Seeing What is ‘Invisible in Plain Sight’: Policing Coercive Control

Coercive control has emerged as a key focus for researchers and activists working in the field of intimate partner abuse. In England and Wales, the issue has taken on a new urgency. On 29 December 2015, s. 76, Serious Crime Act made &#8216;coercive or controlling behaviour&#8217; a criminal offence. Implementation of the new offence has been slow. The analysis of data generated by empirical work with police and survivors suggests that police need to understand a working model of coercive control in order to adopt what could be a transformative approach to policing intimate partner abuse.

None of this will happen unless the police are empowered to be proactive about enforcing the new offence. The rest of this article reviews how this might be done. It is organised into three parts. Part One uses data from interviews and a focus group with senior police officers to explore why the police might be struggling to capitalise on the new opportunities presented by s. 76. Part Two draws on the academic literature (and in particular the work of Mary Ann Dutton, Lisa Goodman, and Stark) and interviews and focus groups with survivors, their closest advisors, and police, to develop a working model of coercive control for use by the police. Part Three assesses the extent to which the model developed in Part Two might address the issues raised by the research described in Part One.

Methodology
The article draws on an empirical study that consisted of interviews and focus groups with survivors, police and criminal justice independent domestic violence advisors (IDVAs). 3 IDVAs were selected as a category of respondent in addition to survivors and police because of their distinct perspective on the communication channels between survivors and police. Although the study sample was small (30 respondents), the collective experience of domestic abuse was extensive. The majority of the IDVAs were highly experienced, having worked intimately with a high volume of survivors over a period of several years. The survivors, too, were unfortunately similarly experienced: many had experienced abuse from more than one partner and over a period of between ten and 20 years. The majority of the police officers were senior officers in charge of specialist domestic abuse units. The small scale of the study and its geographical specificity (the interviews and focus groups all took place in London and the south-east) prevent claims of generalisation, but the combined extensive experience of domestic abuse allowed for detailed feedback on the issues of importance to this project.
One further point needs explanation here: it is recognised that this project focuses on female (rather than male) survivors. The focus is intentional, and is justified by reference to research referred to in the introduction that shows that coercive control is perpetrated almost exclusively by men (Dobash and Dobash 2004;Johnson 2006;Stark 2007). It is accepted that more research is needed into the use of control within same sex relationships, but this research is outside the scope of this project.

Part One: Coercive Control and the Police
A low prosecution count in the first year of operation of a complicated new offence is, in some ways, to be expected. Indeed, the senior domestic abuse specialists within the police that were a part of this project were, for the most part, positive and enthusiastic about s. 76. P01, for example, commented that: 'this should be a fantastic new story. And it is. We have got a piece of law that now lets us deal with some of the most nasty people that you could come across'. But together with that enthusiasm came a recognition of challenges presented by what officers saw as an entirely new approach.
To understand these challenges, it is important to understand that police historically have responded to intimate partner abuse within a framework that Stark calls the 'violent incident model'. This model targets discrete assaults, and rests on an assumption that 'the severity of abuse can be gauged by applying a calculus of physical harms to these incidents' (Stark 2012, p.200). P01 agreed that this is the case: what you used to see before coercive control came in, just giving you a sense of what the culture change that is needed, is a lot of the time you would get officers called to a domestic related deployment. They would get there, they would speak to both parties, they would update on the response log: 'verbal only domestic no offences disclosed'. That was your standard response to something where there wasn't violence. Because it was -we turn up, we deal with violence, and we move on.
In many ways the 'we turn up, we deal with violence, and we move on' approach works for the police because it 'fits' with the way in which police like to work. 'Typical, normal policing' (P01) complements what P01 described as 'black and white thinking'.
Thus the 'normal' domestic abuse incident (involving violence) crosses clear 'black and white' normative and descriptive boundaries. P01 went on to explain: when it comes to violence -for me -there is an obvious line. Violence is never acceptable in a relationship. Actually, we need to be taking formal action every time we become aware of it -and it is a fairly obvious line. And that kind of works neatly. And for more serious violence -you've got injuries, you've got bruises, you've got broken bones and things. Because officers like things black and white. When you look at -assaults for example: you've got a really clear barrier between GBH, ABH, common assault. So even just at the level of injuries it's black and white, or black and blue. It's really clear: this is what it is like, this is what we are dealing with. P01 found it easy to place physical violence between intimate partners on the 'wrong' side of a normative boundary: 'violence is never acceptable in a relationship'. He also found that violence is straightforward from a descriptive perspective: 'you've got injuries, you've got bruises, you've got broken bones and things'. P01 points out the clarity of the hierarchy -he suggests that a discrete assault, for example, rests at a point along a spectrum of severity of physical injury and has an associated incident specific legislative and common law framework ('GBH, ABH, common assault'). This clarity 'fits' with the police mindset of 'we see a problem, we fix it' (P01). As P01 explained, above all, 'police officers like to get stuff done'.
That the violent incident model has much to recommend it from a policing perspective is clear. Also clear, however, is that it 'bears little resemblance to the forms of oppression that drive most abused women to require outside assistance' (Stark 2012, p.201). This mismatch, between the violent incident model and the empirical reality it confronts, creates longterm problems for the intersection between criminal justice and intimate partner abuse. Stark points out that the violent incidence model trivialises abuse because the kinds of ongoing low-level violence experienced by a high proportion of abused women are dismissed as 'minor' when viewed through an incident specific lens (Stark 2012).
Perhaps the most significant negative consequence of the trivialisation of abuse in this way is the impact on the risk assessment process that police utilise to try to keep the most vulnerable victims safe. In the focus group that was run with senior police officers, FGP02 explained that: 'it always surprises first responders and front line officers, when you have a "low risk" they are likely to be the murders'. Two other senior officers in the room (FGP01 and FGP02) agreed that domestic homicides are almost always initially (incorrectly) assessed as 'low risk' cases. When pressed on this point FGP03 went on to give as an example a recent domestic homicide in a neighbouring town. He said: 'it was a female who reported on six or seven occasions low risk harassments . . . when I say "low risk" it sounds so wrong . . . every job, every contact that she had with XXXX Police was "low risk"'. When asked: 'what made them low risk?', FGP03 responded: 'because they were looked at as individual isolated incidents . . . but, she's now dead'. He concluded: 'so, now it's easy to say, well, six different incidents . . . when officers are presented with just that "low" thing -it's "low risk"'. Risk assessments are a necessarily evil -a way to ration finite police resources. But as Stark (2012) points out: 'the level of control an offender is exercising is a far better way to ration scarce police resources than the level of violence' (p.202).
Another long-term difficulty exacerbated by the violent incident model is that it leads to 'the assumption that victims . . . exercise decisional autonomy "between" episodes' (Stark 2012, p.200). This means that victims who fail to capitalise on that (assumed) autonomy are perceived as responsible, at least in part, for the ongoing abuse that they experience. However, as Dutton (1992) explains: 'abusive behaviour does not occur as a series of discrete events' (p.1208). Instead, the wrong of abuse is enacted on the terrain of the relationship and it is the abuser's strategic intent that provides the thread that connects and organises the acts.
Each violent incident is, to many women, relatively unimportant in the context of the 'state of siege' (Dutton 1992(Dutton , p.1208 imposed by their abuser. There is no 'between' episodes: while the violence might be sporadic, the fear it engenders is not. Stark describes the effect of this kind of mismatch between police assumptions and empirical reality as follows: 'some police officers attribute the woman's apparent inability to "leave" to a deficit in her character and consider her expressions of fear exaggerated, fabricated, or as the byproduct of mental illness, particularly in contrast to the relatively minor nature of the incident to which they are responding' (Stark 2012, p.205). P01 commented simply that he understood that 'by and large if you've got physical injuries with no psychological effect it's easier to leave'. Getting his officers to understand was more of a challenge: 'how do we get them to know that?' (P01).
Frustration with victims who are 'un-cooperative', difficulties with risk assessment and mistakes with safeguarding all have an effect on the ability of police to prosecute offenders and keep victims safe. For these reasons Stark (2012) concluded, before the introduction of s. 76, that 'even the most rigorous enforcement of current domestic violence law is largely ineffective against coercive control, the most prevalent and devastating form of partner abuse' (p.213). He anticipated in 2012, three years before s.76, that 'reframing domestic violence as coercive control changes everything about how law enforcement responds to partner abuse, from the underlying principles guiding police and legal intervention, including arrest, to how suspects are questioned, evidence is gathered, resources are rationed . . . ' (Stark 2012, p.213). Data from this research project suggest, however, that legislative change is not, in and of itself, enough to effect the kind of 'reframing' that Stark correctly identifies is needed. Police understandably need help to move from the violent incident model to a coercive control model, despite the existence of legislation that makes that shift possible.
That there is a need for such a shift was unanimously recognised by the senior (specialist) police officers spoken to as part of this research. P04 commented that 'the mindset of what we were taught as investigators many years ago has to change'. P04 referred to the new mindset as one of 'professional curiosity'. P01 similarly stated that the policing of coercive control was a 'mindshift challenge'. But equally, there was an awareness among these officers of the challenges that the coercive control model would bring. During the focus group, FGP03 explained that even the basic procedure of taking a statement was problematic in the context of the coercive control model: 'I think the challenge for first responders is you are not asking them to take a statement about an event. So if it's an assault, or a criminal damage there's an event. Whereas obviously with coercive control you are telling a narrative, a story -that's always going to be much more difficult'.
In addition the narrative of coercive control, its 'story', was perceived as 'murky', 'a kind of grey area' or a 'mindset challenge' (P01). While the legislation itself was 'straight-forward' the behaviour being regulated was not. P01 explained: it's not as simple as -just training people on what the law is. Actually -telling people what the offence is is relatively straightforward. It's not a difficult offence to explain. And anyone who has got any knowledge of stalking -it mirrors it, it's fine. But actually how do you apply it? P01 went on to explain that his officers: P01 is right to appreciate that his officers need help. Coercive control is complicated. It is made up of interlinking behaviours, with which police may, or may not, be familiar; behaviours that are organised around an abuser's strategic, controlling intent. It is not possible to locate that strategic intent in time and space in the same way that you might locate, for instance, a domestic assault. But coercive control is not, or does not have to be, a 'grey area'. It is possible to determine where (to use P01's language) coercive control 'sits'. It should be possible to empower officers to make decisions about someone else's life and relationships as they do in the context of violence. There are 'thresholds'.
The 'help' police officers need is an accessible, working model of coercive control expressed in the language in which they will encounter it. The next section of this article shows how this could be done.

Part Two: A Working Model of Coercive Control
The idea of a working model of coercive control is not new, but it is an idea that is still in the process of development. Work began on modelling control in the 1980s with the well-known 'power and control' wheel developed by Pence and Paymar in Duluth in Minnesota (Pence and Paymar 1986). This work has been continued by, among others: Anderson (2009) (2008); Stanko (2006); Stark (2007); and Tadros (2005). The Duluth power and control wheel is chiefly descriptive, and consists of labelling the different constituent parts of coercive control. The work of later authors, and in particular that of Dutton, Goodman, and Stark, is 'moving away from merely listing the abusive or controlling behaviour of batterers to a more theoretical approach focused on the concept of coercive control' (Johnson 2008, p.14).
The empirical work was an opportunity to develop the theoretical work of later authors, and to consider how it might be tailored more specifically for use by police. For example, Dutton and Goodman organise coercive control into three distinct theoretical dimensions: 'setting the stage', 'coercive behaviour', and the 'victim's response'. The survivors and IDVAs interviewed agreed on the existence of these structural dimensions, and they were comfortable with the labels 'coercive behaviour' and 'victim's response'. They were not as comfortable with 'setting the stage' as a label, which they felt was too neutral, and too located in time and space. Instead, they preferred the label 'grooming', which they felt better reflected its ongoing nature. FGP03 pointed out that 'there are very clear parallels in the grooming process' with the work that officers do in relation to child sexual exploitation. Grooming as a term, she concluded, is 'very useful. It's very similar'. 'Grooming', 'coercive behaviour', and 'victim's response', were then used as reference points around which survivors were encouraged to frame their personal experiences. Each is reviewed in turn, below.

Grooming
Grooming is a critical component of coercive control. Victims of coercive control are vulnerable, but not because they are weak, character-deficient, or mentally unwell. They are vulnerable because they have been groomed. Survivors and IDVAs conceptualised the perpetrator tactics that constituted grooming around the labels: 'courtship', 'fear', 'emotional abuse' and 'isolation'. These four constituent parts to grooming will be looked at in turn.
Courtship and disclosure IDVAS and survivors described 'courtship' behaviours that were not as they first seemed. Behaviour that was interpreted as romantic confused and disorientated the survivors. SV02 commented that 'the messages were so hard to follow. Because him wanting me seemed to be -I think I confused that with love. You know. He was really attentive in that kind of way'. Survivors expressed their confusion; that they felt flattered and disorientated. The attention was often welcomed, because the significance of the behaviour was not apparent.
Survivors spoke about the grooming period as intense. SV02 said: 'and I would say that the first three months were like being in a complete bubble. We were inseparable. And everything just got put on hold'. During that intense period, SV02 was encouraged to share details about even the most intimate parts of her life. She explained: 'he would talk about his values of needing absolute loyalty, absolute transparency . . . having any sense of space or privacy or independence was kind of seen as not acceptable'. This was a common theme across survivors' stories. Dazzled by the intensity of the courtship, women gave perpetrators access to their lives: to their homes, their families, and even their most intimate secrets.

Fear
For many survivors, recognition that they had been groomed -that is that their abuser's initial attention was more about 'grooming' than 'courtship'came as a result of the onset of fear. This is often a transformative moment. IDVA05 told the disturbing story of a client who was badly frightened on her honeymoon: Her story was that everything was groovy, no issues, they got married they went on their honeymoon, and he strangled her with the bathroom towel. Really, really badly. There was a horrific, traumatic incident when he strangled her almost to death with the bathroom towel . . . So then after that for that six years of their relationship -. . . he never ever again used physical violence on her but whenever there was a moment of tension he would go to the bathroom and he would bring out a towel, and he would put it on the table. And that was the sign; and then she would just be, like, 'and then I would just give in -I would just do whatever it is he was trying to get me to do'.
The towel incident is a good example of the significance of groominginduced vulnerability. The placing of the bathroom towel on the kitchen table throughout the six years of their relationship was an instrument of terror but without context would seem innocuous.
Exhaustion and emotional abuse Living in a state of fear is exhausting. The exhaustion is further exacerbated by the perpetrator's emotional abuse. For the respondents, emotional abuse was a visible tactic used by the perpetrator that facilitated and accompanied the grooming and also (as is explained below) the controlling or coercive behaviour.
The emotional abuse took similar forms for many of the survivors and IDVAs interviewed. The gendered emotional 'put-downs' contained criticisms of victims' roles as mothers and homemakers -criticisms of the way they dressed, the way they cooked, the way they looked after their children. At its most extreme, emotional abuse is humiliating and degrading. FGP02 described an incident that had been attended by a colleague. The victim had reported a domestic rape, and the colleague had gone to her house to interview her. The interview took place in the kitchen, and FGP02's colleague noticed a dog bowl on the floor by the fridge. She noticed that the victim did not appear to have a dog, so she asked about the bowl. The victim explained that that was where she ate her supper. The alternatives (for that victim) were worse.

Isolation
In every case emotional abuse was accompanied by the removal of emotional support. Survivors talked about the removal of support in terms of separation from family and friends and a crippling sense of isolation. IDVA06 described the process by which the perpetrator removes emotional support from the victim as follows: Although it does happen with friends, especially with male friends if you are a female victim, but they close down their friendship group. Initially it's people that aren't good for them because they think they are being used. 'They use you, it's only because you are so nice . . . ' kind of thing. And then it's 'he fancies you' . . . so the woman disengages from the males in her company so it can go that way as well.
Emotional abuse has a profound and long-term effect on a victim in the way that it increases vulnerability. Without support, it is difficult for her to resist the world-view that he has and wants her to share (Stark 2007). As IDVA06 put it: 'drip, drip, and it's become her voice, as well'. This has emotional and cognitive repercussions that are discussed in relation to the victim response part of the model, below.
Overall, grooming is an essential component of coercive control. Initial 'courtship' gives way to behaviour that terrifies, emotionally abuses, and isolates survivors. In this way perpetrators create survivor vulnerability. As is explained below, often it is the vulnerability that makes the difference between a threat, and a threat that is, for the survivor, a 'credible' threat. It is when a threat is credible that a demand becomes coercive.

Coercive Behaviour
Coercive demands are thus the product of a relationship between demands, credible threats, and surveillance. Survivors described demands that were small, or big; but pervasive: controlling your entire life, how you eat how you sleep, what you wear. When you answer your phone, everything to how you look, your hair, how you cook you know, just literally your life is consumed by control. (FGSV03) Survivors often talked about demands linked to the perpetrators' creature comforts, the way in which she looked after him and his children, and the way that she presented herself to the world around her -what she cooked, how she cleaned, what she wore. One perpetrator bizarrely hid gold coins around the house for his partner to 'discover' as she did the housework. When he got home from work in the evening he would hold out his hand for the coins. If she had not found the correct number of coins, it would mean she had not cleaned properly, and he would punish her. SV02 talked about her evening regime: Thus, demands are not only obvious and discrete: 'find those gold coins'; they are also 'integrated seamlessly into the day-to-day interactions of the partners' lives' (Dutton and Goodman 2005, p.749), as in the way in which SV02 knew that her husband would have expectations for his dinner, as he did on a daily basis, without him warning her specifically each day that this would be the case.
In every case, the demands were coercive because the threats (both discrete and integrated) were credible. Survivors learnt (with good reason) to be fearful. This is where the grooming, and an appreciation of context are important. Survivors do not 'give in' to perpetrator demands because they are inherently weak or flawed as individuals. They obey, because they are rightly fearful of the consequences if they do not. They understand the threat posed by the perpetrator because they know that he has access, and that he is dangerous. The victim who has to find the gold coins could tell you why she was afraid. She could tell you what might happen if she missed one. The survivor on her honeymoon could explain the significance of the bathroom towel on the table. The survivor in the police story could tell you why she agreed to eat her supper from a bowl on the floor, like a dog.
The final piece of the jigsaw is the surveillance -she also might have reason to be fearful that he will know if she disobeys him because he is spying on her. Some form of surveillance was present in the lives of all of the survivors interviewed. As with everything else, it can seem innocuous if it is not contextualised. One senior police officer spoke of a perpetrator's hourly calls to his wife's work, that seemed innocent enough at the time that the victim called police for help. She was later murdered, and the calls were reassessed once their significance was better understood. IDVAs and survivors all spoke of time monitored at home, at work, even time spent on the toilet. Technology gives perpetrators new powers -tracking devices on mobile phones, administration access to all email accounts, remote access to the home computer.
The most important point for police to understand about the dimensions of coercive control is the way that they work together. Towels, dog bowls, and gold coins in unexpected places, are examples of insidious behaviour patterns that are bespoke. They would therefore not be useful items on a behavioural checklist. What is useful is the recognition that these three survivors have much in common: namely, the way that they were groomed, and the insight the grooming gave them into perpetrator capability. The last part of the model explores the impact that the grooming and the coercive behaviour is likely to have on its victims.

Victim's Response
The survivor response to coercive control is the third dimension to the theoretical model. Survivors talked about fear, instability, and personality change. Every single survivor spoke of the adverse effect on her daily life in terms of the fear and instability at its core. SV02 reported that: 'it was very much setting up this sense of "here are my rules", and immediately, although you can't see them, you are walking on eggshells'. FGSV03 commented that: it's like living on a roller coaster. It's like going on a train journey and never knowing which stop you are going to get off -if it's going to be a nice stop? Or a bad stop? And the day is like that every day. Basically. You don't know how the day is going to start. And you don't know how the day is going to end.
For most survivors, accompanying the instability roller coaster is an elusive sense of personal control: if only they could behave differently, the perpetrator's abusive behaviour would stop. This perception on the part of the victim of the link between what she does and how he might react goes to the core of the relationship between a victim's fear and the impact it has on her daily life. The day is spent trying not to crush the eggshells underfoot. IDVA07 elaborated on this: one of the reasons that they blame themselves is that they feel then like they have a degree of control, like they can prevent it from happening again. So it's like I was raped because I did X, so if I don't do X again, then I won't get raped. Which means that I can now have control over my life so that I don't get raped. The reality is that's not why she got raped. She got raped because he's a perpetrator. Part of blaming yourself is about giving yourself back a degree of control.
The most significant short-term part of the impact that the control has on the survivor is, therefore, the way that she moderates her behaviour: the 'X' in 'so if I don't do X again'; but the long-term context is more profound. SV02 explained how she would blame herself for the abuse, 'and of course anything I did or didn't do, wasn't just wrong whether I did it or didn't do it, but it was also an example of my badness, my passive aggression, my withholding, my darkness . . . '. SV02 internalised the abuse even as she was describing it in an interview environment: 'my badness, . . . my darkness'. This internalisation goes to the heart of the impact that controlling or coercive behaviour has on its victims as it affects the way that they see themselves and the world around them: 'underneath something emotional about you has changed' (IDVA02). In other words: they are at a point where they are pretty much believing what has been said over a period of time to the extent that they found it hard to see him as guilty of a crime because the blame was entirely on themselves and it informed who they were. (IDVA06) Stark (2007) puts it: 'he changes who and what she is ' (p.262, italics added).
The impact of the control on the victim is devastating. She exists in a constant state of fear that she has not moderated her behaviour sufficiently to avert catastrophe for herself and her children. Her fear is real and not imagined, as it is based on a realistic appraisal of the perpetrator's capabilities. But for the survivor respondents the fear was not the worst effect of the abuse. Dutton talks about the way in which the psychological impact of abuse goes beyond symptom-focused conditions such as anxiety to include 'the ways in which battered women have come to think about the violence, themselves, and others as a result of their experiences' (Dutton 1993(Dutton , p.1217. Survivors explained that worst of all was how they learned to blame themselves for the position in which they found themselves and lost confidence in their ability to make decisions about their own and their children's lives. IDVAs emphasised that the survivor will not always be aware of these changes in herself when she first reaches out for help: 'the time when they are first talking to us they are probably not the person who they really are but I don't know if they are able to identify if that is the case' (IDVA07).
The police who meet the victim for the first time as she reaches out for help are not in a position to judge personality change. But assessing fear levels and instability, and recognising that there might, in fact, be significant personality change, are all very helpful in terms of addressing some of the hurdles identified in Part One of this article. Part Three, below, explores how an understanding of the three dimensions of control described heregrooming, coercive behaviour, and victim response -might help police address the challenges identified in Part One.

Part Three: Moving from 'Force' to 'Control'
Part One of this article explained how empirical work with police identified that the introduction of s. 76 has not, in itself, facilitated a smooth transition from the violent incident model to the coercive control model. While senior officers were quick to grasp the opportunities that moving to a coercive control model presented: such as a more nuanced (and therefore empathetic) understanding of the victim experience, and better safeguarding decisions; there was a lack of certainty about what is expected in the context of policing coercive control. A lack of perceived boundaries around the 'new' criminal behaviour is of concern to officers who prefer to deal in 'black and white -or black and blue'. Moving from an incident focus to a relationship focus again raises issues -an incident has a degree of specificity (in time and space) that a relationship does not. All officers agreed that better and more informed 'professional curiosity' was needed but were not clear on how this might be achieved.
It is the position of this article that police need to understand the model of control set out in Part Two in order to do the necessary reframing of the way they approach domestic abuse. Such an understanding will not 'solve' all of the 'problems' identified. Investigating a relationship is more complex and time-consuming than investigating an incident. Coercive control is more complicated than other types of crime. But time spent investigating an incident is not time 'saved' if it is not the incident that is, in fact, the problem; a criminal offence cannot be overlooked because it is complicated.
In general, understanding that coercive control has a clear infrastructure will give first responders the confidence they need to judge the relationships they encounter. Understanding that coercive control is ongoing, rather than incident specific will help alleviate the frustration that is felt by police as a result of mistaken assumptions of decisional autonomy. More specifically, understanding the relationship between grooming and coercive behaviour will help direct officers to asking the right questions. This process is not watertight. Perpetrators are manipulative and cunning -'some of the most nasty people you could come across' (P01). No amount of 'understanding' can guarantee that an officer will think to question the presence of a dog bowl. But what is certain is that without that understanding, the significance of the dog bowl will remain invisible. Understanding, in theory, how the different dimensions of coercive control work together will help officers to 'see' it. And finally, understanding that the fear that a victim expresses is real, and not imagined, and is based on a realistic assessment of what the perpetrator might do to her will help responders (and their supervisors) make a more accurate assessment of risk. 'Professional curiosity' (P03) is, indeed, the answer, and it is best nurtured via an understanding of where to direct it.

Conclusion
This article has reviewed the police response to a new law that presents opportunities and challenges in equal measure. It was, in part, creative police use of the Protection from Harassment Act 1997 that paved the way for the inclusion of the coercive control legislation in the Conservative government's 2015 crime bill, nearly 20 years later. Originally designed to deal with 'stalking' (Finch 2001), it became apparent within three years that police and prosecutors were using the harassment regulations as a valuable tool to address insidious behaviours arising from domestic incidents where couples had separated (Finch 2001;Harris 2000). S. 76 was introduced specifically to fill the gap left by the insistence of the courts that the harassment legislation did not apply to couples in a continuing intimate relationship. 4 Stark warned, in 2016, that 'bluntly put: coercive control can disappear as quickly as it surfaced if police respond passively or actively resist enforcement, as some officers have already stated they will do' (Stark 2016, p.347). The early indications are that senior police are excited by the potential of s.76. Nevertheless, there is a risk (also picked up by journalists) 5 that in the absence of mandatory training for police, coercive control will 'disappear as quickly as it surfaced'.
The College of Policing, together with domestic abuse charity Safelives (formerly Co-ordinated Action Against Domestic Abuse (CAADA)) designed a one-day training programme on control for first responders entitled 'Domestic Abuse Matters'. It was piloted in Hertfordshire Constabulary in England in 2015. The training is focused almost entirely on coercive control, with an emphasis on 'upskilling' officers, and on cultural change. Unfortunately, the training is not compulsory. As it is expensive, and involves the release of officers from active duty for the period of the training, it is likely that the forces that are short of time and resources and who need the training the most will be the ones least likely to get it. P01 explained that recent government cuts put further pressure on staffing levels: 'we just don't have enough officers to be taking officers out to do that training'.
It is important not to understate progress. The development of the way that police respond to intimate partner abuse is a story of progress, but it is a story that needs to continue to develop. It is well known that, tragically, two women die each week in the UK at the hands of someone with whom they are, or have been, intimately connected (Walby and Allen 2004). A recurrent theme in the literature on police culture is that many police officers join the force because they want to keep the public safe (Cockcroft 2013;Loftus 2009;Mclaughlin 2007;Reiner 1984;Westmarland 2001), a finding supported by this project. P01 explained: 'but by and large people do the job to make a difference, and by and large, they join to protect vulnerable people'. He concluded: 'and I think what we are seeing now is absolutely where it needs to be -which is -we put vulnerable people at the heart of what we are doing, and protect them first and foremost, and then we deal with what comes out of that'.
Dealing with 'what comes out of that' is moving to a coercive control model of policing intimate partner abuse. It is the next chapter of the developing police response to coercive control. On its own, s. 76 will not achieve this. Essential national-level compulsory training in coercive control must be implemented immediately to make it happen. 6 Notes 6 Acknowledgements: The author would like to thank Professor Heather Keating, Dr Tanya Palmer, and Ken Donner, for guidance and support, with particular thanks to Ken Donner for his editorial assistance. The author also acknowledges receipt of financial support from the Economic and Social Research Council, UK (grant no, ES/J500173), for the research, authorship and publication of this article. Finally, the author would like to thank the courageous survivors who took part in the research. The author regrets that data sharing agreements cannot be entered into due to the terms of the informed consent that was agreed with survivors and with the University of Sussex Ethics Board.