Using restorative justice approaches to police domestic violence and abuse

Go to section... The use of restorative justice in cases of domestic abuse is highly controversial. While little is known about how restorative approaches are used by the police, recent research shows they are used on a widespread basis to respond to domestic abuse throughout England and Wales ( Westmarland et al, 2018 ). This study delves deeper, to look at 62 cases within one police force. Qualitative police data extracted from force systems were analysed to discover in what type of domestic abuse incidents restorative approaches are used, and what type of approaches are used. We argue that while the language of restorative justice was used, the spirit of restorative justice was often missing. In some cases, the police seem to be using the language of restorative justice to respond to cases they do not think would benefit from further criminal justice intervention (a ‘new NFA’ (No Further Action)). We conclude that further differentiation between different forms of domestic abuse and different forms of restorative approach is needed to enable a more informed conversation. Further, it is essential that the domestic abuse sector work more closely with restorative justice communities in order to more safely and more appropriately consider the challenges and opportunities that restorative approaches may offer.


INTRODUCTION
Go to section...The use of restorative approaches in relation to domestic abuse is highly controversial.Objections are particularly strong within the violence against women scholarly and practice communities, where many consider that restorative practices may trivialise and re-privatise domestic abuse, reversing decades of progress to ensure domestic violence is recognised as a serious criminal offence ( Coker, 2002 ).Further, many express concerns that restorative practices may re-victimise the already 'vulnerable' and endanger the safety of victim-survivors ( Acorn, 2004;Stubbs, 2002;2007 ).Cameron puts it even more starkly, arguing that introducing restorative justice in such cases is to 'gamble' with women's lives ( Cameron, 2006, 59 ).
Nonetheless, due in part to the acknowledged failings of the criminal justice system (for example, Hester, 2006;HMIC, 2014 ), innovative and alternative justice processes, including restorative justice, are increasingly being used in the policing of domestic abuse ( Westmarland et al, 2018 ).In addition, as domestic abuse is a high-volume crime with significant levels of recidivism and which has arguably been rising over the last decade ( Walby et al, 2015 ), there are increasing pressures on the police and other criminal justice agencies to manage the significant resource demands associated with domestic abuse.
In this article, we analyse the practices of one police force area in the United Kingdom which deployed restorative approaches across all offence types, including domestic abuse.Our previous work in this area detailed the widespread use of restorative approaches in policing domestic abuse -this paper provides a more detailed look at the type of cases and restorative justice practices and considers the implications not only for policing but also for restorative justice researchers.We briefly discuss research, policy and practice on restorative justice and domestic abuse and then describe the methods and analysis employed in this study.In our results, we describe the different forms of restorative approach used by the police force, as well as the nature and variety of domestic abuse cases under examination.Qualitative case examples are used to illuminate the typology data, giving a deeper, more nuanced exploration of the context of domestic abuse cases to which the police are responding using different restorative approaches.
RESEARCHING RESTORATIVE JUSTICE AND DOMESTIC ABUSE Go to section...As introduced above, concerns have been raised that restorative approaches risk women's physical safety, as well advances in societal acknowledgement of domestic abuse as a significant issue.These critiques also emphasise the very distinct nature of domestic abuse compared with many other crimes, rendering it less suitable for restorative approaches ( Stubbs, 2002;2007 ).For example, the emphasis in some restorative practices on reparation, or 'restoration', does not adequately address the primary concerns of victim-survivors of domestic abuse who emphasise safety, denunciation and validation in terms of justice outcomes ( Stubbs, 2007, 181; Lewis et al, 2000;Herman, 2005 ).It is this lack of understanding of the specificities of domestic abuse that has led Stubbs (2014) to express particular caution when examining generic restorative justice programmes.Further, while the evidence base is broadly positive for restorative justice in general ( Sherman and Strang, 2007 ), there is a lack of research investigating its specific use and impact in domestic abuse cases ( Ptacek and Frederick, 2009 ).Ptacek has commented that it is 'astonishing' there is so little research on projects involving domestic abuse and, even where there is some element of evaluation, surprise that the methods employed are 'remarkably weak' (2014, 7).Such caveats notwithstanding, the evaluations we do have reveal some interesting findings.
Coker examined 20 cases of intimate partner violence and revealed a number of benefits for the women taking part including community support both to challenge the abuser's conduct and to maintain familial ties ( Coker, 1999 ).She did, however, find a sense of coercion among some women and expressed concern that a focus on offender rehabilitation may detract from the women's interests and needs ( Coker, 1999 ).The family group conference model developed by Pennell and Burford drew on extensive community engagement including feminist antiviolence activists, offender programmes, police and criminal justice personnel ( Pennell and Burford, 2000;2002 ).Research with 32 families reported positive outcomes for the majority of families including reduced levels of abuse compared with a control group, an improvement in children's development and an extension of social support ( Pennell and Burford, 2000 ).
A study of a variety of restorative practices in New Zealand interviewed 20 victims of family violence most of which involved intimate partner abuse ( Kingi et al, 2008 ).The victim-survivors reported positive experiences, including being treated with respect and being able to express their views ( Kingi et al, 2008 ).However, analysis of 30 cases in an Austrian diversionary project found little evidence of influence on offender behaviour and a mixed reception from victimsurvivors ( Pelikan, 2002 ).A study of a 'Circles of Peace' project in Arizona found no evidence that it reduced incidents of domestic abuse or violence compared with standard domestic abuse perpetrator programmes ( Mills et al, 2013 ).Nonetheless, a subsequent study of Circles of Peace, although only involving three cases of intimate partner abuse, concluded that 'restorative processes have some ability to create positive changes in cases of intimate partner violence, when grounded in the experiences and contributions of the battered women's movement' ( Gaarder, 2015, 363 ).
Evidence on the use of restorative practices in cases of family abuse more generally is more common, though numbers involved in evaluations remain small.One project in Western Australia reported positive outcomes for its communitarian model of restorative and transformative justice, though it only involved seven domestic abuse conferences ( Goulding and Steels, 2006 ).Interestingly, across the project as a whole, there were a proportionately higher number of women offenders involved, almost half, with two thirds of the community conferences involving women offenders ( Goulding and Steels, 2006;2013 ).
In New Zealand, family group conferences focussing on youth offending include cases of family violence and, based on cases studies and small-scale evaluations, these have a positive track record of reducing re-offending and victim satisfaction ( Morris, 2002;Morris and Maxwell 2000 ).A detailed study of the restorative practices in Rotorua in New Zealand (including some domestic abuse cases) similarly found positive impacts, though the evaluation only involved interviews with six offenders and eight victim-survivors ( Tisdall et al, 2007 ).Daly and Wade's study of restorative conferences following sibling abuse, while recognising a number of limitations, revealed 'many positives, including bringing forward the impact of the offence on victims, censuring the offending, and facilitating a frank conversation among perpetrators and family members' ( Daly and Wade, 2017, 193 ).This echoes the findings of a case study involving sexual abuse by a child against another child in the family which revealed transformative outcomes for the victim-survivor ( McGlynn et al, 2012 ).
While we are not able to draw clear conclusions from this survey of the field, Ptacek comments that the studies considered above provide 'suggestions of effectiveness' (2014).
Overall, therefore, while the evidence base is growing, and we know that a variety of forms of restorative justice are being used across the spectrum of domestic abuse cases, including intimate partner violence, evaluations are only indicative of the possibilities of this justice process.Evidence of positive experiences and good practice can be identified, as well as concerns regarding safety, coercion and longer-term outcomes.However, in none of the studies identified above are the restorative approaches being undertaken by the police.In examining police practice, therefore, this study considers a new area that has recently been shown to be particularly widespread across England and Wales ( Westmarland et al, 2018 ).

POLICING DOMESTIC ABUSE IN ENGLAND AND WALES USING RESTORATIVE JUSTICE
Go to section...The term 'restorative justice' encompasses a broad range of approaches and practices, commonly defined as: 'a process whereby all parties with a stake in a particular offence come together to resolve collectively how to deal with the aftermath of the offence and its implications for the future' ( Marshall, 1999, 5 ).The 'process' of restorative justice can also take many different forms including restorative conferences which often involve a face-to-face meeting between offender, victim and supporters, facilitated by a trained practitioner, and which are most commonly associated with 'restorative justice'.There are many other forms including shuttle mediation, victim and offender mediation and sentencing circles.Restorative justice has been used for all offence types, not just 'low-level' neighbourhood crimes with which it is often associated, and it is used across youth and adult offending.It can be a diversion from the criminal justice system, adopted in parallel with, or after, criminal proceedings such as when used pre-sentence, or entirely outside the criminal justice system ( Jülich et al, 2010 ).
In the context of policing in England and Wales, the professional standards and guidelines relating to restorative approaches are set out by the National Police Chiefs Council (NPCC) and College of Policing.The NPCC (formerly the Association of Chief Police Officers (ACPO)) states that a restorative resolution involving restorative justice requires: (1) the offender to take responsibility; (2) the involvement of the victim or other affected party; (3) a structured process that establishes what has occurred and what the impact has been; and (4) an outcome that seeks to put right the harm that has been caused, or an outcome that provides other reparation that may not be directly related to the original case.On the basis of these principles, the guidance further details three different 'levels' of restorative approach as detailed below: Level 1: Refers to an instant or on-street disposal, where police officers or PCSOs use restorative skills in the course of their duties, to resolve conflict in minor crimes and incidents.In Level 1, restorative justice is conducted as an alternative to a formal criminal justice process.
Level 2: Refers to measures such as restorative justice conferences, and may involve more participants, risk assessments and seek longer-term solutions.A Level 2 restorative justice response can occur either as an alternative to criminal justice proceedings, or in addition to criminal justice proceedings, as part of a formal crime disposal.Level 2 responses take place for incidents not appropriate for Level 1 resolutions, in order to tackle more serious or persistent matters.
Level 3: Refers to resolutions that take place in addition to criminal justice proceedings, mainly post-sentence.They may occur for cases that involve serious, complex or sensitive incidents, or where offenders are being monitored by an offender management team and/or are deemed at risk of continued offending.A Level 3 resolution can be undertaken presentence, but this must be in formal conjunction with other services, such as: Probation, the Crown Prosecution Service etc. ( ACPO, 2012 ) It can be seen from this guidance that police engagement with restorative justice can take a variety of forms and take place at various stages of the criminal process.The guidance provides detail on the factors to consider when deciding whether to undertake a restorative approach, but does not specify particular levels for types of offence.
College of Policing guidance states that the use of restorative justice is 'rarely appropriate in domestic abuse cases and not recommended in cases involving intimate partner abuse' ( College of Policing, 2015, 4.2 ).It states that domestic abuse is 'among the most hazardous of cases because of the risk to victims of re-victimisation or serious violence and the potential effects of controlling or coercive behaviour'.Further, it states that all forces 'should establish a policy for any use of restorative justice in domestic abuse cases' and that any officer 'considering the use of restorative justice in a domestic abuse case must take advice from supervisors and other agency experts' ( College of Policing, 2015, 4.2 ).In addition, while the guidance acknowledges that victims can request restorative justice, it notes that it is not always suitable and the officer must, therefore, assess suitability 'based on the sensitivities and vulnerabilities of the victim, especially in domestic abuse cases' (our emphasis).In particular, the guidance warns officers that a request for restorative justice from a victim may be indicative of being under the 'influence of controlling or coercive behaviour and making the request to please or appease the perpetrator' ( College of Policing, 2015, 4.2 ).Finally, the guidance states that standard guidelines on the use of restorative justice require contact with a victim to discuss completion of agreed outcomes.This follow-up, the College suggests, should be used as an opportunity for further risk assessment ( College of Policing, 2015, 4.2 ).Despite these concerns, however, it has been known for a number of years that the police have been using restorative approaches in domestic abuse cases, including those involving intimate partners.For example, the HMIC's 2014 investigation into domestic abuse noted that 'there are a few examples from the force inspections where use of restorative justice in intimate partner abuse was happening routinely' ( HMIC, 2014, 101 ).The most recent and comprehensive study found that across all police areas in England, Wales and Northern Ireland, police are using restorative approaches and community resolutions in response to a range of domestic abuse cases -by some forces to respond to up to one in twenty domestic abuse cases ( Westmarland et al, 2018 ).Indeed, this research prompted the House of Commons Justice Select Committee and subsequently the Government to recommend that it be 'reaffirmed' that Level 1 restorative justice is inappropriate in cases of domestic abuse ( House of Commons Justice Committee, 2016; Ministry of Justice 2016 ).
RESEARCH METHODS Go to section...This study used qualitative police data to conduct an exploratory investigation into the characteristics of domestic abuse cases with a restorative justice outcome.The police force we worked with is a small, high performing force, where all police officers had received training on restorative approaches and were actively encouraged by senior officers to use it in their day to day work.According to the training, to proceed with a restorative approach, the suspect would be approached first, and s/he would have to accept responsibility and volunteer to participate in the restorative approach.The victim would be approached second; the restorative approach would only proceed if the victim agreed to it.Risk assessments would be carried out on all parties.An audit of the training slides showed that Level 1 was not recommended for domestic abuse cases and Level 2 and 3 restorative conferences could be used in 'exceptional cases' for domestic abuse where approval was sought from a senior officer and it was done as part of a multi-agency, risk assessed, process.
Cases of domestic abuse (those flagged by the police as meeting the Government definition of domestic abuse, including intimate partner and familial) with a restorative approach listed as part of the disposal were identified from 1 April 2014 to 31 January 2015.This resulted in 62 cases, for which data was extracted from the crime recording system by reviewing the case management entry, victim management entry and domestic violence form for each case.It is possible that some cases were not correctly flagged as a) domestic abuse cases and/or b) as using a restorative justice outcome and therefore this number should be considered the minimum count.For each case, the following information was gathered: the relationship between the offender and victim, gender and age of offender and victim, crime type, disposal of crime, whether the offender was a previous domestic abuse reported offender, whether the victim feared the offender, if the victim was supported by the local domestic abuse agency, the 'level' and type of restorative approach used, and whether the restorative approach was authorised.In addition, an anonymised description of the case and the type of restorative approach used was collected from the crime recording system.We were reliant on the level of information given by officers in these descriptions, which was variable in nature and length.
Ethical approval was granted by The School of Applied Social Sciences research ethics commmittee at Durham University.Particular consideration was given to data access and storage.We did not request consent from the victim-survivors or perpetrators for their involvement in the research, and access to the data was brokered entirely through the police.Although this approach has some limitations, it is the standard way of conducting criminological research of this nature and is not out of step with other research investigating the policing of domestic abuse.Analysis consisted of descriptive statistics for the quantitative data.Thematic analysis of the qualitative data was undertaken so far as data volume/quality would allow.In choosing which cases to focus on in our results within our overall thematic analysis, we used a theoretical sampling approach which places greater emphasis on emerging and exploratory theory development than on representativeness.
We invited those that had arranged the Level 2 and 3 conferences (police officers and/or restorative hub workers) to participate in research interviews, however we had a low take up (only 3 of the 11 cases).This was due to the length of time that had lapsed between the incident and the interview request, and many officers declined on the basis they could not accurately recall the incident in question and others declined on the basis of time restraints.As this is an exploratory study, we include this data because it helps to illuminate some of our findings.Nonetheless, we emphasise that this data is not necessarily indicative of police experiences and views generally (for which further research would be needed).
An action research framework was taken throughout the research process.The research question itself -about the way restorative approaches were being used in cases of domestic abuse -had been identified in multi-agency domestic abuse meetings which included attendance by one of the research team and senior police officers.The data were collected from force computer systems by a researcher and police officer and inserted into an excel spreadsheet for analysis.The results were discussed with the police and followed by a large multi-agency conference for local stakeholders to discuss the results and how best to move forward in a safe yet progressive way.Even though the results, in our view, raised important concerns about how restorative approaches were being used in domestic abuse cases in this force, the police engaged in constructive discussions as to reform, without seeking to challenge or redact the results.In addition, the research led to important changes to the police protocols and practice around the use of restorative approaches in cases of domestic abuse.

RESULTS
Go to section...

Types of domestic abuse relationship
There was a fairly even divide between the intimate partner and familial cases, with partner relationships (current and ex-partners) accounting for 58% (n=36) of the cases and 42%, n=26 being familial relationships (family members aged 16 or over who are not partners or ex-partners).For both types of relationship, around two thirds of the offenders were male (69% for partner cases, 65% for familial cases).Similarly, most of the victims were female for both relationship types, although this was more pronounced for partner cases (72% for partner cases and 58% for familial cases).

Types of domestic abuse offences
Overall, restorative approaches were used for (ordered by most frequent) assault, criminal damage, theft and harassment domestic abuse offences.However, and as shown in Figure 1 below, for intimate partner violence cases, criminal damage offences were more common.Domestic abuse is characterised by repeated incidents over time: some of which may not appear particularly serious, harmful, or 'risky' when viewed individually or out of context ( Stark, 2007 ).For this reason, we looked at whether the offence where a restorative approach was used was a repeat offence.We had a lot of missing data for this variable and we therefore merged the 'no' with 'not known'.The results for this variable should therefore be viewed as minimum numbers.Overall, for familial and partner relationships combined, just over half (55%, n=34) of the offences involved people who were repeat offenders (the remainder being either logged as 'no' or 'not known').This was slightly higher -nearly two in three -for the intimate partner cases (64%, n=23) than for the familial cases (41%, n=11).Nearly half of all the cases (44%, n=27) involved a Level 1 restorative approach which consisted of either an apology, a financial agreement, or a combination of the two.These were used to respond to 11 assaults, 15 criminal damage offences and one theft.Level 1 restorative approaches were used in half of familial cases and 39 per cent of partner cases.We found that the vast majority of these disposals (all but three) did not have permission from a senior officer recorded on the incident log.This may be connected to the fact that force level policy and training did not recommend Level 1 approaches to be used in cases of domestic abuse (in line with national guidance).Four of the victims who participated in a Level 1 approach were recorded as being supported by a local domestic abuse agency (this should be considered a minimum count as some may not have been recorded).In a quarter of the cases, the restorative approach was accompanied by a formal criminal justice outcome (for example, a caution or a prosecution).
Level 1 'street-level' approaches therefore covered incidents not only of criminal damage and theft, but also assaults.It is difficult to characterise these as 'minor crimes and incidents' (as per the Level 1 definition), whether used in familial or partner abuse cases.The following section examines in more detail some examples of these incident types providing more context.

Financial agreements
The most frequent pattern found in the qualitative data relating to Level 1 approaches was one of financial agreements being made in a partner relationship following an act of criminal damage.This accounted for 11 of the 27 cases.The example below is one that we identified as being particularly problematic.
Example 1: 'Tom and Sophie', both aged in their early 30s.
After an argument with his partner Sophie, Tom (who was a previous known domestic abuse offender) smashed the glass panel in the front door leaving glass and blood on the floor.Tom was known to police for previous non-domestic assault as well as the previous domestic abuse.Sophie was pregnant and told the police that she feared Tom.Police arrested Tom for criminal damage.The case outcome was an Adult Simple Caution and adult restorative approach.No senior level authorisation for the restorative approach to take place was logged.Tom admitted guilt and agreed to pay Sophie for the damage.The restorative approach was recorded in the police officer's pocketbook.Tom was warned to stay away from Sophie's house.
This case clearly has a number of risk factors that suggest that a 'street-level' restorative approach may have not been appropriate.First, the offender is a known domestic abuse (and non-domestic abuse related violence) offender.Second, the victim is pregnant, and domestic abuse is associated with a range of negative maternal and fetal outcomes ( WHO, 2011 ).Third, the victim reported to the police that she feared the suspect.Fourth, the suspect is warned to stay away from the victim's address but no formal domestic violence protection order (DVPO) is issued.
Another case that involved partner domestic abuse and a financial agreement was one in which the male partner 'Matt' had stolen £150 from his female ex-partner 'Paula'.According to the police officer who attended, Paula's main concern was to have her money returned.This was one of the cases where we were able to interview the attending police officer, who explained his rationale for deciding that the restorative approach was appropriate in this case: 'Basically, I'm always led by what the victim wants.At the start when I joined the police it was always what the police wanted out of a job.What I've learnt is that people want different things.That what they want isn't necessarily the police outcome or the courts outcome -it's what they want.Obviously, she couldn't afford to lose that money, £150 I think it was -this way she wouldn't have to go to court and would get the money back, hopefully.'(Police Officer) The officer explained that he had not been able to trace Matt as he had left the area, so he had gone to Paula's house to update her and when he got there he found them back together and both at the property.The next excerpt demonstrates some awareness that the action taken did not exactly follow force policy or training: 'So, we did a face to face at the house, and we made an arrangement there and then.It was done at the house, but again it's difficult to get people at the police station, she didn't want to go to the police station, she just wanted to do it there.So, we do our best to follow the training, but we have to apply it in the real world.'Although feeling that both Paula and Matt were happy with the approach he had taken following this incident, the police officer said that he would be reluctant to use a financial restorative approach again because of the complications involved in ensuring compliance -'we're not there to collect a debt'.

Apologies
Of the level one responses, 15 included an apology.Most of these (12 of the 15) concerned assaults in familial relationships.Example 2, below, shows how an apology was used where there seemed to be a dual assault between siblings.This was the only case where there was an apology following a dual assault -the pattern was otherwise family members in one directional assaults.There were also a small number of cases (three) where apologies were used following partner assaults.One example is that of 'David and Anne':

Example 3: 'David (aged mid 40s) and Anne (aged mid 30s)'
David, who was a known previous domestic abuse offender, is said to lose his temper, grab his partner Anne at the front of her throat.He then pushes her away, causing pain and discomfort and reddening of skin.Anne then screams at David to leave the address and he does.David wanted to apologise to the victim so the officer spoke to Anne with a mental health worker present.David apologised to Anne, with the case outcome logged as a restorative approach disposal only.
Again, it is difficult to characterise either of these cases as 'minor crimes and incidents'.Both involved the male 'grabbing' the female by the throat, for instance, which could constitute Actual Bodily Harm (section 47 of the Offences against the Person Act).Further, research has shown non-lethal strangulation to be an important predictor of lethal violence ( Glass et al, 2008 ), leading to it becoming a standalone criminal offence in Queensland, Australia, in 2016 ( 55th Parliament Legal Affairs and Community Safety Committee, 2016 ).
Level 2/3 Restorative approaches (including restorative justice conferences) Level 2 and 3 approaches are those mostly associated with conferences or other planned, organised restorative approaches -the difference being whether the approach happens pre or post sentence -and we therefore discuss these together as the substantive outcomes are the same.However, it is important to note that because a Level 3 approach happens postsentence, it is possible that these are not always recorded on the police crime reporting system and the Level 3 approaches in particular should be considered an under-count.According to the data we had access to, Level 2 (no or pre-sentence) or 3 (post-sentence) restorative approach conference was used in 18 per cent of cases (n=11).These conferences were used to respond to seven assaults, three cases of criminal damage and one theft.Level 2/3 restorative approach conferences were used in 27 per cent of the familial cases in our sample and 11 per cent of partner cases.In three partner cases and three familial cases, the suspect was a previous domestic abuse offender.Eighty-two per cent of these cases resulted in a disposal of restorative approach only, 9 per cent also included an Adult Simple Caution and 9 per cent also included criminal charges.Overall, 64 per cent of the offenders who participated in conferences were female.This was lower for familial domestic abuse (57%) than for partner relationships, where three quarters (75%) of the offenders were female.Clearly, this does not match the pattern found in either police or non-police data where male offenders far outweigh female offenders.
The most frequent scenario for the conferences was an assault that took place within the family (accounting for 5 of the 11 cases).

Example 4: 'Mandy (aged early 50s) and Cassie (young person aged 16-21)'
Mandy used her clenched fist to punch her daughter Cassie's arms approximately ten times and then Mandy used her slipper to strike Cassie on her forearm several times causing no visible injury.The crime was recorded as assault without injury.The case outcome was restorative approach only, and the permission from a senior officer was recorded.The restorative approach included Mandy, Cassie, Social Services, and two Level 2 trained police officers.The conference resulted in a full apology and a package of support provided by Social Services.
In another case involving a mother and daughter, the police were called to an incident between mum 'Margaret' and adult daughter 'Laura'.We were able to interview the conference facilitator in this case.The facilitator explained the situation as 'a family dispute, not domestic violence per se…it wasn't the husband and wife that was the domestic violence'.Although not formally recorded as mutual abuse (Margaret, the mother was listed as the offender and Laura as the victim), it appeared that both had used a form of physical violence against each other: 'The daughter did apologise, she'd pushed her in the chest when she had tried to get in the house -then the mother pulled the daughter's hair, and the mother had then ended up being arrested.
Interviewer: Do you think it was clear in the conference who was the victim and who was the offender?'It was a bit of both really, with domestic ones there's always two sides to every story and its which one you believe and getting the acts that took place out really.'It is possible that it is cases such as these which are not 'domestic violence per se' in the words of the facilitator, and/or those that involve mutual abuse, that are being seen by police officers as particularly appropriate for restorative approaches.
The only Level 3 (post sentence) case in our sample was one that involved serious violence against a child and threat to kill the mum (the offender's ex-partner).The conference took place in prison where Ben was serving his sentence following conviction: Example 5: 'Ben (aged mid 30s), Toby (child aged 10-15) and Susan (age unknown)' Ben was already known to police as a previous domestic abuse offender when he punished his 11-year-old son Toby by hitting him 15 times to the right-hand side of his face and hitting him with a belt all over his body.These injuries caused severe bruising to Toby's face, arms, legs and body.Ben also threatened Toby's mother Susan with a knife (hence the domestic abuse flag).The crime was recorded as assault with injury and the disposal was a criminal charge and restorative approach in the form of a conference.The conference was recorded as being authorised by a senior officer, and was also approved by a social worker.During the conference Toby's mother Susan and Susan's brother brought Toby's questions to the conference for Ben to answer.Ben also went to prison.This example shows how restorative approaches can be used alongside criminal justice sanctions -in this case prison.During the multi-agency conference following this research, the police officers who participated in this conference reported that, in their view, this was a very positive example that showed how a conference can be used to give a voice to the victim, while hopefully reducing the likelihood of recidivism.It is a potentially useful example of where victims may not need to be present at the conference in order to ask questions and convey their views, in this case Susan and Toby's brother attended and raised Toby's concerns.Another issue raised by this case is that it appears that the threats to kill Susan were not fully acknowledged as the focus was solely on the physical violence towards the child, given the specifics of the charge and conviction.However, it appeared from the officer's account that Susan was very keen to participate in the conference.
In the final example in this section, we see another female offender, Cath, in a partner relationship.While there is not an allegation of mutual abuse around the time of this particular incident, the victim in this case (Billy) is a known domestic abuse offender who has himself participated as the offender in a restorative approach in the past but continues to use domestic abuse against Cath.In this incident, it is Cath who uses physical violence against Billy:

Example 6: 'Cath and Billy (both young people aged 16-21)'
Cath slapped her partner Billy on the back of his head five times.She then punched him.Billy was a known domestic abuse offender, who had participated in a restorative approach in a previous case.Cath and Billy agreed to a restorative conference, which was logged as authorised by a senior officer.The conference included an apology from Cath to Billy, and Cath was referred to mental health charity Mind.
It is clearly difficult to assess from police data only, the context of this violence and abuse, but it is nevertheless a troubling case.It is possible that this is a relationship where mutual violence and abuse operates.It is also possible, as will be expanded upon later in the discussion, that the police are using restorative approaches in such situations to avoid criminal prosecutions of women who are otherwise known victim-survivors.However, it is also possible that the restorative approach was requested by Billy as some form of 'pay back' for when he had to do the same and apologise to Cath.In other words, the restorative approach itself could have become part of any coercion and control operating within the relationship.

'Other' approaches
We categorised 19% (n=12) of the sample as 'other', because the description of the restorative approach used did not specify a level and did not clearly match a Level 1, 2 or 3 approach.These 'other' approaches were used to respond to seven assaults, three cases of criminal damage, one harassment, and one theft.In 8 per cent of familial cases and 28 per cent of partner cases, 'other' restorative approaches were used.In four partner cases and one familial case the suspect was a previous domestic abuse offender.Sixty-seven per cent of the offenders were male and the same percentage of victims were female.Sixty-seven per cent of cases ended in a restorative approach disposal only, 25 per cent included a caution in the disposal, and 8 per cent included a summonsed.
Half of the cases (6 out of 12) were of an assault within a partner relationship.These cases were troubling in different ways.Two that were particularly problematic are described below in Examples 7 and 8.

Example 7: 'Pete and Becky' (both aged in their mid 30s)
Pete punched his wife Becky in the back during an argument.Becky composed a letter to Pete that she wanted him to read and as a result Pete agreed to go to counselling.This approach was recorded as an authorised restorative approach by a senior officer.

Example 8: 'Jill (aged mid 20s) and Mark (aged mid 30s)'
Jill punched Mark in his shoulders and struck him to the left side of his face.Additional background information showed that Mark was due to attend the Crown Court for a serious assault he had perpetrated on Jill.Jill denied abusing Mark and they told the police they were living together as a couple again.Police officer explained that they would file the outcome as a restorative approach because they were living together again.
In both of these cases, the police used the 'language' of restorative justice to dispose of the case, but did not undertake any actions which would be understood as part of a restorative process.Further, the outcome of Example 7 arguably requires more of Becky than of Pete, as she had to write the letter and he only had to agree to attend counselling.There is also no actual restorative action in Example 8, with the fact they were back together as a couple being enough to deploy the language of restorative justice.More research is clearly needed to come to any firm conclusions, including interviews with police officers and with victim-survivors and offenders.Nonetheless, we suggest that in some of these cases, the police are attempting to circumnavigate the system to get what they believe to be the best outcome primarily for the victim-survivor (and also in some cases, secondarily, the offender).In Example 7 this was presumably because the victim wanted her partner to 'get help'; and in Example 8 this was because the offender was more often the victim (echoing some of the previous examples).In these cases, ticking the restorative approaches box opened up the opportunity to do something different outside of formal criminal justice procedures and without it being logged as 'no further action'.
DISCUSSION AND CONCLUSIONS Go to section...This study provides new evidence regarding the type of domestic abuse cases in respect of which the police are using a variety of restorative approaches.While it has already been established that the police are using such approaches in a widespread fashion ( Westmarland et al, 2018 ), this study delves deeper into the case reports from one police force to give more detailed background and information.As an exploratory study, the findings raise as many questions as answers in two particular research areas, namely policing domestic abuse and restorative justice studies.

Implications for policing domestic abuse
In terms of policing domestic abuse, the findings add weight to the argument that a new policy or legal definition is needed of domestic violence and abuse ( Kelly and Westmarland, 2014 ).Although coercive and controlling behaviour was added to the definition in 2013 and a new criminal offence was introduced in England and Wales in 2015, the assessment process for the cases in this research was not formally geared towards identifying or screening out those cases involving a system of the restriction of women's freedoms.Instead, there existed a great 'melting pot' of different forms of domestic abuse -some of which may have been appropriate for restorative approaches and some which were not.Having singular definitions and singular approaches, however, means that cases such as 'Mandy and Cassie' and 'Margaret and Laura' -both mother and daughter cases with co-accusations of physical violence and few obvious risk factors for homicide -are discussed in the same breath as expartners 'Tom and Sophie' who carry many risk markers since Sophie was pregnant and feared Tom who was a repeat offender.This is not as straightforward as simply separating out familial and relationship cases however, which the College of Policing (2015) guidance had already proposed.It is a more complex matter of identifying which cases include controlling or coercive behaviours and of preventing individual officers being the ones who differentiate between 'genuine domestics' and those that 'hit the definition' -as found in Myhill and Johnson's (2015) research.However, for so long as all cases are given the same name and flag of 'domestic abuse', the important differentiations are rendered largely invisible and this has important consequences for responses.
Hence, while we did find evidence of some cases which embody the concerns of critics who are apprehensive about any decreased safety of victim-survivors ( Coker, 2002;Acorn, 2004;Stubbs, 2002 ), this is not the only story, nor where the story ends.The interview with the police officer raised the proposition of doing what was right for the victim-survivor primarily, but also for the offender, if possible.It is possible that the cases that appear, at best, shoehorned into restorative approaches are not just examples of an 'acceptable' way of closing a case, but of cases where the police consider that no further action by the criminal justice system would be beneficial.Although more research is needed, it is possible that one example of this would be in dual reports where both parties are recorded as victim-offenders -cases of this nature account for around one in twenty recorded domestic abuse incidents and present a particular challenge to police responses ( Brooks and Kyle, 2015 ).Other examples are those cases where the female 'offender' was more often the victim-survivor in a pattern of violence, abuse, coercion and/or control against them.In other words, having a new 'box' to tick that is somewhere between a NFA (No Further Action) and positive arrest policy, opens up new opportunities for police discretion to be used -either positively or negatively.This could explain the apparent overuse of restorative approaches to respond to female offenders in this sample and in previous research ( Westmarland et al, 2018 ).The findings therefore support previous calls to consider gender and domestic abuse perpetration over time, not as discreet incidents ( Hester, 2009 ).

Implications for restorative justice researchers
When this broad definition of domestic abuse is added to such a stretched (and sometimes extremely marginal) nod to the spirit of restorative justice, the discussion becomes a complicated one to unpick.Just as the domestic abuse community must start to differentiate between different forms of domestic abuse, the restorative justice community must enter into greater dialogue with the police about the broad ways in which the police are using the term 'restorative justice' in practice.On a policy level, some have argued that the police Level 1 is not enough to constitute 'restorative justice' as we might commonly understand it.In these cases, restorative justice has arguably become watered down to effectively be a 'new NFA'.This echoes our previous research which found some out of court resolutions (including restorative approaches) may be being used to inappropriately 'clear up' some domestic abuse crimes, thereby improving crime detection performance rates ( Westmarland et al, 2018 ).Considering these are often happening alongside a lack of proper risk assessments and without formal authorisation, it is important that the restorative justice community takes a closer look using a critical lens at how the police are using restorative approaches -this should not solely be the concern of the domestic abuse community.Myhill and Johnson (2015) have shown how considerable discretion is linked with variable decision making around how the police record domestic related incidents and crimes -our research shows that this is also happening at the other end -at the incident closure stage under the name of 'restorative justice'.This is a clear problem for those seeking to improve the policing of domestic abuse, but it should also represent a serious concern to the restorative justice community.Just as police discretion was able to construct to some degree what domestic violence 'is' through their interpretation and recording techniques in Myhill and Johnson's research, we argue that they are similarly given considerable discretion to interpret at best, and alter and co-opt at worst, meanings of restorative justice.While further research is needed, the pessimistic interpretation of this is that such co-option is being used to cover-up poor policing practices.
The use of Level 1 approaches not only being used in domestic abuse cases but being the most commonly used is in direct contradiction to the force's own policy and training and raises serious questions about the ability of the police to monitor the introduction of new policy and practice.The examples described in this paper add further weight to the call to strengthen policy and reinforce guidance against the use of Level 1 restorative justice in domestic abuse cases that include coercive control ( Westmarland et al, 2018 ).Further, the ability of the police to monitor restorative solutions, such as financial agreements, raises particular problems for domestic abuse because of its repeated nature and in cases where there are shared family finances.Further consideration must be given to the appropriateness of different approaches for different offence types.

Feminist praxis and controversial topics
This research adds additional weight to the argument that the violence and abuse community must start to fully consider the opportunities and challenges that restorative approaches may bring.It has long been known that there are problems with traditional criminal justice responses to domestic abuse ( for example, Edwards, 1989, Hamner et al, 1989;Grace, 1995 ): this research demonstrates that there are also problems with the way the police are using restorative approaches.In most cases, the police approach stood distinct from the carefully planned, risk assessed, supported conferences that most restorative justice projects operate as standard ( Johnstone, 2013 ).
It is imperative that feminist scholars become part of the conversation in order to shape responses from a critical feminist perspective and do not simply stand outside of the process even when (or perhaps especially when) those processes are considered dangerous and risky to victim-survivors.By taking an action research approach and with full police cooperation, we were able to research more fully an issue we had identified as problematic, start a multiagency conversation on the subject based on actual cases, and ultimately see the police change their policies and approach.Police responses to domestic abuse are still a problematic area, as this research and other research and inspections continue to show.But many police are also keen to engage in research and reform to improve practice and enable better outcomes for victim-survivors.Feminist criminologists continue to have an important role to play in both documenting these problems and supporting change.

Figure 2 :
Figure 2:Click to view

Example 2 :
'Rachel (aged early 20s) and Martin (young person aged 16-21)' During a verbal argument, Rachel slapped her brother Martin in the face.Rachel then punched Martin and then Martin grabbed Rachel by the throat and punched her in the face.Both were arrested on suspicion of assault and both wrote each other letters of apology.No other criminal action was taken.