Gender, risk assessment and coercive control: Contradictions in terms?

Abstract

In December 2015, the criminal offence of coercive control was introduced in England and Wales. Whilst, in this legislation, this concept is presumed to be gender-neutral, there is widespread agreement that coercive control is gendered. Using empirical data gathered in one police force area in the South of England, this paper offers an exploration of the feasibility of the extent to which existing risk assessment practices and understandings of risk embedded within them, can incorporate the phenomenon of coercive control. The findings highlight concerns about gender-blind, incident-led (rather than process-led) approaches to assessing risk when these approaches are set against victim/survivor concerns. These concerns highlight the inherent problems embedded in the contemporary gender-blind embrace of the concept of risk as assumed in practices of risk assessment.

Introduction

Risk assessment tools have pervaded policy responses and practices to women as offenders and women as victims (of crime) for some time. In relation to their use for women as offenders, Davidson and Chesney-Lind (2009) conclude that because of the inability of such tools to ‘see’ gender, they have the capacity to both under-classify and over-classify women at risk all at the same time (see also Chesney-Lind and Pasko 2004). This work reflects the well-established evidence documenting the gendered nature of risk (see inter alia Walklate 1997; Chan and Rikagos 2002; Hannah-Moffat and O’Malley 2007; Hannah-Moffat 2015). However, little of this gendered thinking has been transferred to the use of risk assessment tools in relation to intimate partner violence (IPV; but see Walklate 2018). There is a wide range of such tools available and deployed in relation to IPV (e.g. the spousal risk appraisal guide, the Propensity for Abusiveness Scale and the Partner Abuse Prognostic Scale (all reviewed in Hoyle 2008: 327). Currently, the Domestic Abuse, Stalking and Harassment, and Honour Based Violence (DASH) model is favoured by most police forces in England and Wales and is the one featured in the discussion that follows. Interestingly few of these tools have been subjected to empirical evaluation as to their outcome efficacy (McCulloch et al. 2016). However, recent work by Medina et al. (2016), Turner et al. (2019) and Graham et al. (2019) point to some shortcomings in the implementation of them in relation to their predictive outcome, if not in relation to their conceptual framing. The purpose of this paper is to focus attention on the conceptual framing underpinning these tools and to re-centre understandings of gender in relation to risk and risk assessment practices as one way of making sense of the potential for risk assessment in relation to coercive control.

To be clear, given the increasing presence of concerns to criminalize coercive control across the globe (proposals for moves in this direction are evident in Australia, Canada and in some states of the United States, it already having been criminalized in England and Wales, Scotland and Ireland, with similar laws existing in Tasmania), this paper is concerned to challenge and extend what is taken for granted when two contested concepts, risk and coercive control, are conjoined. Based on empirical data, it considers how they might be better framed and progressed if their respective gendered assumptions were placed in the foreground rather than either simply assumed and/or left unacknowledged. In pushing for a more conceptually nuanced understanding of risk in particular, the possibility for improved practices and implementation may follow.

In order to do this, the paper falls into five parts. The first offers a brief overview of the efficacy of risk assessment practices, as currently conceived in relation to IPV¹. The second raises a series of questions concerning the potential of these practices to address coercive control. The third considers findings from a British Academy-funded project conducted in conjunction with a policing partner in the South of England. This project paid particular attention to comparing and contrasting policing practices and victim–survivor understandings of risk, risk assessment and coercive control as experienced and implemented within that particular police force. In the fourth section, we reflect on these findings and the dilemmas they pose for contemporary risk assessment practices generally. In the fifth and final part of this paper, following the work of Merry (2016), we suggest that the conceptual challenges posed for risk-assessing coercive control are the problems of risk inertia and time inertia (Walklate et al. 2020). We go on to suggest that these problems may only be overcome when the conceptual limitations of risk and risk assessment are fully acknowledged by IPV researchers, policymakers and practitioners alike.

The risk ‘turn’ in criminal justice policy

The criminal justice policy embrace of risk is now well established and, as Short (1984) observed some time ago, it reflects assumptions about what is actionable, measurable and how this might be done. Whilst a measurable and actionable understanding of risk can serve the interests of policymakers (Carlen 2008), it stretches the concept of risk in ways that invite partial understandings on the nature of social reality (Mythen 2014). Specifically, as Mythen (2014) has cogently argued, at the conceptual level, risk reflects a limited understanding of power and power relations emanating from risk theory, which results in partial understandings of human agency and a tendency to stretch the concept by putting it to uses for which it cannot be applied (see also Walklate 2018). In terms of risk assessment tools, e.g., and as much work on the ways in which professionals utilize risk assessment tools illustrates (see inter alia Ansbro 2010; Kemshall 2010; Werth 2017), judgements and feelings frequently supplement what is measured by such tools (see also Walklate and Mythen 2011). Thus, there is a well-evidenced tension between the claims made for the efficacy of risk assessment tools per se and how they are used in practice. In other words, what such tools are really ‘measuring’ can be both complex and complicated. In a similar vein, there is a wide body of evidence (some of which is cited above) pointing to the gendered dimensions of risk and risk assessment and this partial vision of risk, as it is embedded in risk tools, is particularly pertinent to their use in relation to IPV.

The partial dimension of risk and risk assessment of particular interest to this paper is the failure to ‘see’ gender. Such a failure inevitably takes its toll on the capacity of policy and practice to make sense of either men’s or women’s lives but renders women’s experiences of victimization (the focus of this paper) particularly difficult to identify, assess and respond to (see also Chan and Rigakos 2002: 756). Casting a gendered lens on risk is revealing insofar as it facilitates the identification of what counts as risk, who counts as risky, where and when (see also Walklate 1997; ,2018; Chan and Rigakos 2002; Shalhoub-Kevorkian 2016). Taking the failure to ‘see’ gender together with the focus on what is measurable, both of which are inherent in risk assessment practices of all kinds, affords a vision of risk and its measurement, which is incident focused as opposed to process appreciative (see inter alia Westmarland and Kelly 2016). Binding risk with risk measurement in this way takes as given a conceptually gender-blind understanding of risk, whereas much work has demonstrated that risk is not gender blind but gendered (see inter alia Davidson and Chesney-Lind 2009; Mythen 2014; Hannah-Moffat 2015). Although ‘risk factors’ informing domestic abuse risk assessment tools may have been informed by feminist research, it is the conceptual apparatus of risk itself that fails to ‘see’ gender, thus rendering the gendered nature of risk hidden from view (for a fuller discussion of gender blindness and its impact on criminology and victimology, see Fitz-Gibbon and Walklate 2018: chapter 1). These conceptual limitations fundamentally frame risk assessment practices in relation to IPV in very particular ways.

Risk assessment and IPV: problems and possibilities.

It is frequently unclear as to whether the risk assessment tools deployed in response to IPV are intended to predict, prevent or protect the victim at risk of such violence. Indeed, the many and varied tools focussing on IPV might inflect attention in any and/or all these directions. Moreover, given the focus of this paper, it is important to note that such tools can be differently informed by feminist work and can also differently engage women themselves in assessing their own level of risk. Indeed, there is evidence to suggest that including the victim’s own understanding of their level of risk is at least as accurate as many of the tools claiming a predictive capacity (Wheeler and Wire 2014). As Day et al. (2014: 581) report:

There is some evidence to suggest that partner estimates of risk can [also] consistently predict future victimization, with approximately two thirds of victims correctly identifying their assessed level of risk.

This is a finding consistent with a wide range of work on IPV (see e.g. Kirkwood 1993; Smith et al. 2010). Nevertheless, the capacity for criminal justice professionals to listen and hear women’s voices remains problematic (Wheeler and Wire 2014). This disjunction might be associated with practitioner understandings of the nature of IPV (Robinson et al. 2018) and/or with the time constraints in which front-line police officers (usually the first responders to such incidents) operate (Brennan et al. 2018). Both factors can make listening and hearing victim-survivors difficult. However, it is also feasible that this disjunction might be inherent in the conceptual assumptions embedded in the kinds of tools in use.

As indicated above, there are a wide range of risk assessment tools available and used in the context of IPV. Currently, the DASH model is favoured by most police forces in the United Kingdom. McCulloch et al. (2016) review this tool along with nine others with European Institute for Gender Equality (2019), offering a detailed review of the use and deployment of the tools most used in Europe. Whichever risk assessment tool is deployed in the context of IPV, its purpose is, at a minimum, to inform future practice responses for those deemed ‘at risk’ of further violence. Indeed, Campbell et al. (2007) suggest that the key feature of risk assessment is to identify the seriousness of prior violence (particularly physical injury) and the nature of its escalation. These tools (their predictive limitations notwithstanding, see Medina et al. 2016; Turner et al. 2019) produce scores, and these scores are intended to convey the level of risk present at the time the assessment is made. These scores are usually translated as high, medium or standard risk. The extent to which such scores are accurately constructed and thereby inform future practice meaningfully has been questioned (see inter alia Thornton 2017; Walklate and Hopkins 2019). However, the production of these scores rests on the assumption that not only is such risk measurable by the instrument in use but also that risky relationships, as defined by the tool in use, escalate into riskier relationships. Thus, the higher the risk score, the higher the chance of IPV becoming at least a repeat demand-led issue for policing responses, or in a worst-case scenario, becoming intimate partner homicide. Yet, as McPhedran and Baker (2012) have pointed out, neither of these assumptions have been effectively evidenced or tested. Yet, they can and do inform practice.

There is, however, a further assumption in risk assessment tools. This is the assumption of a turning point in a relationship. This presumes that it is a particular type of incident that can be indicative of violence escalation and/or it is a particular type of perpetrator whose behaviour is indicative of this. Incidents of non-fatal strangulation are frequently taken to be one such turning point as is the point of relationship separation (Dekeseredy et al. 2017). In either case, the presence of physical violence, as far as front-line officers are concerned, is key in informing what Robinson et al. (2016a) have called the ‘constellation of risk factors’ that can underpin their views of the level of risk present. It goes without saying of course that physical violence is often visible and measurable and, so, lends itself most readily as a marker of either the presence of escalation or a turning point.

To be explicit, the partial and gender-blind understanding of risk embraced within risk assessment tools reflects an incomplete understanding of women’s everyday lives (qua Shalhoub-Kevorkian 2016) and their relationship with and understandings of risk (qua Day et al. 2014). Consequently, such tools, in their endeavours to measure and assign a numerical value to risk, can result in false negatives and false positives (see inter alia Shalhoub-Kevorkian 2016; Barnham et al. 2017). Put at its starkest, this means that many women die at the hands of their partners who have either never been risk assessed since they have never had any contact with the criminal justice system (Thornton 2017) and/or who, on their risk journey through the criminal justice process, have been inappropriately risk assessed (see inter alia Hester 2011; Walklate and Hopkins 2019; Barlow et al. 2020). Importantly, these observations serve to remind us that these tools smooth out gendered understandings of who is a risk and what is risky (as indicated in the discussion above) and simultaneously smooth out the peaks and troughs of people’s lives. Thus, such tools also embed assumptions of relationship linearity over and through time where none may exist. This linear thinking is reflected in the embrace of both the concepts of escalation and the nature of turning points in relationships as measurable and identifiable relationship markers. Of course, both of which may be present in any relationship but may not be in the shape and form assumed by the risk assessment tool, i.e. may not be linear. Thus, the assumption of linearity can also fail to capture the reality of women’s lives and to appropriately assess risk in the light of those realities.

Following Merry (2016), this paper suggests that the observations above constitute risk inertia (see also Walklate et al. 2020). The concept of risk inertia captures the way in which dominant and gender-blind understandings of risk are used as though these understandings reflect what risk means for women living with violence, when it is possible that they do not. Risk inertia assumes that risk is measurable and individual, whereas risk and understandings of risk can also be a product of cultural and social relationships (see Hart 2016; Carrington et al. 2019: chapter 2). The question remains whether the turn to coercive control, which focusses attention on processes (i.e. understanding the nature of relationships over and through time) rather than incidents (thinking about relationships as they are articulated at one point in time), can be measured by such tools.

The criminal justice turn to coercive control

Thinking about IPV through the lens of coercive control is not a new phenomenon (Schechter 1982; Johnson 1995; Stark 2007). What is relatively new is its increasing embrace within the criminal justice policy domain. Early efforts to criminalize one understanding of coercive control were developed in Tasmania (Tasmanian Family Violence Act 2004), with England and Wales criminalizing a different understanding of coercive control in December 2015² followed by Scotland in 2018, reflecting yet another differently emphasized definitional embrace of this concept. These different definitional approaches illustrate that this is a contested concept in relation to its application to the criminal law (see inter aliaFitz-Gibbon et al. 2018; McMahon and McGorrery 2020). In a recent overview, Hamberger et al. (2017) identified 22 different ways of defining and operationalizing coercive control. Nonetheless, in summary, coercive control draws attention to the hidden and intrusive consequences of IPV, including e.g. psychological abuse, intimidation and isolation, which do not leave physical scars. This concept is designed to capture the long-term, ongoing nature of abuse that can pervade women’s routine daily lives over and through time and sometimes across generations at the hands of men (Stark 2007).

Williamson (2010) has pointed out that reframing domestic abuse to take account of its presence over and through time poses (at least) two new, but important, challenges for practitioners. First, it focusses attention on the impact of a wider range of abusive behaviours (some criminalized, some not) on the victim. Second, it moves the focus of attention of criminal justice professionals away from responding to victims in terms of an individual incident-led approach to a process-led one. Coercive control demands that practitioners, and the tools at their disposal, take account of a course of conduct. In other words, it makes clear the need to focus on the cumulative effect of the minutiae of everyday behaviours, some visible, some not, the total effect of which are abusively controlling (Barlow et al. 2019). It is easy to see how this process-led appreciation of the violence(s) perpetrated against women stands in contrast with the incident-led approach embedded in much of the contemporary criminal justice policy embrace of risk and risk assessment tools as discussed above. Thus, as Johnson et al. (2017: 16) argue, coercive control carries significant implications for practices of risk assessment. They say:

Relatively rare events like femicide are difficult to predict but the findings of this study suggest that risk assessments that prioritize assaults that intensify in frequency and severity, in some situations, may overlook the danger inherent in efforts to control, isolate, and terrorize women when assaults and injury are absent or considered by law to be minor.

In what follows, we consider in more detail the capacity for the risk assessment tool currently deployed by the partner police force in this study (DASH), and the officers completing the associated risk assessment check list, to take account of the dangers inherent in coercive control as intimated by Johnson et al. (2017) above.

DASH is the risk assessment tool used by most, but not all police forces in England and Wales. It is available in several languages and intended for professionals to assess the level of risk being faced by an individual in various kinds of relationships, such as some family dynamics and intimate partner relationships. The higher the score generated by DASH the greater level of risk is assumed to be present. To be clear, in what follows this paper is concerned with the tensions posed by understandings of risk and the practices of risk assessment in relation to IPV generally and how these assessments are experienced by the women in receipt of them and not with the DASH tool, and its strengths or limitations, in particular. This just happened to be the tool used by this force. It is also important to note that this data pertains to one point in the risk assessment journey for a victim/survivor: the response by the front-line officer and the women’s experiences of that response. Those women for whom this journey extends beyond this first point of contact may well be listened to differently (e.g. by an Independent Domestic Violence Advisor; IDVA). The ways in which those women referred to multiagency risk assessment conferences, Safeguarding Hubs and/or IDVAs and the ways in which they were differently responded to at various points in their risk journey is the focus of a different discussion (Barlow et al. 2020).

Overview of this study

The data presented here was gathered as part of a British Academy-funded study done with the support of Women’s Aid (England) and a policing partner in the South of England. The study comprised four data-gathering phases: a quantitative analysis of all (anonymized) recorded crimes of domestic abuse (i.e. crimes marked with a domestic abuse ‘flag’ on the partner’s information management system (IMS) from 1 January 2017 to 1 January 2018); an online survey circulated to all police staff; focus groups and semi-structured interviews with police officers and interviews with victim-survivors of coercive control. Ethical approval was granted by Lancaster University’s Faculty of Social Sciences Ethics Committee prior to data collection. Each of these data gathering processes will be discussed in turn before considering the collective themes emanating from them.

The quantitative, force-level data provided important contextual information for the other phases of data collection. After processing and removal of duplicates, this data showed that there were 5,230 recorded crimes of domestic abuse in this policing area for the calendar year 2017.³ 1.8 per cent (93) of cases were recorded as coercive control (see Table 1). This number appears low particularly when compared with other offences (e.g. there were 1447, 27.8 per cent occasioning actual bodily harm (ABH) cases over the same time period), though these figures are very much in line with the generally low use of the offence of coercive control for 2017 (Office of National Statistics 2018). We have argued elsewhere that domestic abuse cases recorded as ABH by police officers often feature evidence of coercive and controlling behaviours that are missed (Barlow et al. 2019).

Given what is already known concerning the propensity of police responses to IPV when physical violence is present (Robinson et al. 2016a), demographic information and police response data relating to coercive control and ABH were compared specifically with this in mind. We wanted to determine if there were statistically significant differences between both offence categories in terms of sex and relationship in an intimate partner context. The sex and relationship variables were tested using a chi-square test, which determined that there is an association between type of offence and victim sex and offender sex and type of relationship (see Table 1).

For context, the police force-level data highlights the well-documented gendered nature of IPV, particularly coercive control, with 95 per cent of victims of coercive control being women and 93 per cent of perpetrators being men. Coercive control cases were also more likely to occur in an intimate partner context (89 per cent). Comparatively for ABH cases, 78 per cent of victims were women, 83 per cent of perpetrators were men and 80 per cent of cases occurred in an intimate partner context. Additional analysis of this data is presented below.

The online survey asked officers to share, anonymously, their understandings of, and attitudes towards, coercive control and IPV more broadly. A web link to the survey was sent via email to all force staff and a reminder sent one month later. One hundred ninety-eight officers (approx. 20 per cent response rate) completed the survey. Respondents were of varying role and rank (including call handlers, front-line responders and senior investigators). Fifty-six per cent of respondents were male, 53 per cent female and 1 per cent unspecified. The median length of service for respondents was 11 years and 79 per cent of participants had received some form of domestic abuse training at some point in their career. To be noted is that 85 per cent had not received specialized domestic abuse training two years prior to completing the survey (i.e. they had not received specific coercive control training since its criminalization). One aspect of the survey invited participants to respond to questions in relation to two vignette scenarios. All officers responded to both scenarios. Scenario 1 involved a coercive control case without physical violence (this was presented first). This featured behaviours, such as monitoring, stalking, isolation and threats of violence. Scenario 2 involved a case of physical violence, with the male perpetrator punching the female victim in the face and pushing her to the ground.⁴ This method emulates that adopted by Robinson et al. (2016b) in which hypothetical vignettes were used to capture a sense of the ways in which non-physical abuse is understood by police officers when compared with physical abuse.

Focus groups and semi-structured interviews were conducted to supplement the survey findings. Recruitment for the focus groups involved an email being sent to all staff, followed by an invitation to a face-to-face briefing session for those who sought further information. Members of staff who wanted to take part contacted the researchers directly via email. A contact in the force facilitated the organization of the focus groups. Five focus groups with 25 participants in total (22 men and 3 women) were conducted. Four focus groups were conducted with front-line, investigating officers and call handlers and one with senior members of staff. Two further semi-structured interviews with senior, decision-making officers (both men) were also conducted. All participants were also asked to discuss the scenarios presented in the online survey with a view to exploring their understanding of, and attitudes towards, coercive control, including perceptions of risk, in more detail.

Finally, 10 interviews with victim/survivors of domestic abuse/coercive control were conducted. These women were seeking the support of Women’s Aid at the time of the interview and Women’s Aid supported the research team with the recruitment of participants. This was to ensure that the women were provided with appropriate support and, if required, counselling after the interviews had taken place. The interviews focussed attention on these women’s experiences of police responses, with a particular emphasis on their understandings of risk. The focus group and interview data were coded and analysed using grounded theory and thematic analysis (Braun and Clarke 2006) to identify over-arching themes in the data. In order to enhance inter-rater reliability, two researchers performed this analytic stage where themes were independently identified within the data and then compared and discussed to reach a thematic consensus. Two over-arching themes became apparent from this data collection process. These themes articulate the tensions between what might be called an incident-focussed approach to assessing risk (or, as Kelly and Westmarland 2016 have called it, incidentalism) and victim/survivors’ concerns with longer-term safety. Each of these will be discussed in turn.

Data Findings

Incidentalism

In comparing the police responses to coercive control to ABH in this police force area for 2017, a chi-square test was used to determine the relationship between variables. The test determined that there was an association between offence type and risk assessment, arrest and whether a case was charged (Table 2).

These results indicate there are significant differences between the police responses associated with ABH and coercive control. Of note is that, despite coercive control cases being more likely to be assessed as high risk (25.6 per cent compared to 15.8 per cent for ABH), ABH cases were more likely to result in an arrest and be charged. The higher risk assessment grading for coercive control cases may suggest that police officers have a good understanding of the risks associated with coercive control. However, the lower likelihood of arrest and solved rates suggest that, irrespective of coercive control cases being more likely to be risk assessed as high, the immediate risk to victims may not have been deemed as significant by police, therefore influencing arrest decisions. The survey and focus group data provided a more in-depth understanding of these different emphases.

When presented with two scenarios as part of the online survey, officers were more likely to assess the coercive control case as medium or standard risk (based on the DASH risk assessment criteria as used by this force). In the coercive control scenario (where no physical violence was present), a small number of officers assessed this as standard risk (11, 5.6 per cent), the majority classified this as medium risk (112, 57.1 per cent), followed by approximately a third, who considered the case high risk (66, 37.2 per cent). When compared with Scenario 2 (physical violence), none of the officers assessed this as standard risk, just over a quarter as medium risk (49, 26.3 per cent) and just over three quarters as high risk (149, 76.1 per cent). Similar risk assessment gradings were evident in the focus groups, with 21 (out of 25) participants grading the physical violence scenario as high risk, 4 medium and none as standard. In this data, the coercive control scenario was graded as high risk by 5, medium by 13 and standard risk by 7.

So, despite force-level data highlighting that coercive control cases were more likely to be assessed as high risk when compared to ABH, the survey and focus group data suggest that police officers perceived the coercive control scenario to be lower risk when compared to the physical violence scenario, echoing the work of Robinson et al. (2016a). The two dominant rationales for the medium/standard risk gradings in the coercive control cases was, first, the absence of physical violence, which was equated with a lower risk of harm for 16 participants in the focus groups, viz: ‘this would have been high, but there was no physical violence in there, so there was no evidence she was at further risk of harm’ (FG2, P1), and, second, the need for more information to definitively ascertain risk for 10 participants, viz: ‘this could have been high, but there isn’t enough information here really. I would need to investigate further before saying for definite what her risk was’ (FG1, P5).

These kinds of uncertainties were not expressed in relation to the physical violence scenario. Interestingly, three of the five participants who graded the coercive control scenario as high risk made their assessment on the basis of the potential escalation to physical violence, viz: ‘an argument has taken place here and even though there is no physical violence yet, you don’t want it escalating to that’ (FG1, P3). In line with other research findings, the presence of physical violence clearly informed the assessment of risk being articulated by these participants and also informed the kinds of (in)actions that they said would follow.

These data suggest that these assessments of risk also influenced the likelihood of arrest. Virtually all participants in the survey (183, 98.92 per cent) stated that they would arrest the perpetrator in the physical violence scenario compared to 76.1 per cent suggesting that they would arrest in the coercive control scenario. Qualitative reasons provided by participants for not making an arrest in the coercive control scenario included not believing arrest was necessary, e.g. ‘at this point, the perpetrator would need to be spoken to as he may not realise what he is doing is a criminal offence’. Similarly, 13 per cent of survey respondents suggested that they would choose to take the perpetrator to one side and talk with them informally in the coercive control scenario rather than arrest (compared to 2 per cent in the physical violence scenario). All participants (25) in the focus groups said they would arrest for the physical violence scenario compared to 16 in the coercive control scenario. The most common reason for making an arrest in the physical violence scenario in both the focus groups and the survey data was the perception that the victim was at risk of the violence escalating. For example:

‘if the perpetrator has already punched her, I would worry what he would do next’ (survey respondent 67);

‘I would want to get him arrested and safeguard the victim asap. This could easily escalate to something even more serious’ (FG3, P4);

‘given the fact there’s an actual injury, I think more officers would be inclined to arrest at this point, purely because of the injury’ (FG5, P2).

In the coercive control scenario, the main reason provided for not making an arrest was the perception that the risks to the victim were not significant, viz: ‘this hasn’t escalated to anything physical yet, so sitting them down, talking to them both would be the best thing to do at this stage’ (FG4, P2).

In summary, incidents of physical violence were perceived as higher risk by police officers when compared with coercive control. Thus, in line with other findings, when assessing risk, these officers tended to focus on what is visible, measurable and actionable: incidentalism. A senior, decision-making officer captured this issue nicely, stating: ‘I know that officers still focus on physical violence when they respond to a domestic. I know they have tunnel vision on injuries and black eyes. It’s difficult to know how to make that shift’.

Evidence of this ‘tunnel vision’ is not new but does serve to illustrate the problematic and partial policy embrace of risk and risk assessment and the tensions that it generates. This becomes more apparent when placed side by side with victim/survivor perceptions of risk.

Victim/survivor experiences: policing ‘here and now’ or women’s longer-term safety?

All ten of the women interviewed had contacted the police themselves about their experiences of domestic abuse/coercive control on at least one occasion, with six of the women having contacted the police on more than three occasions. Two of the women had experienced positive interactions with the police on one of the prior occasions they had called (both women had called the police on multiple occasions). They suggested that this was due to the individual officers being sensitive and empathetic to their experiences of abuse. All women felt on the whole that police officers did not understand their experiences of abuse or the actual risks they faced. They suggested the risk assessment tool used and the police understandings of risk more broadly tended to focus on the perceived risks of ‘here and now’ (i.e. isolated moments in time) rather than considering long-term safety implications. This was manifested in different ways in these interviews. For example, when reporting sustained IPV (rather than contacting the police following an incident of physical violence, for example), six of the women said that they had been told to leave the house when police arrived at the scene rather than the police acting proactively in relation to the perpetrator. Hannah⁵ stated:

they should have arrested him, whether it was the end of their shift and they didn’t want the paperwork, or if he was too mouthy and they couldn’t be bothered. So, it probably was easier for me to go, but it shouldn’t have to be like that. A woman should never have to leave, he should be taken away.

Lucy illustrates this kind of experience in more detail. Lucy’s ex-partner had previously tried to strangle her, but this incident led to no further action because of lack of evidence. However, his continued abuse and stalking behaviour led her to contact the police again:

I had to call the police because he turned up outside the house again and my kids were inside. The police said they would come and it wasn’t until 5 hours later they came to me. He was up the road in the pub the whole time as well, my friend told me. I just remember sitting in my house, staring out the window terrified, waiting for him to come for me, and I was thinking ‘please come’, to the police, ‘please come and make me feel safe’. (Lucy)

In a further example, Annabelle reported having contacted the police about her experiences of coercive control on so many occasions that ‘she had lost count’. She stated:

there were so many times when they said they were going to pass me on to their domestic abuse team. I would always think ‘this time I am going to be heard’. Then nothing. Just the same cycle over and over again, of just not being listened to.

Finally, five of the women interviewed expressed confusion when their case was classified as ‘standard’ or ‘medium’ risk after completing the DASH risk assessment with officers, when they perceived their risk to be high. For example, Gemma stated:

I never understood when they said I was what they called ‘standard risk’. I mean, what does that actually mean? I clearly didn’t view my risk as standard otherwise I wouldn’t have called the police in the first place, would I?

Laura also had similar feelings stating:

it took me so much courage to call the police, and I did as a last resort. I told them what was happening to me, which I now understand is coercive control, they asked me a load of questions and from this they said my risk wasn’t that high. But most of those questions they asked didn’t actually allow me to say much about my experiences from my own perspective if you get me? How can 20 questions or whatever allow them to determine how safe I am?

Laura’s testimony speaks volumes about the tensions between risk assessment tools and women’s lived experiences of risk and experiences of coercive control.

In summary, the data discussed above supports the view that there is arguably no better assessment of risk than that offered by the victims themselves (Wheeler and Wire 2014). Importantly, it demonstrates the value of this view among women victims of coercive control as articulated in the data presented here. Moreover, risk assessment tools widely use a format of a set number of questions with some space for qualitative exploration. Whilst there are guidelines that recommend and provide space for more such qualitative exploration, in which there is evident space for improvement in terms of practice, the capacity for this to occur is limited (Brennan et al. 2018). The views of the women interviewed here consistently highlighted ongoing tensions between what they perceived as police officers focusing on what they can do ‘here and now’ in stark contrast to their (the women’s) desire for understanding and appreciating the longer-term safety implications of their current position. These tensions return us to the thorny issue of risk inertia and its capacity to capture process: time.

Risk assessment, risk inertia and time

As previously noted, coercive and controlling behaviour is dangerous for victims of IPV (Johnson 2017; Monckton-Smith 2019). However, understanding these dangers through the lens of risk, as widely adopted by criminal justice systems across the globe, is inherently flawed. The findings of this study, in line with other work in this area, highlight that police officers understand risk through an incident and physical violence-focussed approach (the here and now). This informs their understanding of threat for all kinds of violence(s) not just IPV and, consequently, influences the kinds of arrest decisions and case outcomes that may result. Their focus is on what is visible, measurable and actionable. This is arguably particularly problematic when assessing risk in coercive control cases, where measuring patterns of abuse is much more complex and cannot be captured by assessing risk at specific moments in time. The victims/survivors interviewed reported feeling that police officers’ understandings of risk, as reflected in the questions they were asked by them, were problematic and contradicted their own priorities for feeling safe and listened to. In particular, the women felt frustrated that the risk assessment tool used with them (in this case, DASH) did not allow their own perceptions of risk and safety to be heard and fully considered.

As discussed above, risk assessment tools, by their nature, cannot capture the everyday practices women use to negotiate their lives and to keep themselves and their children safe (Shalhoub-Kevorkian 2016). In risk assessment practices, who counts as being ‘at risk’ is significantly influenced by a set number of measurable, individualizing factors, such as the presence of physical violence, children, pregnancy and strangulation, but structural issues, such as gendered social relationships, ethnicity, culture and disability are not fully considered. These tendencies point to the presence of ‘risk inertia’ in the understandings of risk as they have been articulated in risk assessment practices generally and here in the extent to which the women interviewed in this study felt that such conceptions did not reflect the reality of their lives. These tensions point to, at the conceptual level, the extent to which such practices embrace a gender-blind understanding of risk (Mythen 2014). In addition, the inherent tensions between incidentalism and process (Westmarland and Kelly 2016) are present in this data for all to see. There is, however, another issue nestling in the tensions between incidentalism and process. This is the question of time or what has been called ‘time inertia’ (Walklate et al. 2020).

Time inertia has (at least) two dimensions relevant to the discussion here. The first refers to the extent to which concepts and/or policies have become ‘locked in’ over and through time (see also Walklate and Hopkins 2019). The second refers to the ‘escalation thesis’ as articulated by McPhedran and Baker (2012) and discussed above. Much work in this field operates with the presumption that violence in relationships escalates over time or that there is a turning point demanding intervention at a point in time. As a result, this work pays scant regard to those relationships in which violence might have featured but has come to an end and/or those relationships in which the first act of violence might in fact be the lethal one. If these presumptions are taken together, it is possible to envisage the ways in which temporality is smoothed out and even erased from the measurement of risk, its prediction and/or prevention: movement in and over time is absent. Thus, risk inertia, translated into an incident-led linear focus of risk assessment, is compounded by time inertia in which time is taken for granted. All of this has happened and continues to happen in the face of the wide-ranging evidence pointing to the complexities of women’s lives who live with violence as ‘just part of life’ (Genn 1988) and the extent to which experiences of risk in relation to such violence is gendered.

Concluding Thoughts

The observations made in this paper are, of course, limited by the context in which they were generated. This was a study based in one police force focusing on their use and implementation of one particular risk assessment tool and how that tool had been received by the women who were subjected to it. The tool used was the DASH tool, and the College of Policing have attempted to address some of the issues raised here by developing a new DASH risk assessment form, in which victims’ voices are better included in this process. At the time of writing, this tool is still undergoing evaluation and in putting emphasize on harm rather than physical violence per se it is hoped that it will result in improving the risk assessment capture of coercive control. There is also clearly space for improvements in police training on coercive control as intimated in the data discussed above. Indeed, improved training may provide ‘guidance’ to police officers in how to use the new tool and it may well assist them in a better understanding of procedure, though it has yet to be established whether training will result in a fundamental improvement in understandings of the risks associated with coercive control more generally. However, the jury is still out on the extent to which an updated risk assessment tool will effectively shift police officers focus from responding to what is measurable (i.e. incidents of physical violence) to process (i.e. coercive control) through an orientation to harm. It is important to note that the implicit acceptance of a forensic concept of risk remains. This raises ongoing and important questions in relation to any future implementation of coercive control legislation and ways to enable criminal justice professionals to respond in the light of such legislation appropriately.

This paper has also focused attention on one point in time in a victim/survivor’s risk journey. That journey itself is neither simple nor straightforward as work by Hester (2011) has alluded to. Ongoing work is engaged in a detailed examination of this risk journey itself with criminal justice and other agencies to offer a more holistic assessment of the problems and possibilities of risk and risk assessment, including the role of the IDVA (Barlow et al. 2020), all of whom may use differently nuanced risk assessment tools and differently nuanced understandings of risk to inform their decision-making. Importantly the purpose of this paper has not been to engage in a detailed analysis of the efficacy or otherwise of one assessment tool. As was outlined at the beginning, the concern here has been to unpack the conceptual heritage of risk as it informs risk assessment in theory and the translation of risk into practice. Thus, the questions of tool efficacy, and/or appropriate tool implementation notwithstanding, the fundamental problem as articulated here is one of conceptual failure (Lewis and Green 1978). The inherent inability of risk assessment tools to acknowledge and take account of structural, cultural and other differences are additional potential areas for such conceptual failure (see e.g. the case of Ewart vs. Canada as discussed by Hart 2016). There is clearly capacity for more work exploring these issues. However, the focus of this paper has been gender, and the consequences of gender blindness have been expressed in this paper as risk inertia and time inertia. Of course, it is undoubtedly the case that some of the risk assessment practices currently in use will speak to some women’s voices better than others. At the same time, of course, women living with the ‘risks’ associated with IPV and coercive control will and do get on with their lives. These women stand as testimony not only to the prevalence of risk inertia in policy responses to their lives but also their resistances to it as the views expressed by the women interviewed for this project clearly illustrate. Their voices serve to remind us of the importance of understanding processes as they are articulated over and through time, in which women know who is risky and when the risks are getting worse (see Stanko 1990). Listening to their voices might afford a route out of the visions of risk and time reflected in the discussion above and into a way of rethinking the oxymoron of risk assessment and coercive control.

Footnotes

Funding

We are grateful to have received British Academy small grant funding to support this research.

Acknowledgements

We would like to thank Prof. David Gadd and Dr. Kate Fitz-Gibbon for their helpful comments on earlier versions of this paper and to the reviewers of this work. In addition, we would like to thank all delegates who attended our panel at the European Conference on Domestic Violence, Oslo, 2019.

References