‘No black and white answer about how far we can go’: police decision making under the domestic violence disclosure scheme

ABSTRACT Domestic violence disclosure schemes are being adopted by police forces in countries around the world, yet they remain controversial and empirically under-researched. This paper presents findings from the largest study of police implementation of such a scheme to date, drawing on in-depth interviews and Freedom of Information data from 12 police forces in England and Wales. We reveal that victims of domestic abuse face a ‘postcode lottery’ of disclosures, with some receiving minimal or no information about the criminal histories of their partners, and others receiving lengthy and detailed descriptions. We identify and analyse two contrasting police approaches to disclosure: ‘risk-averse’ approaches, which are driven by efforts to avoid costly legal action by disgruntled offenders, and to minimise the resource implications of the scheme; and ‘permissive’ approaches, which are more explicitly victim-centred, reflect an increasingly prevalent ‘coercive control’ discourse, and are informed and guided by close collaboration with specialist partner agencies. The discussion sheds light on the shifting culture of domestic violence policing in the UK, yields immediate recommendations for the regulation and best practice of domestic violence disclosure schemes, and has methodological implications for efforts to assess their effectiveness.


Introduction
This paper examines how police in England and Wales are implementing the Domestic Violence Disclosure Scheme to safeguard victims of domestic abuse. The Domestic Violence Disclosure Schemealso known as 'Clare's Law', in memory of Clare Wood who was murdered by a partner with a known history of violence-was rolled out in England and Wales in 2014. It allows police to disclose normally confidential information about a person's criminal history to someone deemed at risk of future abuse, to help them make more informed choices about their safety. 1 Under the Domestic Violence Disclosure Scheme (hereafter, DVDS), police can provide victims with information proactively where they believe it will help safeguard them from future abuse, or in response to a request for disclosure by victims themselves. The DVDS has fast become established as a routine tool of domestic abuse safeguarding in England and Wales, with the number of disclosures made doubling from 3410 in the year ending March 2017 (Office of National Statistics 2017) to 6583 in the year ending March 2019 (Office of National Statistics 2019). It has also been adopted as a model by other jurisdictions, with equivalent domestic abuse disclosure schemes now operating in Scotland, Northern Ireland, New Zealand 2 and parts of Australia 3 and Canada. 4 The DVDS has been lauded as exemplary of a new, victim-centred, preventive approach to the policing of domestic abuse, which seeks to empower and protect victims and not merely to prosecute offenders (Home Office 2019). It has been embraced enthusiastically as a tool by domestic abuse practitioners within policing, as attested by the rise in its use. Groups working for and representing victims of domestic abuse have also welcomed government plans to further embed the DVDS in police safeguarding practice by including it in a much-anticipated Domestic Abuse Act (SafeLives 2019, Home Office 2019). 5 And, while evidence of its impact on victims is very limited indeed, tentative indications suggest that disclosures are valued by those who receive them (Home Office 2016, SafeLives 2019).
In contrast, legal and criminological examinations of the DVDS have been highly critical. From a legal perspective, some have questioned the doctrinal integrity of the disclosure procedure mandated by the Home Office and its compatibility with other important legal principles such as the right to privacy and the rehabilitation of offenders (Bessant 2015, Grace 2015, Duggan and Grace 2018. At the same time, criminologists have expressed scepticism about its safeguarding benefits, with some pointing out that merely telling people about the risks they face does not necessarily help them keep themselves safe (Fitz-Gibbon andWalklate 2017, Duggan 2018). Concerns have been expressed that patriarchal, victim-blaming police cultures will result in women being blamed or treated as lost causes if they do not meet police expectations to act on the information provided (Greene and O'Leary 2018, Duggan 2012, Koshan and Wiegers 2019. Some have argued that the DVDS risks diverting scarce resources away from more effective, tried-and-tested responses to domestic abuse, towards a scheme whose benefits are at best unproven (Wangmann 2016, Duggan 2018.
The aim of this paper is neither to take sides in this debate nor to resolve it. Rather, the aim is to better inform the discussion, by providing an empirically grounded account of how the DVDS is being implemented in practice. Both supporters and sceptics of the DVDS rest their claims on assumptions about how the scheme will in fact be implemented by practitioners on the ground, including assumptions about how police will use it. Yet there remains a paucity of evidence about how police actually interpret and operationalise the DVDS. 6 While national statistics show persistent and extreme divergences in the rate of disclosures by individual forces, they provide no insight into the reasons for these disparities. Nor do they help us understand what kind of information is being disclosed and what rationales and constraints shape police decisions about when and to whom to provide disclosures. At the same time, critical approaches draw heavily on submissions to public consultations, to which police are prohibited from submitting. As a result, the voices of DVDS practitioners are conspicuously absent from this debate. This study fills that gap, presenting findings from the largest in-depth qualitative study of police practitioners' implementation of the DVDS to date. In doing so, it also sheds new light on the culture of domestic abuse policing in the UK today.
Drawing on Freedom of Information data and a series of in-depth interviews with 29 police practitioners from 12 regional forces, the paper reveals that victims of domestic abuse face a postcode lottery of disclosures (as hypothesised by Grace (2018)), with some receiving no or minimal information about the criminal histories of their partners and others receiving lengthy and detailed descriptions, depending in which force area they reside. The findings shed light on the reasons for these discrepancies, showing a divide between two contrasting approaches to disclosure. The first, more 'permissive' approach tends to be more explicitly victim-centred, reflective of an increasingly prevalent 'coercive control' discourse, and informed and guided by close collaboration with specialist partner agencies. The second, 'risk-averse' approach to disclosure appears driven by the concerns of individual forces to avoid costly legal action and to minimise the resource implications of the DVDS. It is argued that the vagueness and indeterminacy of the official DVDS Guidance for police, published by the Home Office in 2016, has allowed forces to define their own disclosure thresholds and constraints, according to these divergent cultures.
These findings suggest clear recommendations for the development of best practice around the DVDS and other schemes like it, and for the planned statutory guidance that will accompany it into UK law. Specifically, they imply that a more permissive approach to disclosure, involving close consultation between police and partner agencies, should be encouraged, and that the rights to privacy of ex-victims, whose abuse might be inadvertently revealed by a disclosure, should be explicitly considered by police when drafting and redacting information about the past. They also suggest that any attempt to reach consistency between forces should not only involve redrafting of existing documentation, but must also contend with the challenges of effecting culture change in policing organisations. Finally, our findings have methodological implications for efforts to assess the effectiveness of 'the DVDS', because they suggest that there is more than one version of the scheme currently in operation.

Background
The introduction of the DVDS in England and Wales accompanies a recent drive to put 'the prevention of abuse and the protection of victims … at the heart' of the UK's approach to domestic abuse (Home Office 2019). This shift reflects a growing recognition that the existing police approach had prioritised the prosecution of offenders at a cost to victim safety, leading police to neglect cases in which the prospect of a prosecution was small, even when the risk of harm to the victim was high. The murder of Clare Wood was just one in a series of high-profile preventable domestic abuse homicides in connection with which police were criticised along these lines. In response, police forces in England and Wales have increased their funding allocations for domestic abuse 7 and introduced significant organisational changes, including the creation of specially trained domestic abuse safeguarding roles within dedicated, non-investigative policing units. Closer formal working relationships have also been established with partner agencies, including victims' advocates, mental health practitioners, probation and social care, to better coordinate risk assessments and safety plans for victims. The adoption of innovative preventive tools, such as the DVDS, which seek to empower and protect victims, are a core aspect of these reforms.
While a recent inspection found that the policing of domestic abuse had, 'improved markedly' and that '[v]ictims are now better supported and better protected' many concerns were also raised, notably about the 'large variations across forces … in the use of preventive measures' including the DVDS (HMICFRS 2019, p. 6). Calls for greater consistency in the police implementation of the DVDS are not new. On the contrary, they have been a feature of every single review and consultation relating to it since its introduction in 2014 (Home Office 2013b, 2016b. Efforts by the Home Office to further streamline and standardise police procedures and decision-making around the scheme, in part by issuing more prescriptive guidance to police about what kind of information could be disclosed and how disclosures should be worded, have not succeeded. Indeed, in 2019, just three years after the publication of the revised Home Office guidance, plans were announced to publish yet another set of guidance for police. Once again, the aim is to ensure that the DVDS is 'used and applied consistently across forces' (Home Office 2019, p. 46).
The current Home Office Guidance is indeterminate and vague, specifying only that disclosures should be reasonable, necessary and proportionate given the identification of a 'pressing need' for disclosure. This leaves a broad discretionary margin of interpretation for individual forces to fill. In what follows, we will argue that this lack of specificity and prescription is, along with the absence of instructive case-law, part of the explanation for why different police interpretations have emerged and indeed persisted. However, police culture also plays a key role in determining the way the Guidance is interpreted. Risk aversion in police culture has long been recognised as a persistent and often obstructive feature of police culture, whose 'advantages of weight and permanence' (Heaton 2011) make reform challenging (Flanagan 2008). Our findings suggest it is also a key determinant of some forces' reluctance to disclose detailed data under the DVDS, which is in tension with and challenged by the increasing adoption of victim-centred approaches to safeguarding and the expansion of 'coercive control' discourses in the field of domestic abuse policing (Monckton-Smith 2019, Myhill and Hohl 2019). This suggests that merely revising the Guidance will not be enough to ensure consistent and fair application of the scheme.

Methodology
The aim of this research was to investigate what kinds of criminal history data UK police forces are disclosing under the DVDS, and what reasons impinge on police decisions to include or exclude different kinds of data. To that end, our research employed a qualitative mixedmethods approach. First, in May and June of 2019 Freedom of Information requests were sent to all UK police forces 8 asking for copies of force-level policy or guidance specifying the kinds of criminal history information disclosed, under what circumstances, and to whom (e.g. current/ex-partners). A document analysis was undertaken by the second author, as a mapping exercise. This involved comparing the policy material received from, or referenced by, a force to the official policy position on the DVDS in terms of national guidance (Home Office 2016) at the time of the research. The exercise yielded a broad but superficial overview of the divergent approaches to disclosure and disclosure thresholds adopted by different forces. Nevertheless, it also revealed important differences in the operation of the Scheme, in particular that a small number of forces were relying on guidance that was out of step with national guidance updated in 2016.
Second, a series of in-depth interviews was undertaken with police officers and staff in domestic violence-related roles over a period of approximately 6 months in 2019. The aim of the interviews was to explore in depth how the DVDS is being implemented, the considerations and drivers influencing its implementation, and police views on its impact as a safeguarding tool. A total of twenty-nine interviews were carried out, across twelve forces in England and Wales (just over one quarter of forces). A first tranche of interviews was undertaken with fourteen officers in one force, all of whom had a domestic-abuse related role. Participants from this force were recruited through an agreement made between the lead researcher's university and the force to conduct a larger programme of qualitative research on police use of data (of which this study is just one part). 9 The remainder, which represented 1-2 interviewees from each of the other eleven forces, were recruited via invitations to participation posted on the members' forum of a domestic-abuse NGO, and direct emails from the lead researcher to regional MARACS, inviting participation. Ethical approval was obtained from the Research Ethics Committee at the University of Essex. The interviews, which lasted between 30 and 90 minutes, were carried out in private by the lead researcher, in most cases at the participants' place of work but sometimes on the telephone. All participants had a role in domestic violence policing, and all but two (who were involved intelligence and investigations) were domestic abuse safeguarding specialists with responsibility for either researching and/or drafting and delivering or authorising disclosures under the DVDS. Seven participants were police staff (civilians), of whom two had previous experience in domestic abuse casework or support. The remaining twenty-two were police officers. 10 Only nine of the twenty-nine participants were men, reflecting broader trends in the gender profile of safeguarding professionals. Two of the female participants disclosed to the interviewer that they themselves were survivors of domestic abuse.
This study was exploratory, as is appropriate to the research of an area about which little is known and little can be assumed. The interviews were carried out, coded (using Nvivo software) and analysed by the lead researcher, with whom ethical approval resided. The FOI responses were analysed by the second author. The interviews privileged the accounts of police practitioners as experts in their field, and sought their individual professional judgements and experiences alongside insights into organisational procedures. As such it was influenced in part by the narrative tradition in social research. Questions were minimal, open-ended and designed to elicit professional views about how the DVDS works and what their experiences of its impact and limitations are. These were followed up with more focused queries if responses had not already covered key aspects of the DVDS, such as constraints on disclosure. The analysis focused on the kind of information disclosed and the considerations informing its selection, which enabled the two overarching themes of risk averse and permissive approaches to be identified.

Limitations
One limitation of the FOI method was the long delays between application time and response: forces took between 2 weeks and 6 months to respond. This delay was extended further by the fact that 31 forces responded by directing the researchers to the public-facing HO Guidance from 2016. These forces were then prompted by the lead researcher to provide more granular, force-level documentation, but this was only provided in 2 cases. This points to a more serious limitation, namely that the FOI process only requires forces to share existing police documentation. In the case of the DVDS, it transpired that existing processes were not captured in guidance or documentation. Instead, the latter was nearly always vague, failing to outline force policies with respect to disclosure thresholds or the content of disclosures. Indeed, as will be discussed in detail in the findings section, forces that provided almost identical FOI responses were later revealed by the interview data to operate wildly different approaches to disclosure in practice. For these reasons, the FOI process could not by itself provide the kind of insight into practices that this study was seeking: for that, interviews were needed.
While it would have been desirable at least to extend the interviews to cover all UK forces, thus yielding a comprehensive view of the divergent approaches to disclosure, capacity constraints prevented this. Nevertheless, the sample size is large enough to provide a good sense of the range of approaches to disclosure currently adopted by police forces in the UK; in other words, it is unlikely that further interviews would have revealed alternative approaches. Further, while only one or more commonly two participants were interviewed for eleven of the twelve participating forces, there are in any case only a small number of individuals dealing with the DVDS in each force, and the approach to its implementation is a matter of the domestic abuse unit or department's policy rather than individual discretion. The interviews therefore represent a reliable account of force practice. However, a broader national picture would have strengthened the study significantly, by revealing which approach to disclosure is more common across England and Wales and therefore enabling the kind of reliable generalisations, which in turn could support more targeted recommendations for reform. It would also have revealed precisely how many and what proportion of forces are still relying on outdated national guidance. The research would also have benefited from a textual review of disclosures themselves. One participating force offered to provide researchers with a folder of redacted disclosures. However, they indicated that it would take two months to anonymise and prepare this. Requesting redacted disclosures from all forces would have increased the burdens of participation significantly, and since interviewees often described in detail the wording and register of disclosures already, sufficient insight could be gained.

Findings
Two main findings can be derived from responses to our FOI requests. The first is the disclosure thresholds and constraints operated by most forces remain de-facto, informal and uncodified. This is demonstrated by the fact that thirty-three of the thirty-nine responses received provided documentation that did not specify any disclosure thresholds or constraints in operation. Instead, responses consisted of repetitions of the 2016 Guidance, internal documents clearly derivative of that guidance, or generic and often public-facing information about the scheme. The second is that a significant number of forces continue to apply constraints on disclosure that are far more restrictive than the official guidance permits. Specifically, of the remaining 8 responses, three provided internal policies that cited out-of-date College of Policing guidance indicating that spent convictions cannot be disclosed; while one wrongly stipulated that ex-partners are ineligible for a disclosure; and one indicated that only victims assessed as high-risk would be given a disclosure. The 2016 Guidance from the Home Office allows for the disclosure of spent convictions in some circumstances (Home Office 2016, pp. 16-17); as well as for disclosures to previous partners (Home Office 2016, p. 24); and does not preclude the disclosure of relevant information to those classed at a medium or low risk (2016, p. 15).
Taken together, our FOI and our interview findings also show significant variance in the way police forces in England and Wales understand and implement the 2016 Guidance around the kind of criminal history information that can be disclosed, at what level of risk, and to whom. Forces diverge on whether they disclose spent convictions, convictions, charges, or other information indicating risk, such as reported crimes that resulted in 'no further action'. They also diverge on whether they make disclosures only to victims assessed as high-risk or also to those assessed as medium or standard risk. And they diverge on whether they make disclosures only to people who are in an intimate relationship or also to those who have ended a relationship but may still face some level of risk from their ex-partner. The level of detail included in a disclosure also varies greatly between forces, with some merely copying and pasting court records and others providing detailed descriptions of the ways in which perpetrators have harmed and attacked ex-partners.
This divergence has led to situation in which recipients of disclosures face a 'postcode lottery' in which 'some victims are told loads of stuff and some in other areas are being told not much at all' depending on the policy adopted by the force in whose are they live (P51 11 ). How and why have these discrepancies arisen, and what can they tell us about the different approaches taken to the assessment of 'pressing need' and the 'balancing' of the right to privacy with the need to safeguard the vulnerable? Our interview data provides further insight into the reasons for the discrepancies and the reasoning behind the divergent interpretations of the guidance.
Participants reported that the indeterminacy of the law and the of the HO guidance around the thresholds and constraints on disclosure under the DVDS had allowed forces to pursue their own independent approach to defining constraints around disclosure, resulting in what one participant described as 'very different disclosure thresholds' (P41). As one participant explained, 'the law isn't exact, it doesn't say what you can or can't disclose' (P47) and 'the guidance is just that-guidance. So its case law we're using and there's no black and white answer on how far we can go (P43). There is indeed scant case law around the disclosure of criminal history information for safeguarding purposes; and even the leading case on the issue does not offer prescriptive guidance, with the judge finding only that an undefined requirement of 'pressing need' must be met, and that 'each case must be judged on its own facts'. 12 Indeed, as that participant explained, these differences entail that two disclosures relating to the same perpetrator could be completely different depending on the location of the victim: there was one where a neighbouring force had a disclosure application for the same perpetrator and we compared and it was eye opening. What they disclosed was minimal (P41). More specifically, our findings reveal a clear divergence between what we term more risk-averse and more permissive approaches to disclosure. These are now examined in turn.

Risk averse approaches to disclosure
More risk-averse forces tend to disclose only to current partners and to provide minimal disclosures of unspent convictions, in the sparse, legalistic wording of a court record. Participants from four forces reported the erroneous view that this approach is mandated by the 2016 Guidance. While the original 2013 Guidance had prohibited both the disclosure of convictions deemed 'spent' by Rehabilitation of Offenders legislation, and disclosures to ex-partners, the 2016 Guidance lifted these constraints. Yet, as one participant reported: 'it says no spent convictions' (P51). This suggests that legal departments in those forces have not updated internal procedures.
The reasons for this failure to update internal procedures vary. In at least one participating force, our findings indicate that it is clearly the result of an oversight by the legal department, but in others, it appears to be a conscious decision by legal teams to interpret the HO Guidance in a way that reduces the risk of legal action against police. Concerns about the risk of legal action were mentioned repeatedly by participants in more risk-averse forces: I'm sure some [perpetrators] have sued the police but I've never heard it (P51) I think it should be used with cautionif we overuse stuff and it gets taken to court then we'll lose it completely (P47) Other responses suggested that reluctance to relax disclosure restrictions reflected more general risk-aversion around revealing criminal history information as a persistent feature of police culture: I think there's very much a culture in the police that we're very scared to disclose people's convictions because we were never previously allowed to do it. So it still takes some getting used to (P46) One participant suggested that limiting disclosures to unspent convictions was a way of reducing the resource implications of the DVDs: "A lot of applications get weeded out because they don't meet the criteria. I can't be in two places at once! I suspect that that criteria will remain quite tight because no one wants to create a role that needs to be resourced more." (P38) It is unclear from our findings how significant a factor resource implications is in influencing force policies around disclosure under the DVDS. Reported resource implications of the DVDS varied wildly, depending on how enthusiastically the DVDS had been promoted to the public in the force area in question, the availability of an online system for Right to Ask applications (applications for disclosures by members of the public), and how proactively the force used their power to issue Right to Know disclosures (disclosures issued proactively by police). 13 However, what these findings do show is that at least some forces are not availing themselves of their full powers to safeguard individuals. Instead, they are instituting default or blanket constraints on disclosure, irrespective of the particular circumstances of the case at hand. Yet these constraints do not appear to be determined by reference to the considerations in the 2016 Home Office guidance, namely a judgement of the proportionality, necessity, and 'pressing need' to disclose.
Default constraints on disclosure of spent conviction or non-conviction data were viewed by participants who had to operate within them as problematic and as inhibiting the safeguarding potential of the DVDS. Participants consistently expressed frustration with the limits placed on their ability to communicate risk to victims and thereby to help safeguard them. For example, when asked whether it happens that they would like to disclose incidents that fall below the threshold one participant, who was prohibited from disclosing any data apart from unspent convictions responded: Yes, loads of things, all the time, yes. … there are things that you'd like to disclose because they've been quite violent to a partner, but we can't … we have to work within the regulations and the law and we can't breach that. (P09) This echoes findings from a study with one UK force not represented in our research, in which officers in what appeared to be a risk-averse force expressed wanting to make more expansive and detailed-and therefore more effective-disclosures (Duggan 2018, p. 210).
While spent convictions and non-convictions can be disclosed where the relevant standards are met, confusion and/or excessive risk-aversion around this issue amongst some forces appears to be causing a risk to women's safety. For example, a small number of participants reported that restrictions on their disclosure of spent convictions and of non-convictions compelled them to tell applicants there is nothing to disclose about a perpetrator, when in fact there was a long history of repeated, corroborated incidents of violence indicating serious risk. In one case, a participant reported actively discouraging social care from alerting a victim to the DVDS, for fear that an application would yield 'nothing to disclose' and thus provide false reassurance and increased vulnerability: A perp in my area had several victims and a type of victim, so we knew who was vulnerable, and he was entering into a new relationship with someone who adult social care were desperate for us to make a disclosure to, and we were saying "we cannot disclose anything and it will just empower him as he'll say 'there's nothing there'". So we had to say "don't tell her about this scheme" [because] if we'd received the application we would've had to process it. That was hairy and difficult and partners were saying, "you've told us about this and promoted it and now we want to tell someone about it and you say not to!" If we'd said "we've got nothing to tell you" even though we knew he had a glittering history of high risk behaviour, that would have put her at risk. (P47) Other participants reported devising workarounds to enable them to communicate risk to victims without violating the constraints imposed by their force. For example, in one force the prohibition on disclosing convictions meant reports of crimes were disclosed but convictions for those same crimes were withheld: The legal department says you're allowed to tell them [about the violence/abuse] but not allowed to tell them it was a conviction. Legal advice is that information like "in 2007 his ex-partner made a report that he punched her in the face and gave her a broken nose" can be disclosed. Normally they [victims] don't ask if the conviction happened. … … there's also cryptic conversations, where you say [to victims] "I'm not going to make a disclosure, there are certain thresholds that have to be met". (P51) And in another, participants reported issuing disclosures that avoided mentioning any incidents but instead consisted of vague warnings of 'risk': However, as the above quote illustrates, workarounds were generally seen as a poor substitute for the kind of concrete information about specific risks whose disclosure the DVDS was established to enable. More generally, as our examination of more permissive disclosure practices will now show, applications of the DVDS that involve blanket constraints on the disclosure of spent and non-conviction data are likely to result in missed opportunities to safeguard vulnerable victims.

Permissive approaches to disclosure
In stark contrast to these restrictive or risk-averse approaches, a significant sub-group of participating forces operate with far more expansive and permissive rules around disclosure. Participants from these forces reported routinely disclosing spent convictions and reported crimes, but also incidents that resulted in no police further action, and indeed detailed descriptions of the modus operandi used by perpetrators. Many of these participants criticised strongly the more restrictive approach employed by risk-averse forces, arguing that the safeguarding potential of the DVDS would be undermined by failures to disclose expansively.
One key reason why the ability to disclose spent conviction data was seen as important was the relative speed with which many domestic violence and abuse convictions become spent. As one of our participants reported, 95% of offences in domestic violence are spent because they're low level (P51). But even more concerning for many participants was the low charge and conviction rate in domestic abuse cases, which means that reliable information indicating that a person poses a high risk of harm to another would not be disclosed if disclosure was limited to convicted crimes: As we all know, a lot might be reported but not prosecuted … if you just went on convictions then we wouldn't do some [disclosures] when we probably should. We're very clear in the disclosure that often DA is not prosecuted but there are reports. We try to be fair to both parties but while getting across the point that there is information that is relevant for them to keep themselves safe. (P54) Along similar lines, many participants pointed out that restricting disclosures to convictions or charges would leave out information about other superficially minor kinds of abuse, which could nevertheless be a precursor to violence and therefore relevant to victims' appreciation of the risk faced: … other forces they just say 'In 1993 Joe Blogs was convicted and found guilty of GBH. They don't expand on what he did to the person. But the thing is it's not only about physical abuse it's also psychological, it's financial, coercive controlling behaviour, and those things are missed where forces don't disclose-where there have been 40 DA-related call outs in a year but he was never arrested. It misses out on that pattern. Because it's still abuse. It's about giving an insight isn't it-because you can be in a relationship with somebody and have just verbal arguments that are then reported and graded as standard risk. And then one day you have a row and he flips and hurts you or kills you and no one is told about the little things that have led to the big explosion. (P42) Participants in more permissive forces were also often passionately committed to providing victims with access to information about the modus operandi and circumstances of the crime, which they saw as vital to the safeguarding value of the disclosure. Without this kind of detail, they said, the disclosure would fail to alert victims to the fact their own abuse fitted a pattern of behaviour by the perpetrator, and therefore that 'it wasn't something they did but it was inherent in the perpetrator' (P41). The potential of disclosures to expose the flaws in the narrative of the offender and to trigger this realisation was seen as vital to the potential of the DVDS to transfer some of the power back to victims: It's trying to illustrate to them, this is about the subject and his or her behaviour rather than about you, trying to shift that blame … it can make the difference of them staying in the criminal justice system or disengaging. When we were looking at our processes our management were saying "you need to cut some of this out, just look at PNC [police national computer] convictions, end of". But I fought back hard with that. From [my experience] seeing it, when you tell them about the conviction, when they can see a long list filling the page and you can add context, you add detail, they can see and relate to it, it resonates more and they start to believe "I'm not the only one". They can see -people do minimise-the minute they see it has happened to numerous other people, it isn't about them, it's not something they did wrong that made them deserve that kind of treatment. … You have one chance of delivering it and if it's not done in the right way it might be lost.
For me, [it's important to disclose] if there's biting or strangling or sexual stuff that the victim can identify with … It's important to say if there's been more than one victim because we know that when we meet up they'll justify that one thing-"she lied about him, she's mad", but then you can say "there's three others". It's much more impactive. (P53) Indeed, one participant spoke for many when they depicted it as against the spirit of the DVDS to restrict disclosures to a mere report of conviction data: … without the level of detail and information you're not doing the Clare's Law as much as you can. You need to take the opportunity while you've got it, to say "they are a risk and this is why they are a risk". The MOs [modus operandi of the perpetrator] are crucial in that, when you see "he jumped on a stomach when someone was 8 months pregnant" that is hard-hitting and the detail will make the difference rather than "an assault". And [we also disclose] the reason they weren't convicted. (P50) In contrast to participants from risk-averse forces, participants from permissive forces were confident of their rights and discretion to disclose any kind of criminal history information if necessary to protect someone from crime. For example, one participant with 18 years of experience in policing domestic abuse described their approach to disclosing detail and modus operandi under the DVDS as merely continuous with disclosures made prior to its implementation, under common law: There doesn't seem to be a set criteria [for what can be disclosed under the DVDS] so I've gone back to doing what I would've done before Clare's Law under common law … for example I had a girl 12 yrs ago and she was pregnant and she had a saucepan mark on her stomach where he had burnt her which had happened "by accident". And she was saying "this is a freak accident, I made him do it". But he had burnt the stomach of a previous partner who was pregnant, with an iron. It demonstrates a propensity to a certain crime, it was an extreme MO [modus operandi]. So I could say "this is a dangerous man, this wasn't an accident, this is what he does". So I'll take my chance with the court and the girl with the saucepan, I'll justify telling that girl all day long, it's common law and we have a right to disclose it. (P57) Alongside the value of detail and modus operandi, many participants emphasised the importance of wording disclosures in simple conversational English. Indeed, they derided disclosures that merely replicate court records or police data, or which use 'police language' as meaningless, incomprehensible, and potentially alienating for victims: If they just look at police printouts it won't work. It has to be translated otherwise it just creates a barrier. You go in plainclothes otherwise all they see is police, they don't see someone sitting down and relating in a human way. A part of them will think "all you want is me to help you to get him nicked". It's got to be about them and their family, giving them information. Removing barriers is important. If you use [words like] "police, charges, assault and battery" it means nothing to them. If they saw "he took the baby away and wouldn't let it feed" [then they'd think] "oh my god he did that to me". It gives them a link and makes them say "they're talking in my language and, well he sort of did this to me". (P44) However, it was also widely acknowledged, by participants from both risk-averse and permissive forces, that there are costs to delivering a 'powerful' and 'hard-hitting' (P50) disclosure. In particular, such disclosures can, for some victims, 14 increase the fear and terror experienced, especially if recipients do not feel able to take steps independently to safeguard themselves: We need to think about the bombshells we drop-one of my colleagues delivered [a disclosure] and the lady was physically sick. … There can be quite a lot of fallout from what they hear (P41) … often there's children and we're saying "5 years ago he strangled an ex-partner", and what can they do? It's life changing news so you've got to be careful. (P54) It's like being given a ticking timebomb, you need to make informed decisions but you may not have the mental equipment to deal with it. (P47) Concerns about the potentially negative impact of shocking or terrifying revelations on victims prompted many practitioners to consult closely with domestic abuse caseworkers (who serve as a victim's first point of contact and safeguarding advisor and are also known in some forces as Independent Domestic Violence Advisors) and mental health practitioners before making a disclosure. The role of these partner agencies in providing insight into the circumstances of the individual victim was seen by many as vital to ensuring disclosures did not unintentionally undermine the safety of victims. For example, some participants reported withholding or delaying disclosures on the advice of professionals, or indeed allowing a trusted party to deliver the disclosure: Is it the right time to know … [Sometimes] the outreach worker will say "Thank you but can we put it on hold? She's not ready." They have a bigger role in this, they can say "No" or "Yes, but let us do it" (P44) A few participants also underlined the importance of the advice of mental health practitioners in ensuring that disclosures that were anticipated to be frightening or disturbing were not made in circumstances in which they might undermine rather than promote safety: What we don't want to do is open them up to traumatic information and send them out on their merry way without some plan in place. I've had one where we've had a victim who is highly suicidal so in that case telling them information that could push them over the edge is not something I want to do. We think: Is this the right time or do we hold it until they're more stable? Or just withdraw it completely? It may be that we look at what support comes with them, a particular practitioner they know, someone specifically there to look at their mental health and factor their mental health into safety planning. (P50) As these responses illustrate, practitioners operating within permissive disclosure regimes rely less on predetermined thresholds and constraints to determine the content and wording of a disclosure than those operating within risk-averse regimes. Instead, their judgments about what to disclose, how, and when were more discretionary, tailored more closely to the particular circumstances of individual victims and informed by closer consultation with other specialist safeguarding practitioners. Unlike more risk-averse forces, the disclosures were carefully formulated with a view to achieving specific safeguarding objectives, such as correcting a false narrative peddled by the perpetrator, or exposing a modus operandi. In addition, their decisionmaking and procedures prioritised the safeguarding of the victim over the rights of perpetrators or any perceived legal risk to policing more strongly than risk-averse forces. This may in part be explained by the fact that forces with more permissive approaches were also more likely both to separate safeguarding roles from investigative ones, and to appoint domestic violence safeguarding specialists to DVDS-implementation roles, (including in two forces civilian practitioners with a background in victim-facing casework). The discourse employed by these specialists indicates that they are better-versed in the theories of coercive control, which has emerged as a key framework for understanding domestic abuse in recent years (Monckton-Smith 2019, p. 9, Myhill and Hohl 2019).

Balancing safeguarding and rights to privacy
Our findings also reveal that, across all participating forces, the content of disclosures is determined by assessments of the potential risks of disclosure not only to the right to privacy of perpetrators, but also previous victims. While the Home Office guidance states that 'sharing of personal information about a potential perpetrator may be an interference with a person's right to a private and family life' (Home Office 2016, p. 45), this overlooks the fact that information about 'a person's known history of violence and abuse' is also information about the victim of that abuse, and potentially also about any children in their care.
As well as balancing the need to safeguard victims against the rights to privacy of perpetrators, participants reported redacting disclosures in order to protect the privacy of the perpetrator's previous victims. Many participants reported concerns about the fact that disclosures often include dates of convictions or reported crimes, which a recipient can combine with their existing knowledge of their partner's personal history to deduce the identity of the survivor. A range of measures, such as limiting details of the offense or the date, taken by participants across the board, to protect the privacy of survivors: You don't want to disclose stuff that makes it obvious who the victim is … We'd say 'since 2009' but no more dates. If there was a 3 year-old child present at a sexual assault they might be able to work out who the victim is and that's not fair. And if it's a sexual assault where the victim has reported but refused to make a formal complaint and doesn't want it recorded then we won't disclose if there's a risk of them being identified. (P53) The protection of survivor identities was also seen as important from a safeguarding perspective. Concerns were expressed that perpetrators might come to know the content of disclosures and use this as a reason to retaliate against ex-partners. One participant reported struggling to balance their duty to disclose with the need to protect the rights and safety of previous victims: We have to be careful because we can reveal the identity of the victim so we try to be as sensitive as possible, especially if it's an NFA [no further action taken by police] not because they've been undetected, but purely because the victim has withdrawn consent. Generally [recipients of a disclosure] will be aware of other victims-"I know he has issues with his ex" -but they've heard it from the perpetrator's perspective. So it's very difficult not to breach the confidentiality, but … even if it's NFA we would disclose an MO but cleanse it so that it shows not too much about who the victim is. (P50) Some participants reported having adopted 'rules of thumb' to help them navigate this tension. For example, one reported deciding to 'only disclose stuff the suspect knows he has been accused of' (P51). For one participant, concerns about the privacy of survivors weighed as heavily on their decisions about what to disclose as those about the rights of perpetrators: Our disclosures are much smaller than other forces. We don't want an applicant to look at a job and say "oh yeah that would've been Tracy or Sharon"-we want to protect victims. Are we being too cautious or could we disclose more? That's a discussion we'd like to have. (P56) This last comment is also representative of the strong appetite amongst most participants for discussion and debate with other practitioners about which approaches to disclosure were getting the various aspects of this judgement 'right'. In general, participants expressed keen awareness of the different approaches to disclosure and a desire for more consensus and indeed more prescriptive guidance from the government to support this.

Discussion
As our findings indicate, the vast majority of our participants advocate a permissive approach to disclosure under the DVDS. Therefore, though we are reluctant to promote a specific approach to disclosures under the DVDS in the absence of further research, it seems probable that the results of any future study would support a permissive approach. Any shift in police practice to embrace such an approach would inevitably imply greater resourcing for the DVDS, as disclosures would be made more often, in more detail, and supported by more extensive criminal history checks. This raises the question, already posed by others (Fitz-Gibbon andWalklate 2017, p. 18, Duggan 2018, p. 203) of whether such investment would represent a sound use of scarce resources, in a field of policing in which there is undoubtedly still much room for improvement. We have not been able to address that question here.
Neither have we been able to shed light on how other agencies, apart from the police, engage with the DVDS. Concerns are often raised about the potential for victims to suffer adverse consequences if they choose not to end a relationship after receiving a disclosure, as a result of victimblaming cultures in children's social care and the criminal justice system (Refuge 2012, pp. 2, 6, Duggan 2012, Grace 2015, Koshan and Wiegers 2019. Notwithstanding the closer collaboration between policing and these agencies, the diversity in professional remits, constraints, and cultures means that little can be inferred from our findings about these concerns. Nevertheless, our findings yield clear recommendations for the development of best practice around the DVDS and other schemes like it, and for any future statutory guidance that accompanies it into UK law. Specifically, they suggest that the provision by police of detailed and comprehensive disclosures of both conviction data and reports of crime, supported by close engagement with specialist partner agencies, is most likely to achieve the kind of safeguarding benefits the DVDS was set up to deliver. This should be explicitly recognised in any future training, best practice, or statutory guidance to reduce both inconsistency and the risk that victims will be given false reassurance through excessively cautious and uninformative disclosures. They also suggest that future statutory guidance for police should emphasise the need to consider, as many forces already do in practice, the rights to privacy of survivors of domestic violence, alongside those of perpetrators. And they provide ground on which to challenge the roll-out of schemes which, like that adopted in New South Wales, impose strict limits on the disclosure of spent convictions or crimes committed in other jurisdictions (Wangmann 2016).
Our findings also shed doubt on the claim, made repeatedly by the Home Office, HMICFRS, and others, that increasing consistency in application of the DVDS is simply a question of making the numbers align upwards, by encouraging those forces with comparatively lower rates of disclosure to increase their use of the scheme (HMICFRS 2019, p. 64, Graham 2018. Not only does this position conflate more frequent with better use of the scheme, but it belies an assumption that all disclosures are equally effective and fair and that therefore forces committed to victim protection should simply seek to make as many as possible. Yet our findings have shown that there is significant divergence both in disclosures themselves, and in practitioner views about what constitutes a fair and effective disclosure. This suggests that we need first to investigate (and indeed conceptualise) in more depth what effectiveness in a DVDS would look like, before we try to achieve it through the publication of legally binding guidance. Such an investigation should take the form of a nationwide, systematic evaluation of the DVDS of the kind that would combine the police perspective with specialist caseworker insights and, most importantly, feedback from survivors. The results would be formative to both practice and regulation in the UK and beyond.

Conclusion
In addition to the more practical and methodological implications just considered, our findings should be a source of both concern and reassurance to critical scholars of the DVDS. For they suggest not only that there are likely to be cases in which both disclosures and failures to disclose increase the vulnerability of victims to further abuse, but also that many police practitioners are keenly aware of these risks and seeking actively to minimise them in their own disclosure practices. More generally, they reveal an apparently growing alignment between both the professional culture and identity of domestic abuse policing in England and Wales, and domestic abuse casework; a development that can only be positive for victims. Only two qualitative studies of the DVDS have been undertaken in the six years since its roll-out. The first was by the Home Office (2016b), and the second was by Duggan (2018). 7. Police spending on 'public protection', within which domestic abuse safeguarding sits, rose by 22 percent between 2017 and 2019 (HMICFRS 2019, p. 9), a figure that is notable in light of the fact that policing has undergone real-term budget cuts of 19% in the period 2010-2018 (National Audit Office 2018, p. 7) 8. A total of 43 regional forces. This included England and Wales well as Police Scotland and the Police Service of Northern Ireland, though in both these latter jurisdictions the legal basis of the local version of the Scheme is different. 9. These were part of a larger, ongoing qualitative study into the ethical and human rights implications of police use of data and technologies 10. Ranks ranged from PC to Detective Inspector, though the majority were Sergeants. 11. Participant numbers were allocated as part of the larger study, which is why they do not correspond to the number of people involved in this study, e.g. 1-29. 12. Munby LJ in H and L v A City Council [2011] EWCA Civ 403 at para 37 found there was an inherent link between a 'pressing need' test for disclosure and the 'proportionality' requirements of Article 8 ECHR for disclosures, noting that: As the authorities show, each case must be judged on its own facts. The issue is essentially one of proportionality. Information such as that with which we are here concerned is to be disclosed only if there is a 'pressing need' for that disclosure. There is no difference in this context between the common law test and the approach mandated by Article 8. The outcome is the same under both.
13. For example, one (risk-averse) force employs two police officers as DVDS disclosure specialists on a full-time basis (P02) with a more senior officer devoting time to authorising the disclosures. Another force employs a team of 5 police officers, who devote about 60% of their full working time to implementation of the DVDS, plus authorisation from senior officers and a discussion at a MARAC (P53, P54). 14. It is important to note that nearly all participants emphasised that disclosures are met with a very wide range of reactions, from utter indifference to shock and devastation, and that generalisations about the impact on victims are therefore very difficult to make.

Disclosure statement
No potential conflict of interest was reported by the author(s).

Funding
This work was supported by the Economic and Social Research Council [grant number ES/M010236/1].