrators who make threats towards the victim (or
others that are important to her, such as her friends or family, or even their children) have also been
viewed as more dangerous and at risk of committing additional violence (or even homicide). Although
the empirical support for this notion is mixed, threats do precede physical attacks and have been
included in risk assessment tools that are good predictors of future violence (Hemphill et al., 1998).
One of the more well-known risk factors for severe harm or homicide is when the relationship is being
dissolved. Wilson and Daly’s (1992) research found that women are at greatest risk of homicide after
they separate from a violent partner. A related concern has to do with child contact, which also can
provide the opportunity for conflict (and violence) to continue.
8 The Danger Assessment Scale was developed by Campbell, J. C. (1986) “Assessment of risk of homicide for
battered women” Advances in Nursing Science, 8(4), 36-51.
9 For a good overview see J. C. Campbell’s (1995) Assessing Dangerousness: Violence by Sexual Offenders,
Batterers, and Child Abusers. Thousand Oaks, CA: Sage Publications.
In a recent review of the research on pregnancy and domestic violence, Jasinski (2004) argues that
the available evidence is contradictory about whether pregnancy increases a woman’s risk of domestic
violence. For example, while hospital or clinic-based studies find that pregnancy may increase a
woman’s risk, other national studies do not. However, she notes a lack of research that is specifically
designed to look at this issue. In addition, she notes that while there may not be clear-cut empirical
evidence that supports the notion that pregnancy increases a woman’s risk of domestic violence, the
anecdotal evidence from health care providers and victim advocates suggests that it is a time of
increased risk. Furthermore, it is certainly not a time of decreased risk to the woman, despite
potential injury to the foetus.
Sexual abuse is only recently being recognized as a significant component of domestic violence. It is
commonly held that sexual abuse is a risk factor that signals a troubling escalation of the abuse. The
recent evaluation of the WSU (Robinson, 2003) showed that perpetrators with previous domestic
violence complaints were much more likely to inflict sexual abuse on an intimate partner. Specifically,
about 1 in 3 perpetrators who had previous domestic violence complaints sexually abused their
partners, compared to 1 in 10 perpetrators without previous domestic violence complaints. Risk of
future violence and risk of sexual abuse perpetrated by the same person would therefore appear to be
strongly correlated.
Risk to victims is also shown to play a role in police decisions to make arrests at domestic violence
incidents. Kane (1999) modelled 668 domestic violence incidents in Boston, and found that police
were more likely to make arrests when they perceived victims to be at greater risk of further violence.
In his study, risk was measured as use of fists, feet, or a weapon against the victim. This study
reminds us that assessments of risk are made on a daily basis by professionals on the ground. The
SWP risk assessment tool can therefore be seen to enhance existing knowledge about which victims
are particularly vulnerable.
To conclude, the following table summarizes risk factors found in past research and how those are
measured on the SWP risk indicator form.
Table 1: Overview of Risk Factors
Identified in Past Research Question on the SWP Risk Indicator Form
Past physical abuse Q1 Does partner/ex-partner have a criminal record? If yes, tick
box if DV related.
Q11 Has partner/ex-partner attempted to strangle/choke past or
current partner?
Escalation of abuse Q12 Is the abuse becoming worse and/or happening more
Weapons Q3 Has the incident involved the use of weapons? If yes, does
this cause significant concern?
Unemployment Q4 Is assailant experiencing/recently experiencing financial
Alcohol and/or drug abuse Q5 Does the assailant have/had problems with alcohol, mental
health and/or drugs? (specify)
Pregnancy Q6 Is the victim pregnant?
Psychological abuse Q7 Has the assailant expressed/behaved in a jealous or
controlling way? If yes, does this cause significant concern?
Separation Q8 Has there been/going to be a relationship separation
between victim and assailant?
Q9 Is there any conflict with the partner/ex-partner over child
Threats Q10 Has partner/ex-partner ever threatened to kill anybody? If
yes, does this cause significant concern?
Q13 Has partner/ex-partner threatened/attempted suicide?
Sexual abuse Q14 Has the assailant said or done things of a sexual nature
that makes the victim feel bad or that physically hurts the victim?
Suicidal thoughts Q15 Does the victim have suicidal thoughts relating to the
The SWP risk assessment tool includes these risk factors as well as one other important one: the
victim’s own fear or concern for her own safety. At the bottom of the risk form officers are asked to
write additional comments about this issue, as well as whether the victim seems particularly isolated
(either geographically or socially). The debate about the importance and accuracy of victim
intuition is discussed below.
Clinical Judgment vs. Victim Intuition
Some authors have promoted the idea that predictions made by victims of domestic violence about
their own risk of re-victimisation must be the most accurate because they have more contact,
knowledge, and history of the abuser than anyone else (de Becker, 1997; Hart, 1994; Walker, 1984).
The cycle of violence that many victims experience over time with an abuser means that they could be
especially attuned to the ‘warning signs’ of impending violence. In other words, the daily experience
of surviving in an abusive relationship means that victims are aware of certain risk factors.
On the other hand, there are several logical reasons why victims might make inaccurate predictions
about the likelihood of future violence. Psychological abuse and trauma may make them desensitized
to the actual danger they face (Campbell, 1995). Additionally, the very nature of psychological abuse
means that women are faced with degrading verbal attacks on their character and intellect. Some
women may internalize these repeated attacks and subsequently doubt their own judgment. Dutton
and Dionne (1991) also make the persuasive argument that women who (for whatever reason)
choose stay with an abuser may minimize the danger in order to cope with their circumstances.
Weisz et al (2000) empirically assessed the validity of these competing arguments by documenting the
accuracy of victims’ predictions of re-assault compared to risk factors supported by previous research.
Using data from 177 victims, they test three hypotheses:
§ Predictions of DV can be best made by the victims themselves;
§ Predictions of DV are most accurately made statistically using many factors;
§ Predictions of DV are most accurate using a combination of the above methods.
Victims were asked during an initial interview “How likely would you say it is that your partner will
become violent with you during a dispute in the next year?” and they could respond using a scale
where 0=no chance of this happening through 10=sure to happen. They were re-interviewed 4-
months later.
While analysis of the data supported the last hypothesis that many risk factors (including victim’s
own predictions) are necessary to accurately predict future domestic violence the victim’s own
prediction was the strongest correlate with future violence.10 This validates the SWP policy of
incorporating victims’ level of fear into their risk assessment tool. Indeed, the final question on the
risk assessment tool is the key indicator in that, regardless of the victim’s overall score, if she is
extremely frightened for her own safety or that of her children then a MARAC is triggered.
The evaluation period was 6-months in duration (Oct 2003-Mar 2004). The evaluation includes two
components: a process evaluation and an outcome evaluation. Each is describe in the sections that
10 As an interesting aside, research conducted by Martin et al (2000) found that women’s predictions of their risk
of returning to an abusive relationship (rather than predictions of future violence) were “biased by unrealistic
optimism.” In other words, they were not good at recognizing their own vulnerability for returning to the abuser,
even though they viewed it as a problem for “most battered women.” Given that returning to an abusive
relationship would put a woman at higher risk for future violence, the authors urge advocates and support
workers to carefully explore the ambivalent feelings that women might have about their abusers, as the process
of leaving an abusive relationship is usually a complex, ongoing process.
Process Evaluation
The process evaluation aimed to identify how the MARACs draw on the various abilities of the agency
representatives to create harm reduction strategies for victims of domestic violence. Interviewing
participants from the various agencies and observing six monthly MARACs provided the data
necessary to reveal the strengths (and limitations) of multi-agency partnerships. For example, how do
the various agencies share information, cooperate, and develop proactive strategies as a team? This
part of the evaluation also identified hurdles that reduce the effectiveness of the MARACs, leading to
recommendations for future efforts to prevent harm to high-risk victims and their children.
Six monthly MARAC meetings were observed. The meetings typically last from 9am to 2pm. Notes
were taken of the background information of the cases, as well as what actions were agreed.
Typically the format would be for the police representative (either based in the DVU or WSU) would
review the relevant details of recent incidents of domestic violence. Then other agencies would share
any relevant information, after which actions would be agreed and assigned to specific agencies to
carry out.
The actions agreed at the MARACs were coded into an SPSS database. The database was organized
so that the type of actions typically assigned to each agency were coded in order to enable an
understanding of how the workload is distributed across agencies, as well as what type of actions are
commonly undertaken by each agency. Attendance registers also provided details of which agencies
attended the MARACs.11
An interview schedule was devised to structure interviews with key informants, and is contained in
Appendix C. It provides an opportunity for participants to comment on the strengths and weaknesses
of the MARACs, describe examples of how MARACs assisted victims in specific cases, and to comment
on the key ‘ingredients’ necessary to implement a successful MARAC-type process. The interviews
typically lasted about 1 hour.
Representatives from the following agencies were interviewed:
§ Police (n=3)
§ Probation (n=1)
§ WSU (n=1)
§ Social Services (n=1)
§ Housing/Homelessness (n=2)
§ Health (n=1)
§ Women’s Aid (n=1)
§ TOTAL (n=10)
Data from the 6 monthly observations and 10 interviews were designed to complement the
quantitative data collected in the outcome component of the evaluation.
Outcome Evaluation
The aim of the outcome evaluation was to document what the MARACs are able to accomplish. Of
particular concern is to what extent these multi-agency meetings are able to reduce harm to high-risk
victims and their children. To determine whether safety was increased and fear and/or violence had
been reduced, several forms of data were collected:
§ Police files were pulled for the 6-month sample of MARAC victims.
§ Police call-outs were documented for the addresses of the 6-month sample of MARAC
§ Telephone interviews were conducted 6-months after the MARAC was held for a sub-sample
of victims (Oct 03 victims were interviewed in April 04, and Nov 03 victims were interviewed
in May 04).
11 These were unavailable for two of the six months included in the evaluation period.
Police Data
Police files were obtained for the 6-month sample of MARAC victims, or 146 women. Information
from the police complaint that ‘triggered’ the MARAC was coded into an SPSS database. This provides
information about the risk factors that responding officers identified at the scene. The dates of
previous domestic violence complaints from the same perpetrator were also noted. Additionally, any
post-MARAC incidents were identified and the one nearest in time to the MARAC was also coded in
order to provide some information about victims who did experience future violence. The overall
purpose of this exercise was to determine the level of domestic violence known to the police post-
It is well known that police may not record all domestic violence incidents, and that not all domestic
violence is reported to police. To supplement the police incident data, the number of police call-outs
to victims’ addresses over the 6-month evaluation period was also collected with the assistance of the
Incident Management Unit. This provides an overview of police attention to the victims’ homes that
may not have been recorded as ‘crime’ but is still important to document as an indication of potential
trouble, disturbance, or general stress that is present in the households.
Follow-up Interviews with Victims
To supplement the quantitative side of the outcome evaluation, interviews were also conducted with a
sub-sample of victims. The victims discussed at the October and November 2003 MARACs were
telephoned by representatives of the WSU in April and May 2004 to provide a 6-month check on their
quality of life and whether there had been any improvements in their levels of safety and security. In
total 27 women were interviewed. The interviews were semi-structured around the following
§ Are you still in a relationship with (name)? If so, how is the relationship? If not, are you in
a new relationship? How is that going?
§ If so, has there been any additional violence since October/November? If so, was it
reported? What happened?
§ Are you experiencing any emotional abuse from him?
§ How is your quality-of-life generally? How are your children?
§ Any especially significant events over the past 6 months, either positive or negative?
Analytic Strategy
The outcome evaluation provides answers to the following questions:
§ What is the average level of risk documented for MARAC victims? Is there a risk factor that is
most prevalent?
§ What proportion of victims experienced post-MARAC violence, as measured by additional
domestic violence complaints and/or police call-outs? Are any risk factors significantly related
to future (post-MARAC) violence?
§ How do the MARACs feature in victims’ experiences of domestic violence over time, e.g., if the
violence continues is it reduced in severity or becoming less frequent?
§ What are the most frequent actions agreed at the MARACs, and which agencies agree to take
In conclusion, several forms of data were collected in an attempt to provide an accurate picture of
whether the lives of victims have been improved as a result of the MARACs.
Findings from the Process Evaluation
Workload Issues
What’s Done?
The MARAC process definitely has implications for how agency representatives do their jobs. It was
universally felt that workloads had increased substantially from being involved in MARACs: all
respondents estimated an additional 2-3 days per month. Almost 1 day of this is used attending the
MARAC itself, and the other time is spent preparing for the meetings (e.g., checking the names of
victims against their own agency’s records, pulling files, etc.) or feeding back the action plans from
the MARAC to their agency (e.g., delegating responsibilities to specific workers).
To hold a MARAC requires, at a minimum, that:
§ Police identify the very high-risk victims,
§ Police circulate the details of these victims and their children to participating agencies (the
MARAC ‘list’),
§ Police pull the files for these victims and bring them to the meeting,
§ Minutes of the MARAC meeting are taken (currently by police, formerly probation), and
§ Minutes of the MARAC meeting are circulated to participating agencies.
Additionally, it is expected that:
§ All participating agencies will check the MARAC list against their own agency’s records, in
order to collate all the evidence available for the victims/perps/children,
§ Some agencies, such as the WSU, will also pull files and bring them to the meeting,
§ Representatives will take notes at the MARAC, in order to delegate actions to workers,
§ Actions agreed at the MARAC will be taken (although this is currently a matter of trust rather
than official review), and
§ Individual files held at agencies will be updated.
Who’s There?
The protocols contained in Appendix B state that the following agencies will always be invited to a
MARAC: Police, Social Services, Probation, Health, Education (where relevant). Other statutory or
voluntary agencies may also be invited depending on whether they have any specific involvement with
any of the victims (e.g., Youth Offending Teams, Community Psychiatric Nurse, NSPCC, Women’s
Safety Unit, Women’s Aid).
In practice roughly 15 people attend the MARACs. The following agencies are consistently
represented: Police, Probation, Social Services, the Women’s Safety Unit, Health, and Housing. Other
agencies that have made appearances include Women’s Aid, NSPCC, Community Midwifery, Mental
Health Psychiatrist, and the Homelessness Office . Additionally, sometimes other members from the
‘standard’ agencies will attend (such as members of the High Risk Team from Probation, Social
Workers, the Child Protection Unit at Social Services, Designated Nurse Child Protection, etc.).
The attendance of a Mental Health representative at the March 2003 meeting was viewed as a very
positive addition. Recent developments include the inclusion of an Education representative, and soon
CAFCASS are expected to join.
Perceptions of the Work
Much of the work associated with MARACs is administrative in nature and is performed in addition to
people’s everyday workloads. In many cases respondents worried that their line managers would one
day pull them off of MARACs because it was taking too much of their time… and none of the
respondents seemed to benefit from a reduction in their other duties in order to devote time to the
MARACs. One respondent put it this way:
I think the agencies involved in the MARAC are underestimating how
important the work is. They are not allowing the staff to have the time
to devote to MARAC. Agency underestimation it’s lip-service, we
recognise its valuable but we are not giving it the time or the resources
It was clear however that the respondents themselves viewed the MARACs as invaluable, and that the
added work they undertook did ‘pay off.’ There was the sense that MARACs in many ways improved
the performance of the involved agencies, in that the agencies could assist victims more efficiently.
The actual benefits of the MARACs are described in more detail in the following sections, but some
comments from respondents are useful to present here:
“Its made it much easier, when you are around the table with all the
other agencies, discussing the family, the DV issues adds strength to the
case to allocated to social workers.
“The MARACs have changed how people work if we all pulled out of
these meetings I think everyone would go back to their old ways. It does
hold people to account, even if it is only once a month.”
I feel that we are doing a much better job now because we are aware of
the cases. Prior to this we were not aware of men on our caseloads [that
were also committing DV]…. But that has changed now thank
The interviews made it clear that MARACs facilitate the accomplishment of many key objectives
§ Information-sharing between agencies,
§ Contributing to victims’ safety, and
§ Identifying key contacts within agencies.
Specific information with respect to these objectives is described in the sections that follow.
Information-Sharing between Agencies
Respondents were clear that they viewed the main ‘output’ of the MARACs as information-sharing.
This was viewed as the key ingredient necessary to provide high-risk victims of domestic violence (and
their children) the assistance they require from many agencies to be safe. This respondents
description of the process is typical:
I think it’s purely an information-sharing process, to make sure that all
the information that all the agencies are sitting on is shared. That
everyone is linking together, telling everybody the information they have
got to make the victim safer, to make the children safer, and to make the
workers safer, give them the knowledge that they are potentially walking
into a dangerous situation.
Information-sharing means that gaps in knowledge are filled so that agencies have a more holistic
idea of what is happening in a particular victim’s life, or in a particular household. Each agency is
seen to contribute a unique and valuable perspective, and in combination the information can provide
an accurate assessment of the risks faced by a victim and also how different agencies can contribute
to her safety.
“Some agencies may have snippets of info that on their own don’t raise
any particular concern, it’s only when the jigsaw of info is pieced
together that the risk factors begin to be understood.”
For this reason it is vital that non-CJS agencies are well represented at the MARACs. For example,
health visitors and others visiting the home will have a different perspective than police or probation,
which are typically only aware of criminal incidents. Victim-oriented agencies such as the WSU and
Women’s Aid are able to provide information from the victim’s perspective. Social services can
provide information about children, and take actions on their behalf. Police and probation are able to
offer information about the perpetrator’s history, and presence of other aggravating factors such as
drugs or weapons. There is a usually a wealth of information held in the community about all the
people impacted by domestic violence in a particular household, but it takes a MARAC-type process for
that information to come together in a way that can actually create meaningful difference in people’s
Some practical examples of information-sharing between agencies include:
§ Police hold pre-cons on perpetrators, enabling probation to reclassify them as high-risk and
other agencies to be aware of the potentially violent person on their caseloads.
§ Health and Social Services are able to provide information about particular needs of victims,
perpetrators, and/or their children which may be important factors to address (e.g., neglect of
children, mental health needs of victim or perpetrator). They are also able to conduct home
visits, which can provide an additional source of information about the situation.
§ Housing can provide information about the tenancy of a particular address, which can inform
whether victims need to be re-housed or whether perpetrators can be evicted and/or
§ The WSU can provide information from the victim’s perspective, for example whether her
priority is relocating to another city, following through with the prosecution of a case, or
getting help for her partner. The goals of victims ultimately influence what the MARAC can
and cannot accomplish.
Facilitating the information-sharing between agencies is that the MARACs have enabled the key
contacts in agencies to be identified. This was reiterated many times in the interviews as being
extremely beneficial for improving efficiency and effectiveness. For example:
“Before I joined I wouldn’t have had a clue how to get in touch with
probation. We now have points of contact and are on first name terms
with people in the other agencies to find out pieces of information. You
are not afraid to pick up the phone, and also people from other agencies
phone us to find things out.”
You can put faces to names that you’ve heard. NHS Trust/SS are huge
organisations, and you can get lost in them sometimes... If you have
one contact who you know can get things done, give you the information
you need, and will know what you are talking about because they were in
same meeting as you then it definitely helps.
“It means that you can have named people that you can ring up, and
they know the case you’re talking about.”
An important issue raised by respondents was that of confidentiality and how this should impact
information-sharing between agencies. Respondents recognized that this was a barrier to
information-sharing in the past, but were adamant that effective multi-agency working required
agencies to be able to share confidential information with each other. As respondents explained:
“The whole ethos is one of not being afraid to tell each other things
because at the end of the day the welfare of the victim and the children
is paramount…. In the past agencies have been reticent to divulge
confidential information… at the end of the day though, we are all
professionals working for professional organisations and we’re purely
there to help the victim.”
Its important that people don’t hide behind issues of human rights,
privacy. When we talk about human rights, we should be looking at Arts
2 and 3, the freedom to live without threat of torture, freedom of life…
We need to think about the issue of proportionality. In terms of victims,
we are looking at the critical few. You’ll find that some agencies hide
behind issues of data protection and Art 8 (right to a private life).
Agencies need to look a bit deeper at their reasons for not sharing info,
and not being prepared to work together. We looked at these issues and
understood that we couldn’t hide behind data protection. You need to
look at data protection as your reason for sharing information. Other
parts of the UK have not reached this realisation yet.
Agencies involved in the MARACs willingly share information because they trust the other
representatives sitting at the table. Trust was a theme raised frequently in the interviews. Good
working relationships are vital for the success of MARACs:
“You have different agencies all around the same table, all to the best of
their ability leaving the baggage of each agency outside the door. I don’t
think that would have been possible if it hadn’t been for the DV
Forum12there were a lot of people around the table who already knew
each other. There was a long period of time necessary for trust to build
up [before the MARACs began].”
“The agencies feel comfortable [with each other] in other areas there is
a lack of trust between agencies and thus a lack of information-sharing.”
“The MARAC process has ensured that we have much better working
relationships. We are working for the same ends. The process gives DV
the recognition it deserves, and recognizes the impact that it has on
This is an important lesson for other communities wanting or attempting to set up multi-agency
partnerships (for any purpose). Abstract ideals or goals are accomplished by working partnerships on
the ground that are imbued with trust and understanding. Only by trusting each other can the
agencies share sensitive information and jump over bureaucratic hurdles in order to provide the best
and most timely service to victims.
Contributing to Victims’ Safety
As mentioned previously, the overriding aim of the MARACs is to engage in actions that will help
protect victims and their children from further abuse. This theme was very apparent in the interviews,
as all respondents mentioned that this was the main objective of the MARACs and what they hoped to
accomplish by attending the MARACs and implementing the action plans agreed at the MARACs.
To further illustrate what ‘multi-agency working’ means in practice, respondents were asked to
comment on a specific case where they thought the MARAC was able to help a victim and/or her
children. Some selected examples are presented here:
With regard to “a victim who has suffered 13 years of horrific abuse…
through the MARAC process and discussing the issues we discovered that
the guy had breached his probation order. He turned up at probation
and we were then able to arrest him. In that way it has opened up lines
of communication with agencies whereby we have contact points within
each agency. Good result.”
Housing was contacted to find out who holds the tenancy and it turns
out to be him [the assailant]. So the housing agency can kick in then
and make him give up the property because he is breaching conditions
(code of conduct) of staying there because he is being violent to his
partner. They’ll evict him now. That agency is saying to that perpetrator
because of what is happening we will evict you. It’s not just the police
that are on to you, WE also know that your behaviour is unacceptable
and therefore you cannot stay in this house. The onus is therefore taken
off of the victim.
“The main benefit has been that we are able to put conditions on
people’s licences. If we are made aware that there are DV issues, there
can be a condition upon release from prison that states he is to have no
12 The DV Forum was resurrected in 2000.
direct/indirect contact with named (ex) partner. That is obviously very
powerful because if he breaches that on his release then we could have
him recalled. It offers much more protection to the victim. She knows
that he has got that on his licence, and it is very easy for us to do.
These are just a few examples, but should make it clear that multi-agency working is the most
effective way of helping victims of domestic violence.
Raising Awareness about Children
MARACs also have the parallel function of providing a space where the safety and needs of children
can be discussed. As respondents noted:
“It’s a DV forum, but it is good that child protection issues can be raised
so the risk to children can be highlighted and plans can be put into place
to support especially older children in families where the father has
moved on.”
“The focus on children is essential. Almost the first ¾ of cases on the list
involve children under 5, the impact that DV has on their lives, their
education potential is such a huge knock-on effect.”
“The MARAC process has made me think more about the emotional and
physical impact of DV from a child protection point of view.”
Most of the respondents’ comments about the need for representatives from education, school nurses,
etc. to attend the MARACs also reflect the belief that the welfare of children can be enhanced by the
Limits of the MARACs
Victim Cooperation
In spite of these successes, there are still limits to what can be accomplished particularly if victims do
not want assistance or do not admit there is a problem. The respondents were very clear that they
felt that the MARAC-process is effectively ‘stopped in its tracks’ by the victims themselves. For
“There’s always a certain point where we can’t do anymore if the victim
is not prepared to accept our help, still in the relationship and not
prepared to break the cycle. All we can do is keep offering the support
and monitor the situation; they need to be proactive. This is often
difficult because often they feel it is better to be in a violent relationship
that no relationship at all.”
“The MARAC is very dependent on the cooperation of the victim. If the
victim is on board and working with the agency then I think you can take
it all the way. The very nature of DV means that it takes a while for the
victim to reach the stage where they are ready to break away. This is
the stage where all the help and inter-agency support kicks in really.”
“I think if the victim hasn’t got young children in the house then there
isn’t much that you can do… if there are then the situation has to be
moved forward. Sometimes this is about giving the victim choices to
make and she has to put the children first. From a child protection
perspective, if you have a victim in denial or refusing agency help then
that automatically places the child in danger.”
Respondents placed the responsibility squarely with the victim, but do so in a way that is both realistic
and generally sympathetic to the victim (recognizing the complex and often chronic nature of the
abuse). Fundamentally, if victims want to stay with their (abusive) partners then the agencies must
accept if not respect this choice.
A widely acknowledged issue was the administrative burden imposed by the MARACs, both on
individual agencies and on the MARAC meeting itself. Drawing up the list and circulating the minutes
were viewed simultaneously as very important and very onerous tasks. For example:
Minutes need to be typed out and distributed earlier… It’s a general lack
of resources. If I want to record on our records what the actions are it
means that I would have to write up everything that happens at the
meeting. By the time that the minutes come through, I then have to go
back and write everything on. It’s a lot more work and I don’t always do
it because I just haven’t got the time.
“The amount of work that has been generated has been underestimated
by all of us, and there has been a problem with the transferral of
information, of liaising. It is the practical issues that have been difficult,
such as drawing up the list in time.”
In terms of the impact on police resources, it seems evident that the administrative burden is actually
reducing the amount of service police that the DVU can provide to the victims themselves. As one
officer explained:
A clerical member of staff would help because the majority of the
MARAC work is admin they could take minutes, do filing, send out
letters which would leave us with more time to spend with the victim,
find out exactly what is going on. We would have more time to do one-
to-one house visits because you break down more barriers when you
speak face-to-face rather than on the phone... you are able to get the
bigger picture of what is going on in their lives.
While the MARACs could not take place without the administrative work necessary to organize them, it
seems apparent that the administrative burden should not be shouldered by the same people who
need to carry out actions for and on behalf of victims and their children. Another issue that was
perceived to impact the effectiveness of the MARACs by one respondent was the sheer volume of
domestic violence cases in the community. As she noted:
“I think the biggest threat to women’s safety is the numbers we can
only look at those who are very high-risk as perceived by a risk
assessment which is a crude tool…. I think this is the biggest weakness
but there is no solution there are thousands to deal with.”
This is a particularly important issue given that resources currently permit only very-high-risk victims
being given MARAC attention. Obviously there are many high-risk victims (and their children) who
also would benefit from many agencies working together to increase their safety.
The resource issue also comes to bear on potential action plans that could assist victims and/or their
families, but are not available. For example:
“There are people who I’ve met through multi-agency partnerships that I
can contact and know that I will get feedback, an honest critique, advice
or we can solve issues. But if there is a shortage of fundamental,
practical support then we can’t do it the lack of detox beds in Cardiff is
a prime example.”
While MARACs therefore appear to be a very effective process for facilitating information-sharing that
enhances the safety of victims and their children, they cannot overcome the limits imposed by few
resources and victims who are unwilling to change their circumstances.
“I think the process is really really good. What we now need to focus on
is some outcomes what has been achieved, what hasn’t and why.”
Findings from the Outcome Evaluation
Characteristics of Very High Risk Victims
The six-month sample of MARAC victims included 146 women.13 The proportion for each month is as
§ October 2003 29 19.9%
§ November 2003 23 15.8%
§ December 2003 24 16.4%
§ January 2004 26 17.8%
§ February 2004 24 16.4%
§ March 2004 20 13.7%
§ TOTAL 146 100%
Demographic Information
Table 2 presents background information for the victims and offenders in the sample. All of the
victims included in the sample are women. Their average age is 29 years at the time of the offence,
and almost half are under 30 years old. Three quarters have children in the household. Less than
9% are from minority ethnic backgrounds. Less than 1 in 3 has paid jobs outside the home.
Table 2: Background Characteristics of Victims and Offenders.
Victim Offender
Variable Value N Percent Value N Percent
Sex Male 0 0.0 Male 145 99.3
Female 146
100.0 Female 1 0.7
Ethnicity White Euro 126
86.3 White Euro 113 77.4
Dark Euro 2 1.4 Dark Euro 3 2.1
African-Carib. 2 1.4 African-Carib. 9 6.2
Asian 5 3.4 Asian 4 2.7
Arab 2 1.4 Arab 2 1.4
Mixed 1 0.7 Mixed 0 0.0
Unknown 8 5.5 Unknown 15 10.3
Age at time of offence Under 20 15
10.3 Under 20 4 2.7
21 through 30 57
39.0 21 through 30 51 34.9
31 through 40 35
24.0 31 through 40 43 29.5
41 through 50 10
6.8 41 through 50 24 16.4
51 and Over 1 0.7 51 and Over 3 2.1
Unknown 28
19.2 Unknown 21 14.4
13 This was reduced from 164. Ten victims were involved in more than one MARAC in the 6-month period;
their first MARAC only was retained for analysis. Four male victims were excluded because the couple had a
history of domestic violence but he happened to be the victim for the incident that ‘triggered’ the MARAC. Two
other victims were excluded because their files could not be found.
Employment Status Employed* 40
27.4 Employed* 38 26.0
Unemployed 70
46.9 Unemployed 76 52.1
Other** 23
15.8 Other*** 1 0.7
Unknown 13
8.9 Unknown 31 21.2
Relationship to Victim Spouse 21 14.4
at time of Offence Ex-spouse 15 10.3
Partner 57 39.0
Ex-partner 52 35.6
Mother 1 0.7
# Children in Household 0 35
1 35
2 36
3 19
4 10
5-7 4 2.7
Unknown 7 4.8
* Indicates part-time, full-time or unknown to what extent employed.
** 21 are housewives and 2 are students. *** In prison.
Offenders are slightly older (mean age of 33 years), and more are from minority ethnic backgrounds
(12%). About one-quarter are employed. Roughly half (46%) are the ex-partners or ex-spouses of
the victims.
History of Domestic Violence
As the chart below indicates, the majority of MARAC victims (77%) have previous complaints for
domestic violence on record with the South Wales Police, and more than half (52%) had two or more.
The average victim had more than 3 previous complaints. This is most likely an underestimation of
the actual violence experienced, as the police data are limited to: 1) what is reported to and recorded
by the police, 2) complaints received by the SWP, not taking into account violence experienced in
other jurisdictions, and 3) violence committed by the current offender, not including violence from
other relationships.
Chart 1: Previous Police DV Complaints
Number of complaints
Number of women
For the previous 7 domestic violence incidents, date information was obtained. Analyzing this
information revealed that:
§ The average number of days between the most recent incident and that which triggered the
MARAC was 106, and ranged from a low of 0 to a high of 666.14 Almost 70% of victims had
less than 100 days between the most recent incident and the MARAC trigger.
§ The average number of days between the MARAC trigger and the incident most distant in
time for the victims was 576, with a low of 4 to a high of 2840.15 Almost 70% of victims had
less than 700 days between the ‘oldest’ incident and the MARAC trigger. These are relatively
short histories, indicating a sharp escalation of abuse and risk.
§ The average time between incidents was 115 days.16
Risk Factors and Their Prevalence
Table 3 presents information about the prevalence of various risk factors for the domestic violence
incident that triggered the MARAC:
Table 3: Prevalence of Risk Factors Among Very High-Risk Victims.
Number Percent
Risk Factor "yes" out of total
Partner/ex has criminal record 123 84.2%
criminal record is DV related 51 34.9%
Incident resulted in injuries 66 45.2%
this causes significant concern 26 17.8%
Incident involved use of weapons 18 12.3%
this causes significant concern 11 7.5%
Perp experiencing financial problems 33 22.6%
Perp has aggravating problems 103 70.5%
type of problem: alcohol 74 50.7%
type of problem: mental health 30 20.5%
type of problem: drugs 52 35.6%
Victim is pregnant 9 6.2%
Perp is jealous/controlling 91 62.3%
this causes significant concern 54 37.0%
Has been/going to be relationship separation 112 76.7%
Conflict over child contact 44 30.1%
Perp ever threatened to kill anybody 62 42.5%
this causes significant concern 32 21.9%
14 Ten victims had values in excess of 700 days, but these values were removed because they were skewing the
distribution of this variable.
15 The ‘oldest’ incident was taken for each victim, whether they had only 1 previous incident or 7 previous
16 This was computed by dividing the average number of days from the oldest incident to the trigger incident by
the total number of incidents (all previous incidents plus the trigger incident). Two victims with values in excess
of 700 were removed from the analysis.
Perp ever attempted strangle/choke victim 56 38.4%
Abuse becoming worse/more frequent 77 52.7%
Perp ever threatened/attempted suicide 37 23.5%
Perp ever sexually abused victim 17 11.6%
Victim has suicidal thoughts 17 11.6%
The most frequently reported risk factor was the partner/ex has a criminal record (84%), however
less than half of these were known to be DV related (35%). Relationship separation featured as an
issue for more than three-quarters of victims (77%). Aggravating problems were present in about 7
in 10 perpetrators. Jealous/controlling perpetrators were noted in more than 6 in 10 cases. On
average, victims scored ‘yes’ on about 6 out of these 15 risk factors. Overall they demonstrate the
severity of the abuse experienced by these victims.
The next table displays a correlation matrix for the 15 risk factors. This enables the significant
relationships between the risk factors to become apparent.
Table 4: Correlation Matrix of Risk Factors.
Risk Factor Crim Inj Weap Fin Aggr Preg
Jeal Sep Child Thrts
Suic Sex
Crim record 1.0
Injuries 0.0 1.0
Weapons 0.0 0.3 1.0
Financ probs 0.0 0.0 0.0 1.0
Aggr probs 0.3 0.2 0.2 0.2 1.0
Pregnant 0.0 0.1 0.0 0.1 -0.1 1.0
Jealousy 0.2 0.2 -0.1 0.2 0.3 0.1
Separation 0.3 0.0 0.0 0.1 0.3 -0.1 0.4 1.0
Child contact 0.1 -0.3 -0.2 0.1 -0.1 0.1
0.2 0.3 1.0
Threats to kill 0.1 0.1 0.0 0.0 0.1
-0.1 0.2 0.1 0.1 1.0
Choke 0.1 0.2 -0.1 -0.1 0.1
-0.1 0.4 0.2 0.0 0.3 1.0
DV worse 0.2 0.1 0.1 0.2 0.3 0.0
0.5 0.3 0.0 0.2 0.3 1.0
Perp suicide 0.1 0.0 0.1 0.2 0.1
-0.1 0.2 0.2 0.0 0.1 0.3 0.1 1.0
Sex abuse 0.0 0.0 0.0 0.1 0.1
-0.1 0.2 0.1 0.0 0.1 0.1 0.2 0.2 1.0
Vict suicidal -0.1 0.1 0.0 0.2 0.1
0.2 0.2 0.1 0.2 0.1 0.2 0.1 0.3
Boldface type indicates that the correlation is statistically significant at the .05 level.
Several important trends may be observed from the matrix. Most pronounced is that when
perpetrators are jealous and controlling, 11 of the 14 other risk factors are more likely to occur. In
other words, jealous/controlling perpetrators are also more likely to have a criminal record, to injure
the victim, to have financial problems, to have aggravating problems, to have threatened to kill, to
have choked or strangled the victim, and to have threatened suicide. They are also more likely to be
in relationships that have been or are about to separate and to have conflict over child contact.
Jealous perpetrators increase the likelihood that the abuse is become worse or more severe. They are
also related to victims having suicidal thoughts. These relationships demonstrate the importance of
this particular risk factor, as it is associated with many other behaviours or issues that have been
found to increase the likelihood of future violence or homicide. Therefore even the most basic risk
instrument should include a question about the ex/partner being jealous or controlling of the victim.
Another important trend revealed by the table above is that when the perpetrator has aggravating
problems (alcohol, drug and/or mental health issues), they are also more likely to injure the victim, to
use weapons, and to escalate the frequency or severity of the domestic violence. This mirrors
findings in the evaluation of the WSU that perpetrators with co-occurring problems are more likely to
cause serious harm to victims. Both the criminal justice system and the voluntary sector need to
devote more attention to rehabilitation for offenders for their drug, alcohol, or mental health
problems. Many victims desire treatment’ for their partners and the traditional and most common
sanctions for DV offenders (such as fines or community punishment orders) go no way towards
providing this.
Suicidal thoughts by victims were related to the perpetrator having financial problems, being jealous
or controlling, and making threats to kill. These thoughts are also related to relationship separation,
the abuse becoming worse, and having experienced sexual abuse by the perpetrator.
Pregnancy was the only risk indicator that was not related to any of the other factors. This finding
can be interpreted to mean that either 1) pregnancy is related to other risk factors (that are not
included on the SWP form and it is unclear what these would be), or that 2) pregnancy itself is truly
not a risk factor (and perhaps it is post-partum that is the more dangerous time). Given the
contradictory findings in the research, it should be further investigated.
Post-MARAC Violence and Abuse
Police Data
The charts below show that the overwhelming majority of MARAC victims did not experience any
further incidents of violence that were reported to police. Specifically, as of April 2004, 79% did not
have any additional complaints on file and 70% did not have any police call-outs for domestic
violence. Especially given the extensive histories of abuse (known to police and otherwise), these are
impressive results indicating that most victims are experiencing less violence after the MARACs.
Chart 2: Percentage with post-
MARAC Police DV Complaints
Chart 3: Percentage with post-
MARAC Police DV Call-outs
These indicators are also significantly correlated with each other, meaning that the likelihood of one
occurring increases the likelihood of the other occurring. For example,
§ 97 (66%) of victims experienced neither additional complaint/s nor call-out/s,
§ 33 (23%) of victims experienced either additional complaint/s or call-out/s, and
§ 16 (11%) of victims experienced both additional complaint/s and call-out/s.
Therefore, according to the police data 97 of the 146 women experienced no further incidents of
violence or abuse. Additionally, checking to see whether these levels of ‘safer’ women varied
according to some having a longer opportunity for revictimisation (i.e., the police data reflect a 6-
month follow-up on October 2003 MARAC victims, but a 1-month follow-up for March victims) also
produced relatively consistent findings:
§ MONTH No Complaints No Call-outs
§ October 2003 62.1% 77.8%
§ November 2003 60.9% 52.4%
§ December 2003 79.2% 62.5%
§ January 2004 88.5% 75.0%
§ February 2004 87.5% 87.0%
§ March 2004 100% 100%
Even after 6-months, the majority of victims had no complaints (62%) or call-outs (78%) on record.
Given the importance of the risk factors for identifying very high-risk victims of domestic violence,
their predictive ability was assessed. In other words, which factors were significantly related to future
violence (as measured by police call-outs and complaints)? Four of the 15 risk factors were
significantly correlated with both additional complaints and additional call-outs for DV:
§ Perpetrator has aggravating problems (alcohol, drugs, and/or mental health),
§ Perpetrator is jealous or controlling,
§ Relationship separation, and
§ Abuse becoming worse or more frequent.
However, when all of the risk factors were put into a logistic regression model simultaneously (a more
sophisticated statistical method to determine their predictive ability), the results revealed that 1) only
‘incident resulted in injuries’ was significantly related to additional police complaints and that 2) none
of the risk factors predicted police call-outs. Rather than being an indictment of the risk factors,
however, these findings point to the need of analyzing their predictive ability for the entire population
of DV victims (rather than only the very high-risk victims included in the current sample). Until that is
accomplished, we are assuming (rather than empirically documenting) that the risk factors can
identify those women most at risk of further domestic violence.
Other Data
Two other indicators were also assessed as a further check on the safety of victims post-MARACs.
First, a small number of women (7%) were subject to additional MARACs within the study period,
indicating complex and potentially worse cases that took longer for the MARAC team to address.
Second, there was evidence that some women (8%) were still having problems with the perpetrator
(that was not reported to police). For example, one woman was receiving harassing phone calls and
text messages from the perpetrator that were keeping her in fear of him. This type of information
was present on some cases files, but was not consistently documented. It points to the necessity of
asking the victims themselves to describe what is happening in their lives (see next section).
Chart 4: Percentage with
Additional MARACs
Chart 5: Percentage with Other
Evidence of Abuse or Problems
Follow-up Interviews
Follow-up interviews were conducted with 27 (out of a possible 52) MARAC victims.17 Seventeen were
interviewed from the Oct 2003 sample and 10 were interviewed from the Nov 2003 sample. These
interviews were designed to complement the police data in order to provide more robust findings
about whether victims are safer following the MARACs.
Of the victims interviewed, 60% had no additional police complaints since the MARAC, and 70% had
no further domestic violence call-outs. As a further measure of whether the safety of women has
been increased since the MARAC, victims were asked whether they had experienced any additional
violence or threats since (Oct/Nov). The majority (63%) responded that they had not. This is almost
identical to the proportion found in the police data.
Of the 10 that responded in the affirmative, however, 3 had no complaints and 4 had no police call-
outs (i.e., about one-third of incidents were ‘invisible’ in the police data). If we extend this to the
sample of 146 women, then we would need to inflate the figure of 3 in 10 women being revictimized
(according to the police data) by one-third to about 4 in 10 women. These experiences can be
considered the ‘dark figure’ of domestic violence that goes unreported to the authorities. Some
examples from the interviews include:
Despite continuing physical, financial, and mental abuse over the 6-
month period, [victim] has been afraid to call the police because she
thought Social Services would be concerned about her kids, and possibly
take them.
“Still having problems although [victim and perpetrator] have parted and
live separately. He drove there drunk and tried to smash the door in.
[Victim] is very afraid of him and suffering from agoraphobia.”
Conversely, there are women who have experienced further domestic violence complaints and/or call-
outs, but according to their interviews they are doing better than they were before the MARAC.
“No [further abuse or threats]. He has tried to win her back, and offered
to give the relationship another go, but she feels she’s given him enough
“Currently [victim] has moved in with her mum. Currently quite
stressed, afraid and depressed but is taking steps to improve her
situation. She has been to a solicitor to apply for an injunction, and is
awaiting council housing. She had an abortion because she feared
perpetrator could use the baby as a reason to keep in contact and
continue to harass and abuse her (he was previously arrested for child
cruelty against her son).”
“[Victim and perpetrator] are trying to sort things out. They are not
currently living together but [victim] feels in control of the relationship.
Apart from incident in January there hasn’t been any further abuse.”
“[Victim] has moved away to start a new life. He doesn’t know where
she lives. She has had no further abuse or threats since moving away.”
Then there are those who show up in both the police and interview data as experiencing on-going
domestic violence. While not ‘success stories,’ at least many criminal justice and community agencies
have on-going concerns for these victims and their children, and continue to monitor their situations.
17 Those that were not interviewed could not be reached after several attempts.
“He has isolated her from her friends and threatened her friends. He is
back on drugs and drinking. Only time they get peace is when he is in
prison (when he still makes threatening phone calls). [Child] is worried
about staying at the house and is on edge and anxious. He has taken on
a protective role [for the mother].”
“Although victim is in new relationship, perpetrator attacked her
boyfriend, who dropped the charges because he is so afraid. There is
on-going harassment. Victim is frightened of what he may do in
“The perpetrator phoned while [victim] was in hospital and threatened to
beat them all up. He rings still silent calls. She has panic attacks.”
“On-going physical and emotional abuse (despite the recent marriage of
the victim and perp). Perp has threatened to ‘slice her throat’ and the
victim is on anti-depressants and seen as a vulnerable witness. There
are major concerns for the children, as they have witnessed horrendous
DV (bed wetting and wetting and soiling in school).”
It should be noted that the majority of victims had ended their relationships with the perpetrator (17
out of 27 or 63%). This did not however mean that they were safe of the 10 who reported
additional violence in their interviews, half were still in relationships with the perpetrator and half were
In conclusion, the outcome evaluation found that victims were safer post-MARAC:
§ 66% of victims did not have any additional police complaints or police call-outs according to
police records, and
§ 63% of victims reported in interviews that they had not experienced any further violence or
These are very encouraging results that demonstrate the importance (and effectiveness) of taking a
multi-agency approach to reducing repeat victimisation among domestic violence victims.
Agencies’ Actions
Recall that observations of MARAC meetings were coded in order to determine the most frequent
actions taken by agency representatives. While not a perfect measure of either productivity or
effectiveness, it does enable us to paint a picture of what actually happens in a MARAC.
The first two actions for each agency were coded. These figures were then summed to determine
which agencies tended to take the most actions in relation to the 146 victims. Chart 6 displays the
Chart 6: Number of Actions taken
by Each Agency
Soc Serv
Women's Aid
The chart above shows that the two main players tend to be the WSU and the police (with the MARAC
itself charged with many tasks and MARACs have been chaired by both probation and police). Next
it is clear that social services, housing, probation and health are involved in a similar amount of cases.
As more narrowly defined agencies (offering a particular services), the NSPCC, Women’s Aid and
Midwifery are involved in few cases. This chart is not meant to demonstrate who is most important to
the MARAC process but rather to provide an indication for planning and staffing purposes as to which
agencies are tasked to carry out the bulk of the actions in support of victims. It should also be noted
that some actions take much more time and effort to carry out than do others.
What are the most common actions taken by each agency? Actions can be divided into two
categories: 1) those that involve an agency’s own (normal) duties and 2) those that involve work
between agencies. Each agency present at the MARAC tended to be tasked with both sorts of duties.
The WSU’s most common task was to make or continue contact with the victim (n=34). This was
most often done to get the most up-to-date information from the victim, or to inform her about
actions that other agencies were planning to take on her behalf. The WSU’s ability to provide
information from the victim’s perspective is invaluable to the process. The next most common action
for the WSU to take was to liaise with other agencies about the victim’s situation (n=30). For
example, to liaise with the CPS about other information that may have emerged about the offence or
to advise on whether to witness summons the victim.
The police were most likely to engage in ‘law enforcement’ type actions. For example, their most
common task was further investigation (of the original offence, bail breaches, or potential additional
charges) (n=16). They were also able to provide occurrence markers or to create incidents so that
the police in Cardiff would be more aware of particular situations (n=6). For example, an incident
was created alerting police that a particularly violent offender was due to be released from prison
(and it was felt that he would inevitably confront the victim as he had continued to threatened both
her and her new boyfriend from prison).
Similarly, probation tended to be needed for their criminal justice powers in many cases. For
example, initiating proceedings against the perpetrator (such as contacting his probation officer to
inform of recent actions that would violate his probation, to revoke his community punishment order
due to domestic violence, or to try to recall or get the perpetrator arrested) (n=8). Probation was
also likely to liaise with other agencies such as the police or mental health (n=7).
The MARAC itself was most likely to send letters to other agencies outlining the situations of particular
victims or their children (n=16). They were equally likely to add the victim to the next MARAC or to
continue to monitor the situation. This was most often done when there was either a need for more
information before a strategy could be agreed, or because the victims did not want to cooperate.
Social Services were most likely to refer cases (n=9), such as to the NSPCC or psychiatric care. They
would also assess or conference the victim (n=6) and work with the police (n=5). One issue which
emerged from the MARACs is that non-criminal justice agencies were often working (unknowingly)
with perpetrators that were very dangerous. Joint visits between agencies such as social services and
the police were instigated to increase the safety of the practitioners involved in these cases.
Housing played a very valuable role by performing its normal duties on behalf of very high-risk victims
and their children. They were most likely to alert the Tenant Support Team (n=8) or Homelessness
Team (n=4) to fast-track victims into appropriate housing. They were also likely to work with
Women’s Aid Outreach on specific cases (n=4). The agency’s ability to inform the MARAC process
about whether the perpetrator or victim (or both) have tenancy was very important in guiding the
actions that other agencies would take.
Health was most likely to liaise with the WSU (n=10). This often took the form of using the WSU’s
facility to provide a safe and confidential service to victims and/or their children, or providing
information to the WSU about health issues of the perpetrator, victim, or their children. They were
next most likely to liaise or provide a joint visit with Social Services (n=3).
It must be reiterated that the coding of these actions was a blunt instrument in terms of identifying
the type of actions in which the agencies were engaged. Only the major trends were highlighted, but
it should be remembered that each victim’s situation was treated on its own merits, and strategies
undertaken at the MARACs were individualistic to each victim and/or her children. The actions
described for each agency therefore are combined with the actions of other agencies, in a multitude
of iterations of action. The important point to take from this exercise is that each agency has
something valuable to contribute to the MARACs, either a service or a source of information which
typically no other agency can provide.
The MARACs are an important innovation in the community and criminal justice response to domestic
violence. The multiple and unique needs of victims and their children are recognized in this type of
multi-agency approach. Representatives from many agencies are undertaking substantially more work
in the hope that their actions will lead to reductions in victimizations, as well as homicide prevention.
Their efforts have been effective, and are highly commended. The results presented in this report
make it clear that 1) the MARACs should be continued for the long-term and that 2) a MARAC-type
process would benefit any community’s response to domestic violence. As one respondent noted:
Everyone is keen because nobody really knows how to risk assess. This
is very user friendly, common sense. Its all about getting people to
know what risk is. In the past women have been given dangerous advice
hide the family allowance book, pack a spare suitcase. I’ve worked
with women where hiding the passbook has meant she has received a
beating that could have killed her. As a system, failure to understand
risk has endangered people in the past. The system of risk assessment
needs to be integrated.
The MARACs in Cardiff indicate substantial dedication and progress towards integrating risk
assessment for domestic violence victims and their children into the daily practice of many criminal
justice, community and voluntary support agencies. In the hopes of furthering this progress, several
recommendations are made on the following page.
Review of Cases. This was deemed to be a very important objective by the key informants. There
was a sense that, without review, agencies could not be properly held accountable. In several
interviews respondents expressed displeasure at having to ‘trust’ that the other agencies would do the
actions agreed at the meetings. Additionally, without a review of cases people felt that they did not
know what their hard work was actually achieving (i.e., are victims becoming safer?).
On-Going Evaluation. This is linked to the point above. At the moment another 6-month evaluation is
scheduled for the same period in 2004-05. Afterwards, however, agencies need to develop a plan for
monitoring what happens to victims. An IT package is currently being developed to assist in this
endeavour. It would be housed by the South Wales Police and information on victims updated on a
regular basis.
More Administrative Support. The work performed by the respondents largely went unnoticed by their
agencies, in the sense that they were performing MARAC-work on top of their existing workloads.
This is not only bad for morale, but in the long-run will lead to fatigue and mistakes (that could
potentially impact victims and/or their children). The task of taking minutes and distributing them to
the participating agencies is particularly time-consuming. At the moment this is done by the officers
in the Domestic Violence Unit at the SWP, which seems particularly under-staffed. Given the
importance of the MARACs, and their impressive accomplishments (as evidenced in this evaluation),
all agencies need to think in a more strategic fashion about staffing and resource issues. All agencies
should make the necessary arrangements to ensure the MARACs continue in the long-term.
Monitoring Attendance. Some felt that this would add a degree of accountability that is not currently
present. Some also felt that consistency of attendees was important, in that this enabled more
trusting working relationships to develop. When possible, agencies should therefore ‘ear-mark’
particular people to attend the MARACs, and adjust their workloads accordingly. As mentioned
previously, an Education representative should be found to attend the MARACs to enhance child
Chair Rotation. This was an idea expressed by some as a way for the MARACs to be more
democratic, rather than led exclusively by criminal justice agencies. Chairing the meetings also takes
additional time and effort, a burden that could be shared more evenly. This is an issue that should be
raised for discussion.
Fortnightly MARACs. One common refrain coming from the interviews was a need for fortnightly
rather than monthly MARACs. As one respondent noted, “we need short sharp bursts of MARAC
rather than long, drawn out meetings because people do lose interest.” At the time of this writing this
change to procedure has been agreed.
Documenting Victims’ Intuition. Past research suggests that victims are good sources of ‘predictive
ability’ themselves, and that their perceptions of fear and danger must be acknowledged. While the
SWP risk indicator form has a section for officer’s observations on this issue, the information
documented by officers was of varying quality. Furthermore, some officers might comment on the
‘isolation’ issue or even any other issue they felt was important. The actual sense of what victims
thought might happen to them, or how frightened they were, was therefore not available in a majority
of cases. This risk indicator needs to be clearly and specifically documented on the risk assessment
form so that victims’ perceptions about their safety are obtained in a more reliable fashion.
Bennett, L., Goodman, L., & Dutton, M. A. (2000). Risk assessment among batterers arrested for
domestic assault. Violence Against Women, 6(11), 1190-1203.
Campbell, J. C. (1995). Assessing dangerousness: Violence by sexual offenders, batterers, and child
abusers. Thousand Oaks, CA: Sage.
De Becker, G. (1997). The gift of fear: Survival signals that protect us from violence. Boston: Little &
Dutton, M. A., & Dionne, D. (1991). Counselling and shelter for battered women. In M. Steinman
(Ed.), Woman battering: Policy responses (pp. 113-130). Cincinnati, OH: Anderson.
Goodman, L. A., Dutton, M. A., & Bennett, L. (2000). Predicting repeat abuse among arrested
batterers: Use of the Danger Assessment Scale in the criminal justice system. Journal of Interpersonal
Violence, 15(1), 63-74.
Hart, B. (1994). Lethality and dangerousness assessments. Violence Update, 4(10), 7-8.
Jasinski, J. L. (2004). Pregnancy and domestic violence: A review of the literature. Trauma, Violence &
Abuse, 5(1), 47-64.
Kane, R. J. (1999). Patterns of arrest in domestic violence encounters: Identifying a police decision-
making model. Journal of Criminal Justice, 27(1), 65-79.
Hemphill, J. F., Hare, R. D., & Wong, S. (1998). Psychopathy and recidivism: A review. Legal and
Criminological Psychology, 3, 139-170.
Martin, A. J. et al. (2000). The process of leaving an abusive relationship: The role of risk assessments
and decision-certainty. Journal of Family Violence, 15(2), 109-122.
Robinson, A. L. (2003). ). The Cardiff Women’s Safety Unit: A Multi-Agency Approach to Domestic
Violence. Final Evaluation Report. Cardiff University: School of Social Sciences. (Available online at
Walker, L. E. A. (1984). The battered woman syndrome. New York: Springer.
Weisz, A. N., Tolman, R. M., & Saunders, D. G. (2000). Assessing the risk of severe domestic violence:
The importance of survivors’ predictions. Journal of Interpersonal Violence, 15(1), 75-90.
Wilson, M., & Daly, M. (1992). Homicide. New York: Aldine de Gruyter.
Appendix A
South Wales Police
Domestic Violence Complaint (FSU9)
including Risk Indicator Form
Appendix B
MARAC Information-Sharing Protocols
Appendix C
Interview Schedule for Key Players
Time at agency:
Time working on domestic violence issues:
Please describe your involvement in the MARACs.
Have the MARACs changed how you do your job? If so, how?
Can you give an example of how actions taken by you on behalf of your agency
have contributed to a victim’s safety?
Can you give an example of how a MARAC was not able to increase a victim’s
Are you aware of any actions resulting from a MARAC that increased the
danger faced by a victim?
How much time do you typically devote to MARAC-activity each month?
What do you feel are the aims and objectives of the MARACs?
Are there limits to what the MARACs can accomplish? If so, what are these?
Do you think the MARACs reflect multi-agency partnership? How so? Why
Are there any ways that MARACs in Cardiff could be improved?
What would be the key lessons that members of another community would
need to know to successfully implement a MARAC-type process?
What resources, if any, are needed to maintain MARACs?
Do you have any other thoughts or comments about the MARACs?
... Conferences to enhance the safety of high-risk domestic violence victims and families (Robinson, 2004). ...
... Reasons given are expedited information sharing, improved referral processes, integration of working practices across organisational boundaries, access to professional expertise, and the ability to present a unified response to service users (Burnett and Appleton, 2004;Robinson, 2004;Sloper, 2004;Berry et al., 2011). Partnerships that focus on managing high risk people can potentially provide a better quality of response to service users -due to the synthesis of diverse perspectives on how to respond with the potential to target multiple causal mechanisms driving the problem -and/or a better quantity (i.e., dosage) of response to service users (Robinson and Clancy, 2021;Hester et al., 2017;Rosenbaum, 2002). ...
... Based on our understanding of the IASU's workflow processes from the stakeholder interviews (see Figure 2) and the literature on multi-agency partnerships dealing with highrisk cases (Robinson, 2004), we hypothesised that three possible causal mechanisms could bring about the first aim of MASIP, i.e. reduction in reoffending. The first of these was: ...
... This paper explores the shifts in partnership working on domestic violence and the establishment of this prevailing framework. Whilst much is known about the CCR model in general (Pence and McMahon, 1999;Pence and Shepard, 1999;Shepard and Pence, 1999) and there is a growing body of work around the trilogy of initiatives led by MARACs in particular (Cook et al., 2004(Cook et al., , 2006Coy and Kelly, 2011;Howarth et al., 2009;Robinson, 2004Robinson, , 2006Robinson, , 2009Robinson and Tredigda, 2005;Steel et al., 2011;see also McCoy et al., 2016), what is less explored is how partnership work has developed to focus principally on those victims assessed to be at the highest risk and what this means for those whose experiences remain highly risky but who are assessed to be at lesser risk. Equally, whilst much attention has been paid to the use of risk assessment as an intervention in domestic violence (Hanmer et al., 1999;Humphreys et al., 2005;Richards, 2003Richards, , 2004Robinson, 2003Robinson, , 2004, very much less time has been spent reflecting on how, within the context of partnership work, risk-assessment tools attach a very particular meaning of high-risk to domestic victimisation. ...
... Whilst much is known about the CCR model in general (Pence and McMahon, 1999;Pence and Shepard, 1999;Shepard and Pence, 1999) and there is a growing body of work around the trilogy of initiatives led by MARACs in particular (Cook et al., 2004(Cook et al., , 2006Coy and Kelly, 2011;Howarth et al., 2009;Robinson, 2004Robinson, , 2006Robinson, , 2009Robinson and Tredigda, 2005;Steel et al., 2011;see also McCoy et al., 2016), what is less explored is how partnership work has developed to focus principally on those victims assessed to be at the highest risk and what this means for those whose experiences remain highly risky but who are assessed to be at lesser risk. Equally, whilst much attention has been paid to the use of risk assessment as an intervention in domestic violence (Hanmer et al., 1999;Humphreys et al., 2005;Richards, 2003Richards, , 2004Robinson, 2003Robinson, , 2004, very much less time has been spent reflecting on how, within the context of partnership work, risk-assessment tools attach a very particular meaning of high-risk to domestic victimisation. Understanding is limited about what this means for how, in the context of this work, we intervene in domestic abuse generally and in individual cases particularly. ...
... The first MARAC was held in April 2003 in Cardiff (Robinson, 2004(Robinson, , 2006Robinson and Tredigda, 2005). MARACs have since been implemented throughout the United Kingdom. ...
Full-text available
Whilst pioneering partnership work first took place in the battered women’s or refuge movement in England and Wales, the response that came to dominate in the 1990s and 2000s mirrored that associated with crime prevention more generally and Home Office crime prevention in particular. This reflected the increasing positioning of domestic violence as ‘real crime’ and the moves at this time to view domestic violence through a ‘crime lens’. In the last 10 years or so, there has been a clear shift, with the prevailing approach now dominated by initiatives such as Multi-Agency Risk Assessment Conferences, Independent Domestic Violence Advisors and Specialist Domestic Violence Courts. These initiatives have achieved considerable success in reducing risks to high-risk victims. Yet, in doing so, they establish a very particular framework for responding to domestic violence, positioning and promoting it as high-risk victimisation and moving to see it through an ‘exceptional risk’ lens. This paper examines shifts in the partnership response to domestic violence in England and Wales. It argues that, not only are the vast majority of lower risk women excluded from the prevailing framework but, in focusing on high-risk reduction, intervention within this framework fails to address women’s complicated and often contradictory needs in relation to abuse. The prevailing partnership response rests on a notion of safety as risk cessation rather than one which prioritises expansion of women’s space for action and freedom from the legacies of abuse. It concludes that, whilst partnership has huge practical and philosophical potential as a response to domestic violence, only by seeing domestic violence through the lens of diminished possibilities and with a broader conceptualisation of safety can a partnership framework support women to achieve theirs.
... Some of these current interventions have evolved from previous programmes. For example, the WNPH program draws influence from multi-agency risk assessment conferences (see Robinson, 2004). Some components have been overturned in response to research evidence. ...
This chapter provides a critical review of research supporting what works in the policing of family violence. A problem occurs when researchers limit the concept of family harm, or equivalently, the actual range of interactions when policing family violence, they also diminish the applicability of their findings. The history of key research starting with Sherman and Berk (1984) is offered to illustrate the problem. An alternative approach uses a new Big Data paradigm, and this is presented with several evaluations undertaken in New Zealand. These evaluations use all the available data, consider large populations and are augmented with a Crime Harm Index. New Zealand’s leading-edge approach to policing family harm is deconstructed into its components and reviewed to reveal that successful practice runs ahead of strong research evidence. The 16 parts of New Zealand’s approach are placed on an array to illustrate that the impacts of interventions are multi-dimensional with impacts for both offenders and victims. When a multi-dimensional approach is considered, successful practices are those that enhance victim agency and avoid formal, structured decision-making. The array is offered to practitioners to assess their own interventions. The Big Data approach is recommended as a wayfinder to determine successful practice.KeywordsFamily violenceFamily harmEvaluationBig DataCrime Harm Index
... Accurate measurement of the trajectory between victim, perpetrator and the police provides an important source of data to measure and therefore manage demand, and also crucially enables evaluation of DVA interventions (e.g. Robinson, 2004;Hester et al., 2019). The value of evaluations is therefore reliant upon the accuracy of police data. ...
Full-text available
Targeting police resources at repeat demand may reduce overall demand. To effectively target resources at repetition requires repeat demand to be accurately measured in police data. Using domestic violence and abuse (DVA) as a case study of repeat demand, this study takes a deep dive into the data and information systems used by one police force to identify the key issues that prevent the effective measurement of repeat DVA-related demand. From observations of the police response to DVA and manual review of 325 DVA case files, four key issues are identified: (1) fragmented units of measurement across multiple information systems; (2) inconsistent recording of personal details; (3) multiple methods of identifying DVA; and (4) the embedding of information in free-text. This paper makes recommendations to improve the measurement of repeat demand in police data, with implications for police practitioners and researchers.
... typically, around 10 minutes is allocated to discuss each case(McLaughlin et al, 2014;Acheampong, 2018).Following the MARAC process' introduction (Cardiff, April 2003), several research teams have conducted evaluations in order to understand their outcomes, underlying mechanisms of change and contextual barriers to their function and effectiveness.Robinson (2004) andRobinson & Tredigda (2005) conducted a two-phase process and outcome evaluation of MARACs in South Wales between October 2003-April 2005. Researchers interviewed agency participants and observed six, monthly MARACs in order to create an explanatory model for how MARACs utilise agency knowledge and capacities to develop harm reduction ...
This chapter examines key aspects of Australia’s current national and state-based domestic and family violence (DFV) policy and practice agendas through the lens of a response to infants, children, and young people. Australia’s socio-political landscape is discussed with reference to deeply embedded gender norms and stereotypes and the lack of a holistic view of victim survivors and perpetrators of DFV through the lens of family, community, and culture. The risk and harm to infants, children, and young people experiencing DFV are situated within this milieu, and the barriers and opportunities to progressing adequate responses to their experiences are discussed. Evidence informed relational programs to support children’s healing and recovery are described along with five key features required for system-based approaches. Particular attention is given to policy and programs that address each family member including the perpetrators of violence and abuse through “all of family approaches”. A way forward is proposed drawing on a case study of state-based DFV reform to illustrate key features that underpin the potential emergence of a safe, effective, and appropriate DFV response to infants, children, and young people.KeywordsDomestic violenceInfantsChildrenYoung peoplePolicy reformIntersectional
This chapter describes the origin and evaluates the use of risk assessment for the policing of intimate partner violence (IPV) and other forms of domestic abuse. We know there is significant heterogeneity in the cases of IPV that come to the attention of the police and that there are limited police resources to provide all cases with the same level of response. Thus, the use of risk assessment tools for evaluating the priority given to cases and adjudicating the required level of intervention has become one of the most important changes that has transformed the policing of IPV. Several countries have now embraced what has come to be known as the risk-led policing of IPV. We identify three general models in which these reforms have been implemented across the world, including the growing role that machine learning and predictive policing applications. Our review of the literature suggests that there are significant challenges to be considered when implementing these processes and that there are limits to the predictive efficacy of these tools. Critically, we continue to need a better understanding of what kind of effective interventions can be deployed with victims and perpetrators at different levels of risk.KeywordsIntimate partner violenceDomestic abuseRisk assessmentPoliceMachine learningPredictive modelling
Full-text available
Objectives: In the UK, a range of support services and interventions are available to people who have experienced or perpetrated domestic and sexual violence and abuse (DSVA). However, it is currently not clear which outcomes and outcome measures are used to assess their effectiveness. The objective of this review is to summarise, map and identify trends in outcome measures in evaluations and reports of DSVA services and interventions in the UK. Design: Scoping review Data sources: MEDLINE, EMBASE, PsycINFO, Social Policy and Practice, ASSIA, IBSS, Sociological abstracts and SSCI electronic databases were searched from inception until 21st June 2022. Grey literature sources were identified and searched. Eligibility: We included randomised controlled trials, non-randomised comparative studies, pre-post studies and service evaluations, with at least one outcome relating to the effectiveness of the support intervention or service for people who have experienced and/or perpetrated DSVA. Outcomes had to be assessed at baseline and at least one more time-point, or compared to a comparison group. Charting methods: Outcome measures were extracted, iteratively thematically grouped into categories, domains and subdomains, and trends were explored. Results: 80 studies reporting 87 interventions or services were included. A total of 426 outcome measures were extracted, of which 200 were used more than once. The most commonly reported subdomain was DSVA perpetration, and cessation of abuse according to the Severity of Abuse Grid was the most common individual outcome. Analysis of temporal trends showed that the number of studies and outcomes used has increased since the 1990s. Conclusions: Our findings highlight inconsistencies between studies in outcome measurement. The increase in the number of studies and variety of measures suggests that as evaluation of these services and interventions matures, there is an increased need for a core of common, reliable metrics to aid comparability. Protocol registration:
Trained professionals and volunteers have been key partners in the various developments in support of victims in the criminal justice system. Victim advocacy work is embryonic in this tradition and advocacy services can include offering information to victims, emotional support, help finding resources and completing paperwork, and accompanying victims to court. Concerns about the criminal justice response to rape have prompted the development of a number of different initiatives including victim advocacy services. This chapter draws upon research and evaluation studies of innovative advocacy services introduced to assist victims/survivors of rape to report to the police. Whilst findings suggest advocacy support improves victims’ experience of the criminal justice process and assists sustained engagement in this process, in some cases, it also facilitated making a report of rape to the police in the first instance. The chapter explores the nature, meaning and value of advocacy work from the perspective of practitioners delivering advocacy support and victims who have accessed advocacy services when reporting rape to the police. The chapter focuses on the benefits of advocacy that extend well beyond the emotional recovery of victims; advocacy also assists engagement in the criminal justice process.
Full-text available
Psychopathy is defined by a constellation of interpersonal, affective and behavioural characteristics that should, in principle, be strongly related to risk for recidivism and violence. We reviewed the literature on The Hare Psychopathy Checklist-Revised scales (PCL-R; Hare, 1980, 1991) and recidivism. We found that the PCL-R consistently was an important predictor across inmate samples and was consistently among the best predictors of recidivism. Average correlations between the PCL-R and recidivism, weighted by their degrees of freedom, were .27 for general recidivism, .27 for violent recidivism, and .23 for sexual recidivism. Relative risk statistics at one year indicated that psychopaths were approximately three times more likely to recidivate—or four times more likely to violently recidivate—than were non-psychopaths. The correlation between general recidivism and PCL-R Factor 2 (a measure of the social deviance facet of psychopathy) was stronger than the correlation between general recidivism and PCL-R Factor 1 (a measure of the interpersonal/affective facet of psychopathy). Both PCL-R factors contributed equally to the prediction of violent recidivism. The PCL-R routinely made a significant contribution towards predicting recidivism beyond that made by key demographic variables, criminal history, and personality disorder diagnoses. Across studies, PCL-R scores were as strongly associated with general recidivism, and were more strongly associated with violent recidivism, than were actuarial risk scales designed specifically to predict reoffending. Taken together, these findings indicate that the PCL-R should be considered a primary instrument for guiding clinical assessments of risk for criminal recidivism and dangerousness.
Full-text available
The findings of this study contribute to the discussion about the best method for predicting the recurrence of severe domestic violence. The findings are from a secondary data analysis comparing the accuracy of 177 domestic violence survivors' predictions of reassault to risk factors supported by previous research. The survivors' predictions were associated with recurrence of severe violence in a bivariate analysis. These predictions also added significantly to the accuracy of established risk factors in two multivariate equations predicting severe reassault within a 4-month period. Although not all of the survivors made accurate predictions, this research supports the use of survivors' predictions as an important element that should be included in risk prediction.
Full-text available
As a result of mandatory arrest and no-drop prosecution policies, domestic violence cases are flooding courts throughout the country and overwhelming available resources. Ideally, the most dangerous offenders would be prosecuted and sentenced most aggressively and victims at the highest risk would receive the most intensive advocacy. Thus, judges, prosecutors, probation officers, and advocates are calling for empirically validated methods to assess future risk among arrested batterers. This article describes a pilot investigation of the utility of the Danger Assessment Scale for contributing to the prediction of repeat abuse. Findings suggest that the Danger Assessment Scale, administered to battered women seeking help from the justice system, can contribute significantly to the prediction of short-term abuse recurrence among arrested batterers. Implications for future research are discussed.
Full-text available
Increased attention to the problem of domestic violence has brought with it a swell in the number of cases entering the court system. Particularly when resources are limited, it is critical to assess the dangerousness of such cases. This preliminary study attempted to add to the knowledge base about risk factors for continued abuse among batterers who have been arrested for domestic assault. Specifically, we investigated the predictive power of the level of psychological abuse in the relationship. Results suggest that psychological abuse—in particular, dominance and isolation—may be a useful variable in assessing dangerousness.
Each hour, 75 women are raped in the United States, and every few seconds, a woman is beaten. Each day, 400 Americans suffer shooting injuries, and another 1,100 face criminals armed with guns. Author Gavin de Becker says victims of violent behavior usually feel a sense of fear before any threat or violence takes place. They may distrust the fear, or it may impel them to some action that saves their lives. A leading expert on predicting violent behavior, de Becker believes we can all learn to recognize these signals of the universal code of violence, and use them as tools to help us survive. The book teaches how to identify the warning signals of a potential attacker and recommends strategies for dealing with the problem before it becomes life threatening. The case studies are gripping and suspenseful, and include tactics for dealing with similar situations. People dont just snap and become violent, says de Becker, whose clients include federal government agencies, celebrities, police departments, and shelters for battered women. There is a process a s observable, and often as predictable, as water coming to a boil. Learning to predict violence is the cornerstone to preventing it. De Becker is a master of the psychology of violence, and his advice may save your life. --Joan Price A Q&A with Gavin de Becker Question: In today's world, where terror and tragedy seem omnipresent, the fear of violence never seems more heightened. Is the world a more violent place than it ever has been? Gavin de Becker : Your question contains much of the answer: today's world, where terror and tragedy seem omnipresent... The key word is seem. When TV news coverage presents so much on these topics, it elevates the perception of terrorism and tragedy way beyond the reality. In every major city, TV news creates forty hours of original production every day, most of it composed and presented to get our attention with fear.
"Assessing Dangerousness" is a . . . volume that brings together experts in the fields of health, mental health, and criminal justice with both clinical and research experience in predicting dangerousness. The introductory chapter presents the theoretical and clinical issues involved in predicting violence in general. [The book] discusses the prediction of child abuse using the Child Abuse Potential Inventory, as well as the prediction of homicide in spouse abuse, of further violence by sexual offenders, and of further assault by batterers. The contributors . . . discuss accurate measurement using tested instruments as well as th