Re-tangling the concept of coercive control: A view from the margins and a response to Walby and Towers (2018)

Abstract

This article responds to Walby and Towers’ article, in which they propose a quantitative methodology that evidences gender asymmetry in ‘domestic violence crime’. Through examining core issues including harm, severity and repetition of domestic violence crime victimisation, they argue that Stark’s concept of ‘coercive control’ is obsolete and refute Johnson’s typology of intimate partner violence. However, their conclusions are based on problematic assumptions about, for example, the relative impacts of physical and non-physical violence; the usefulness of incident- rather than relationship-based understandings of domestic violence and abuse and a focus on victim/survivors’ ‘resilience’ and ‘vulnerability’ over perpetrators’ motives. Moreover, their cisnormative operationalisation of sex and gender and neglect of sexuality overlooks important evidence about lesbian, gay, bisexual and/or transgender people’s victimisation. This reinforces a limited ‘public story’ of domestic violence and abuse and arguably creates weaknesses in feminist analyses of domestic violence that could further fuel anti-feminist, gender-neutral approaches.

Introduction

Recently, Sylvia Walby and colleagues outlined a radical new methodology for surveying interpersonal violence (Walby et al., 2016, 2017; Walby and Towers, 2017). Their aims are twofold: to facilitate the production of more valid and robust empirical evidence about experiences and perpetration of what they call interpersonal violence; and to address the ongoing debate about the extent to which gender is implicated within it. They argue that ‘the way forward is to include gender within mainstream statistics and indicators’ (Walby et al., 2017: 3). Most recently, Walby and Towers (2018) propose the concept of domestic violence crime (DVC) as both definition and measure of the behaviours that are, in their view, both the most problematic type of interpersonal violence and the most emblematic of gender asymmetry.

We refer to intimate partner violence (IPV) in this article when we are discussing any acts of violence and/or abuse in intimate adult relationships and from ex-partners. This is because we agree with Myhill (2017) and others that there exist different kinds of IPV that require careful identification in order to best respond to each; and that in order to identify them, motives, impact and the relationship context for the IPV need to be examined. Consequently, we concur with Johnson (2008) and Stark (2007) that coercively controlling behaviours constitute a substantively different kind of violence and abuse than a one-off incident situationally motivated to win an argument or indicate frustration. However, we adopt the term domestic violence and abuse (DVA) to describe the most serious kind of IPV – coercively controlling violence, as Johnson would call it – because this is the term most widely used, including by the Home Office in England and Wales. In addition, while we are aware that in England and Wales, the Home Office definition of DVA includes familial relationships, for brevity – and in line with existing IPV theorisation – we focus only on adult intimate relationships.

In reading Walby and Towers (2018), we have found ourselves in disagreement with much – though not by any means, all – of what they say. We found this surprising because we identify ourselves as feminists and have drawn from the feminist conceptual toolkit during our years of researching IPV in the relationships of lesbians, gay men, bisexual women and men, trans women and men, and gender non-binary people (LGB and/or T+). Our conceptual journey began with agreeing that the feminist focus on power and control as defining characteristics of DVA is crucial in being able to identify those most at risk of escalation, fear, a closing down of ‘space for action’ (Kelly, 2007) and significant physical and mental health impacts. We have understood gender as a core lens through which IPV should be researched in order to make sense of the different experiences, enactments and impacts of IPV, as well as to explore different help-seeking practices (see, for example, Barnes, 2008; Donovan and Hester, 2014). The work of Johnson (2008) and Stark (2012) has underpinned our arguments that identifying the different motives for, and meanings and impacts of, IPV are crucial for the development of best practice in relation to those who are LGB and/or T+.

That the societal context within which IPV takes place is patriarchal is also axiomic. From this follows our agreement that patriarchal influences shape and construct expectations and beliefs about how families, intimacy, gender roles and norms might be enacted and/or experienced in intimate relationships, regardless of identities of gender and sexuality; and that these will be fundamental in understanding experiences and perpetration of IPV and societal responses to it. Our feminism has also been intersectional (Crenshaw, 1994) in its approach, foregrounding an awareness of the different social positioning of social groups that coalesce around ‘race’ and ethnicity, age, disability (physical and learning), faith, social class, nationality/citizenship status as well as gender and sexuality (see Donovan and Hester, 2014).

Walby and Towers’ (2018) key argument is that their concept of DVC captures the harm, severity and gender-specificity or asymmetry of violence (understood as primarily physical and sexual violence) in intimate relationships. For them, this renders unnecessary the concept of coercive control and the need to consider non-physical violence, including non-physical sexual violence, besides harassment. They argue that, rather than focussing on perpetrators’ motives, as Johnson and Stark do, it is more important to recognise the ‘resilience’ (or lack thereof) and/or ‘vulnerability’ of the victim/survivor as measured by their economic and material (i.e. housing) conditions, that can predict escalation in the amount and severity of violence. To respond to Walby and Towers’ argument, our article is similarly structured to address their key issues of contention:

1. Gender

2. Violence and coercion

3. Seriousness and harm

4. Motivation and resilience

Gender

A key concern for us in the debates focussing on gender are the assumptions made by researchers about what is meant by gender or the categories ‘woman’ and/or ‘man’. There is a tendency among some feminist and other researchers in the field of IPV to confuse the separately (albeit overlapping), socially constructed categories of sex, gender and sexuality and use the term ‘gender’ as if it dealt with them all. For example, Walby and Towers (2017; Walby et al., 2018) describe ‘the four gender dimensions’ that should be considered when developing a good quality survey of IPV. These include the sex of the victim, the sex of the perpetrator, ‘the relationship between perpetrator and victim (intimate partner or other family member; acquaintance; or stranger’), and whether or not ‘there is a sexual aspect’ (Walby and Towers, 2017: 13). From our perspective, sexuality is rendered invisible by being subsumed as a ‘gender dimension’. The sexuality of the partners to a relationship is crucial to understanding not just the nature of the IPV being experienced (including whether there is a ‘sexual aspect’) but also their help-seeking and help provision. Without recognition of sexuality, ‘gender’ becomes a proxy for the categories of woman and man, who are too easily assumed to be heterosexual and cisgender. The way in which a problem is defined has important impacts on how experiences are recognised, understood and constituted (Barnes, 2008; Kelly, 1988). This has been a critical feminist argument, for example, in campaigns to criminalise rape in marriage: without explicitly criminalising a husband forcing his wife to have sex, a wife might not problematise the behaviour at all, believing instead that it is within her husband’s conjugal rights.

Elsewhere, Donovan and Hester (2010, 2014) argue that an unintended consequence of the success of feminist scholarship and activism around DVA has been the construction of the public story of DVA. This constructs DVA as a problem of heterosexual (cisgender) men for heterosexual (cisgender) women; a problem primarily of physical violence; and a problem of a particular presentation of gender: a big ‘strong’ (cisgender) man being physically violent towards a small ‘weak’ (cisgender) woman. While this public story reflects the empirical evidence about who is most often victimised by DVA, it nevertheless makes it very difficult for those whose experiences do not fit this story to either tell their story or to be heard. Namely, it excludes not only those who are LGB and/or T+ but also cisgender heterosexual men, and any victim, regardless of sexuality or gender identity, whose experience is not primarily of physical violence. It also omits anyone who has used violence/abusive behaviours themselves in retaliation or in self-defence, or anybody who is physically bigger than their abuser.

Neglecting to be specific in arguments about gender, that is, not specifying that the focus of concern is actually on heterosexual, cisgender women, is theoretically problematic because it implies that the category ‘woman’ (or ‘man’) is stable, fixed and homogeneous, none of which are the case. This is not only because of the existence of trans women and men and of non-binary gender people, but also because feminist analysis tells us that women and men are heterogeneous with intersecting identities across the social groups listed above, thus shaping how IPV is experienced, understood and responded to (Crenshaw, 1994). It is also problematic because this theoretical articulation of gender implies that, in practice, only the victimisation of heterosexual cisgender women is worthy of investigation and intervention. For example, Walby and Towers argue that survey data should be collected from women and men (as we agree), to demonstrate differences in victimisation; and that to address the way the criminal justice system has invisibilised gender, DVC should be mainstreamed within violent crime more broadly, rather than considering it as aetiologically distinct (Walby et al., 2016). They argue that this will make the gendered nature of all violent crime visible so that policy and resources can follow appropriately: ‘[g]ender is woven throughout domestic violent crime, in its scale, distribution, repetition and seriousness, as well as in its associations with access to economic resources through employment and property’ (Walby and Towers, 2018: 18). Yet the data, even that collected using Walby and colleagues’ new methodology, evidences that other groups also experience IPV, including DVC.

Walby and Towers’ analysis of the Crime Survey for England and Wales (CSEW), using their category of DVC, provides a powerful argument that (ostensibly heterosexual, cisgender) women are disproportionately affected by DVC: 74% of DVC victims are women and 82% of DVC is committed against them. However, we do not know the gender or sexuality of the perpetrators, or the sexuality of those victimised. Furthermore, even though the proportions are very low, the numbers of men affected are still substantial (and will include gay, bisexual and/or trans men as well as heterosexual cisgender men) – 79,473 men experiencing 219,118 DVCs. Looking at repeat victimisation, the proportions are similarly stark: 83% of high frequency victims (more than 10 incidents in the last 12 months) are women. Yet, 14,064 men also report high frequency crimes. Considering injury, the proportions are similarly disproportionate with 77% of victims reporting DVC with injury being women and 91% of DVCs with injury being reported by women. Still, this means that 79,549 men report a DVC with injury and 156,441 DVCs resulting in injury are reported by men. Our point is not to challenge the extent of violence against women; it is to emphasise that the analysis does not evidence that only women are victimised and severely injured by DVC, but rather that they constitute the biggest proportion (and numbers) of those victimised. This suggests two things: that a more complex approach to gender that includes an understanding of sexuality is required to make sense of how and why people, other than heterosexual (cisgender) women, are being victimised and that caution is necessary to not privilege a theoretical paradigm over the need for all survivors of IPV to have recourse to appropriate interventions.

We can also look at the proportions of those reporting victimisation in surveys where the sexuality of those victimised is used in the analysis, and this should concern anybody interested in making sense of IPV either methodologically, theoretically, in policy and/or practice. In the most recent Office for National Statistics (ONS) (2018), analysis of which women are most likely to report partner abuse to the CSEW, bisexual women emerge as twice as likely to report (10.9%) partner abuse as heterosexual women (6%), with lesbian/gay women also more likely to report than heterosexual women (8%). Looking at specific types of abuse, bisexual women, for example, are five times more likely to report sexual assault than heterosexual women (1.9% and 0.4%, respectively) with lesbian/gay women also more likely to report than heterosexual women (0.5%). Previously, Donovan and Hester (2014) discussed an analysis by the CSEW of 1000 cases of women and men identifying as LGB (Smith et al., 2010). The reported rates of IPV were more than twice as high for those identifying as lesbian or gay male (13%) as heterosexual/straight people (5%) (cisgender identity was assumed). Furthermore, 12% of lesbian or bisexual women and 6% of gay or bisexual men reported experiencing one or more instances of non-physical abuse, threats or force (excluding sexual assault) in the past 12 months. These proportions are three times and twice as high, respectively, as those reported by heterosexual women (4%) or men (3%) (Smith et al., 2010). Again, the sexuality and/or gender identity of the perpetrator are not reported, making it difficult to do anything other than speculate about why higher rates are reported. Smith et al. (2010) point to age as a factor as 37% of the LGB respondents were aged 16–24 years compared to 21% of the heterosexual respondents and this is the age group most likely to report partner and sexual violence. Donovan and Hester (2014) speculated that the high rates for lesbian or bisexual women might be accounted for by previous male partners (see Ristock, 2011, for similar patterns in Canada and Walters et al., 2013, in the United States). However, it would seem from these analyses that sexuality and not gender predicts the likelihood of reporting victimisation, while reinforcing the importance of collecting data about the perpetrator.

This pattern is repeated in the Fundamental Rights Agency (2014) Violence Against Women: An EU-Wide Survey which asks women participants a question about their sexuality and the sex of their partner. Only 526 participants identified as lesbian, bisexual or ‘other’ (grouped as non-heterosexual in the analysis) so analysis at the individual country level was not possible. Analysis at the European level finds non-heterosexual people reporting proportionately more than double the amount of ‘physical or sexual partner violence’ (48%) than heterosexual women (21%), and nearly double the amount of psychological partner violence’ (70% and 43%, respectively) (Fundamental Rights Agency, 2014: 185). Such dramatically different proportions are surely worthy of further research, suggesting again that it is sexuality and not gender that is the key predictor of reporting IPV.

In their quantitative survey of 1754 young people from across the United Kingdom, Bailey et al. (2018) explore possible links between partner abuse and sexual health. While the numbers of those identifying as having had exclusively same-gender partners (they label these F-F or M-M) or both female (F) and male (M) partners (labelled F-MF and M-FM) was too small to conduct the regression analysis to explore relationships between demographic and sexual health variables, the descriptive statistics tell an interesting story. Higher proportions of F-MF and M-FM report experiencing each of the six types of physical, emotional and sexual violence and abuse asked about than those identifying as F-M, that is, ostensibly heterosexual women; higher proportions of M-M report experiencing five of the six types of violence and abuse asked about than F-M; and F-F reported experiencing more emotional and physical violence and abuse but lower sexual violence and abuse and levels of fear of an ex-/partner than F-M report. The trans participants and those with a trans partner were excluded from the analysis.

In their discussion, Bailey et al. (2018) retain the focus on the numerical fact that young women rather than young men most often report partner violence and abuse, rather than showing any curiosity or concern about their finding that bi/pansexuality predicts the reporting of IPV, not gender. Their stated aim is to discover whether experiencing partner abuse might impact negatively on young people’s sexual health. The correlation is found most strongly with heterosexual young women and less strongly with young heterosexual men. Yet, there is no speculation about the impacts for their bisexual and gay male or lesbian participants. The conclusions do not even call for urgent research to be done with these groups to explore these issues or any methodological speculation about how more young people who are not heterosexual can be recruited into studies such as this so as to improve the likelihood that the same statistical analysis can be done.

Nevertheless, these studies design a methodology identifying those who are not heterosexual which is an improvement, on the whole, on the research methodologies adopted in the heteronormative mainstream field of IPV. Yet it seems that the methodology is used, at worst, to better ‘clean’ the data for an analysis focussed on heterosexual cisgender women and men or, at best, to signal an inclusive research project. The latter aim backfires of course when the subsequent analysis excludes groups because there are not enough to conduct statistical analysis. The end result is the invisibilisation of LGB and/or T people from wider, mainstream discussions of IPV and interventions to respond to this social problem; a similar problem to that which Walby and colleagues (Walby et al., 2017; Walby and Towers, 2018) point to in mainstream approaches to violent crime that render gender invisible, yet one that they repeat in their proposed methodology with respect to sexuality.

Another problematic aspect of a sole focus on gender is the reductionist assumptions made about what gender signifies. Thus, gender is used to signal both binaried sex categories (woman, man) as well as binaried gendered behaviours: women are victimised, men are perpetrators. However, the ways in which gender operates in relationships of IPV have been shown to be more complex than this. In their qualitative analysis of interviews with lesbians, bisexual and queer women, gay and homosexual men, and heterosexual women (all cisgender), Donovan and Hester (2014) identify two ‘relationship rules’ in violent and abusive relationships regardless of sexuality and/or gender identity. The first is that the relationship is for the abusive partner and on their terms, and the second is that the victim/survivor is responsible for the abusive behaviour, the abusive partner, the relationship, the household if they share one and the children if they have them. These rules resonate to some extent with heteronormative constructions of heterosexual relationships that confers on men the role as initiator, decision-maker and relationship authority, while the woman’s role is to be the deferrer, the responder, the one doing the emotional work. Yet, these relationship rules were found in same-sex and heterosexual relationships.

Donovan and Hester (2014) also found that violent and abusive relationships were more complicated than this apparently ‘gendered’ relationship type. They found that while abusive partners set the terms for the relationship, they are also often the most needy of the two partners, eliciting forgiveness, loyalty, protection, special treatment from their partners. Engaging in the latter often left victim/survivors feeling not like the classic ‘ideal victim’; passive, without agency, defenceless (Christie, 1986), but instead as the (emotionally) stronger and responsible partner who manages their abusive partner and the relationship. It would seem useful therefore to consider the consequences for those experiencing IPV of dominant discourses about IPV that confuse sex, gender and sexuality in discussion about what IPV is and rely on narrowly presented binaried assumptions about how gender might be enacted and experienced.

Thus, while we agree with Walby and Towers’ argument that survey data should be collected from women and men to demonstrate gender differences in victimisation, we would also argue that it is as important to establish victim and perpetrator sexuality and the relationship context in order to properly explore the relative importance of these factors on IPV reporting.

Violence and coercion

Walby and Towers argue that violence is always coercive even, as per Gramsci and Weber, if its use is only occasional, and that the law recognises this by default. They therefore dispute Johnson’s (2008) perspective that not all IPV is coercive and Stark’s (2007) emphasis on the non-physical rather than physical forms being most coercive because of their relationship to gender inequality. We argue that this incorrectly conflates violence with the exercise of power over another. The argument that all violence is coercive helps to underpin two wider points that Walby and Towers make: DVC should not be an aetiologically separate field of study to violent crime, and there is no need for separate concepts such as coercive control which, in their view, falsely exaggerate the distinctiveness of DVC from general violent crime.

Yet, to state that all violence is coercive bypasses a huge degree of nuance, conflating violence with a coercive intent and violence in self-defence, anticipation or retaliation. It also fails to distinguish between the impacts of being in an intimate relationship with someone who one knows will readily use violence against them versus a single incident of stranger violence. Furthermore, while it can perhaps be argued that all violence is coercive insofar as it seeks to sway the outcome or course of a situation (even if that is to incapacitate a primary perpetrator), this seems rather questionable when one’s violence is intertwined with a partner’s use of violence. The antecedents to the use of violence are not effectively captured in quantitative research, resulting in a very partial picture of how and why violence is used in intimate relationships, and to what effect.

For example, Myhill (2017) points to the risk that a primary perpetrator could be counted as a victim of DVC if they were to report any physical violence from their partner either to the police or in a victimisation survey. He states that best practice and appropriate interventions rely on being able to understand the context for physical violence because most of the victims presenting at domestic violence services are experiencing coercive control (see also Leone et al. (2007) findings that (heterosexual, cisgender) women at greatest risk and/or fear are most likely to seek help). Establishing the extent of coercive control in those reporting IPV would, Myhill argues, provide important intelligence to use for provision planning and resource requirements. This evidence also has relevance for Walby and Towers’ argument about violence and harm insofar as perceptions of risk might also include perceptions of harm and be implicated in decisions about help-seeking. Similarly, Hester (2013) has shown in a mixed-methods study of police arrests of IPV perpetrators in one English police force that women arrested were more likely than the men to have used weapons, arguably because women were not confident about defending themselves without a weapon. In general, however, women were also twice as likely to be arrested as men. Yet, in Walby and Towers’ (2018) analysis, use of a weapon is counted as among the most serious crimes which could act to distort understanding about women’s violence against men as it suggests they are more likely to enact the most serious crimes. At the same time, their approach seems to undermine feminist exhortations to move away from an incident-based to a relationship context approach to understanding IPV (e.g. Stark, 2007).

Moreover, the presentation of physical violence as the only authentic way of coercing a partner is problematic. Identity abuse provides a good example of (most often) non-physical violence by which abusive partners can exert coercive power and control, which would not be recognised in Walby and Towers’ proposed methodology. Identity abuse is an aspect of DVA that is seldom explicitly researched in studies of cisgender, heterosexual women, although it has implicitly been evidenced in how abusive partners undermine, belittle and punish ex/partners, drawing on cisgender, heteronormative assumptions as justification. For example, Dobash and Dobash (1998) and Stark (2007) evidenced abusive cisgender, heterosexual men controlling the presentation of femininity that the women they victimise are allowed: their hair style, use of make-up, clothing as well as their behaviours.

Similarly, identity abuse on the grounds of LGB and/or T+ identities illustrates the ways in which patriarchal systems are deeply heteronormative and provide opportunities for coercive control. Abusive partners coercively control by, for example, threatening to out their partners, undermining their partner’s sense of self as, for example, a ‘real’ lesbian or trans man, denigrating local LGBT scenes or venues to isolate their partner and/or drawing attention to parts of a partner’s body in ways that are demeaning or run counter to their gender identity (e.g. Donovan et al., 2014; Donovan and Hester, 2014). These can all only be effective in a society that does not fully accept the equal social status of those who are LGB and/or T+. The fact that identity abuse can be particularly effective with those only recently out into their sexuality and/or gender identity, with little experience of life and intimate relationships as an LGB and/or T person points to the situated vulnerability of those who are in their first relationship as an LGB and/or T person (Donovan and Hester, 2008, 2014; Ristock, 2002). Donovan et al. (2014) speak about the use of ‘experiential power’, where abusive partners who are already out, regardless of their age in relation to their partner, use their prior knowledge of the local scene and local subcultures of ‘gay’ life to coercively control them (Donovan et al., 2014). Accounting for these kinds of behaviours must be a part of how we recognise and identify IPV and while we agree with Myhill (2017) that discussions should take place about how to operationalise empirically robust measures of these types of behaviours, we have a broader constituency of victims in mind.

Walby and Towers (2018) also argue that it is better to measure each violent incident than a single course of conduct. We agree that the re-analysis of CSEW data to lift the cap on the maximum number of incidents that can be counted is a critical development (Walby et al., 2016). However, we are concerned about a blurring of what is expedient methodologically with how victims/survivors understand their lived experience. Our research leads us to argue that restricting DVA only to those acts defined as DVC provides a partial picture of the experiences reported by survivors (Donovan and Hester, 2014). Coercive control, a concept arising from qualitative research with victim/survivors (e.g. Walklate et al., 2018), captures the range and/or pattern of behaviours – not only or necessarily ever, physical violence – that cumulatively result in one partner exerting power and control over the other (Stark, 2007). A focus on DVC appears to be a reductionist solution, separating and prioritising material impacts of physical violence over the more holistic understanding of the ways that coercive control can, cumulatively, exert profound emotional and social impacts.

A focus on DVC might provide a methodology that makes it easier to count behaviours, but the data are unstable as a measure either of the prevalence of what an individual might be experiencing in their abusive relationship or of the most serious IPV (see below). In addition, unless there is an assessment of how any (repeated) incident ‘fits’ within the relationship as a whole, the wrong person could be identified as a perpetrator and the victim/survivor’s needs will go unrecognised.

Seriousness and harm

Walby and Towers argue that physical violence is the most physically harmful, and therefore the most serious, type of violence (including stalking), as per the hierarchy of harms designated by criminal law, for example, in sentencing guidelines. This contradicts Johnson’s distinction between physical violence in the context of situational couple violence being (typically) less serious than physical violence in the context of coercive controlling violence (Johnson, 2017).

It is an astounding oversight to read harm only in physical ways. Physical injuries might be more measurable, yet research with (heterosexual cisgender women) survivors foreground the impacts of non-physical violence (Kirkwood, 1993; Wilcox, 2006; Williamson, 2010) including impeding their help-seeking. We accept that problems with the data are not straightforward and agree with Walby et al.’s (2017) argument that surveys relying on the Conflict Tactics Scale (CTS) are deeply problematic for their lack of context; limited attention to impact; and, ironically, their approach to counting incidents of IPV. We agree that in order to construct a profile of IPV across a population, it is important to get a sense of how much and what range of IPV individuals have experienced as well as impacts (see Donovan and Hester, 2014; Hester et al., 2010). Our methodological approach values listening to the voices of those most marginalised in society in order to make sense of the legacies of a society based on hierarchies of gender, sexuality, ‘race’ and ethnicity, faith, age, physical and mental capacity and the legitimising of economic inequalities. It is important to (re-)state the critical contribution of qualitative research to the field of IPV in its own right and also in informing the design, the analysis and interpretation of quantitative data.

While Walby and Towers (2018) develop a robust argument against considering typologies of IPV reflecting the motivations of perpetrators, they nevertheless create their own hierarchical typology of IPV. Their typology makes cumulative distinctions between those who have experienced physical violence that reaches the threshold of an existing crime and those who have not; those whose experiences of physical violence have resulted in any physical injury or not; those who have experienced threats of physical violence that have frightened them or not; and repeat victimisation of these behaviours, in the previous year or ever. These criteria construct what is for them the worst type of IPV, DVC, while side-lining other experiences as ‘not domestic violence crime’.

Walby and Towers (2018) argue that the legal definition of violence captures intent and harm: if there is no harm, there is no crime. They reject Johnson’s (2008) and Stark’s (2007) conceptualisations of the relationship between harm and violence as dependent on particular relationship dynamics. They are also (rightly) critical of the CTS for only considering the act, devoid of the contextual detail of harm to the victim, which they say falsely reveals gender symmetry.

However, what this overlooks are other contextual details such as the meanings of violence. Violence means something different when it is perpetrated by someone whom one loves, trusts, feels compassion or responsibility towards. Violence also means something different, and exerts different, less visible, harms, when it is perpetrated by an intimate partner, in one’s safe spaces, in the presence of/proximity to one’s children. The meanings of violence are also mediated by social identities, biographies and individuals’ resources (Anderson, 2009), resonating with Walby and Towers’ discussion of resilience and vulnerability. But, what victims/survivors decide to report – either in a survey or to an agency or informal source of support – is determined by a much more complex array of factors than simply whether a violent act has taken place and whether it has caused harm. To fall back on how the criminal justice system handles harm is deeply problematic. Little more than a cursory glance at attrition rates, sentencing data and conviction rates is required to know that victims/survivors receive neither equal recognition of their victimisation nor equal acknowledgement of the harms inflicted upon them.

In our research, we see this as a consequence of the public story of DVA where on one hand, only women are victims and only men, perpetrators; and on the other hand, in a relationship between two women or two men, any violence must be equal, or mutual, because they are assumed to be equally matched; and the violence will not be very serious either because they are women and women are not violent, or because they are men and used to fighting and defending themselves. Not only can practitioners make these assumptions (see Ristock, 2002), but also researchers (see Johnson, 1995). Several studies have been conducted in the United States with different groups of practitioners who perceive risk and safety differently depending on the sexuality and/or gender of the victim/survivors and perpetrators. How seriously violence between women is perceived (by psychology students) is influenced by whether the victim/survivor is understood to be ‘masculine’ or ‘feminine’ (Little and Terrance, 2010); the violence of men towards men is recognised as potentially serious by police, but not as serious as the violence of men towards women (Pattavina et al., 2007); the risks of escalation are not perceived by crisis (refuge) centre workers as being as high for survivors in same-sex DVA scenarios as in opposite-sex scenarios (Brown and Groscup, 2009). This evidence can be seen to challenge the premise of Walby and Towers’ (2018) argument that harm, as measured by crime, is somehow factually established. On the contrary, this evidence suggests that perceived harm is subjective and socially constructed, being judged by many factors including the sexuality and gender of the protagonists, and not only the type of physical violence used.

Motivation and resilience

Walby and Towers (2018) ask, ‘are variations in the extent and seriousness of domestic violence best explained by the motivation of the offender or the resilience of the victim?’ (p. 14). They use the arguably individualistic terms ‘resilience’ and ‘vulnerability’ when actually they mean (cisgender, heterosexual) women with limited access to economic and material resource. This is because Walby et al. (2016) demonstrate, by removing the cap of five incidents of violence in reports from the CSEW, that DVC victimisation has been increasing in the CSEW since the 2008 economic crisis.

However, it seems odd, first, to posit perpetrator motivations versus victim resilience as an either/or in analysis of IPV. Without an understanding of what motivates perpetrators, it is not possible to develop appropriate perpetrator interventions, plus motives are important in criminal justice processes, for example, in determining whether a homicide is manslaughter or murder and in ascertaining whether an act of violence was self-defensive. Second, Walby and Towers draw attention to vital economic and material factors that shape experiences of leaving and help-seeking with respect to IPV. Yet, to suggest that only economic factors prevent victims/survivors from leaving abusive partners is far too limiting, and again glosses over quantitative and qualitative evidence demonstrating that a much wider range of factors is at play (Donovan and Hester, 2014). Third, survivors’ so-called ‘resilience’ cannot be separated from perpetrator motives, since the latter can include reducing their partner’s financial independence and shrinking their social networks as well as diminishing their confidence, such that their options for leaving the relationship are restricted. Without understanding the dynamics of coercive control, however, it is difficult to understand how this ‘entrapment’ (Stark, 2007) comes about. Even if unintentional, the language of resilience and vulnerability risks conjuring up neoliberal ideas about the responsibilised citizen (Rose, 2000), responsible not only for their own success but also their own failures, thereby putting the onus on survivors to transform their material circumstances.

It is also unexpected to find Walby and Towers promoting a focus solely on survivors and not including perpetrators because this seems out of step with current feminist and policy thinking that has called for increased perpetrator accountability (Home Office, 2016; Westmarland and Kelly, 2012). Considering the motivations and intent of perpetrators is not to imply that some forms of IPV are more or less acceptable: all IPV is concerning and should be addressed appropriately to ensure that it stops. However, making a judgement about what violence and/or abuse has happened and why, with what intention and in what context is fundamental to ensuring that any response is reasonable, just, proportional, informed, likely to get buy-in from the partners to the relationship and maximises safety and well-being. It is our belief that in order to achieve this, practitioners should be confident and have expertise to draw on that enables them to ascertain what IPV they are being approached for help with and what interventions will be most effective.

Walby and Towers critique (rightly, we would argue) the impression that typologies such as Johnson’s give that different types of violence are discrete and stable, thus neglecting the potential for movement within types. They dispute the idea that there are different types, arguing instead that what is being observed is escalation. Thus, all violence can become what Johnson describes as coercively controlling violence if, due to lower ‘resilience’, the victim/survivor does not have the resources to leave the abusive relationship prior to the frequency and severity of the violence escalating.

Echoing the previous discussion, it is too limiting to suggest that escalation only occurs because the victim does not have the financial means to leave the abusive relationship. For example, arguably at a time of their greatest agency, when they decide they will leave, survivors can face an escalation in violence and abuse (Campbell et al., 2009). The assumption is also made that all victims/survivors want to leave the abusive relationship as soon as violence becomes evident, and yet again this is not what the qualitative evidence tells us (Donovan and Hester, 2014). There is much more complexity in decisions to leave abusive partners, as evidenced in both the research with heterosexual women and research on LGB and/or T people’s experiences of IPV. The impact of love here is also important as it is often given as a reason why survivors remain or return to abusive relationships (Donovan and Hester, 2014).

However, rejecting the idea of typologies in favour of a focus on DVC is a retrograde step. It also seems to contradict Walby and Towers’ own critique of the risk assessment tools used by most police authorities in England and Wales to categorise risk. They argue that such tools can result in not providing interventions to those at lowest risk with the danger that the potential for escalation is missed. Risk assessment scales, like Johnson’s typologies, they argue, assume stable categories (even though Johnson (2017) does not say that the violence identified as situational couple violence could not escalate, only that it is less likely to than coercively controlling violence). In their proposal, however, those deemed not to have experienced DVC are not assumed to have any risk of escalation to physical violence, even though they might have experienced identity abuse, isolation from friends and/or family, financial abuse and/or any forms of degradation, none of which would be counted in Walby and Towers’ (2018) proposed methodology.

Conclusion

This article sets out a response to the claims made by Walby and Towers (2018) in justification of their proposed new methodology to measure and evidence gender asymmetry in DVC. Our response challenges the heteronormative and cisnormative ways in which their analysis privileges a particular operationalisation of gender in order to achieve their goal of designing a methodology that will prove gender asymmetry. Their proposed methodology would reinforce the public story of DVA which we argue would be a retrograde step, privileging physical violence in the relationships of ostensibly cisgender heterosexual people. We have problematised their apparent trust in criminal justice system definitions and categories of crime as a basis for their measurements.

These concerns notwithstanding, a focus on the category ‘woman’ is understandable and defensible, particularly in a context of austerity when there is increasing competition for reduced funds to provide services. While heterosexual cisgender women may not be proportionately the group most likely to report DVA in victimisation surveys, they are numerically, and thus it is women who most often appear in services (e.g. see Myhill, 2017). Again, it is important to remember that it is not clear what sexuality the women are because most police data collection systems do not report on the sexuality and/or relationship type of the victim/survivor reporting. The Westminster Government have insisted that their support for IPV service provision, including women-only services, remains constant, but insist that, ultimately, decisions about local services are local government decisions (Donovan and Durey, 2018). Yet, the evidence mounts that funding for the domestic violence sector and for women-only services particularly is under threat (Davidge and Magnusson, 2018; Towers and Walby, 2012).

We have demonstrated that those who are not heterosexual or cisgender are rendered invisible in research even when the proportions of them reporting of IPV is demonstrably higher than heterosexual women. We suggest that in the quest to address the very real threats to the gains made by feminist activism, scholarship and practice in the last 40 years or so by encroaching gender-neutral, anti-feminist arguments being made about the nature of IPV, the experiences of LGB and/or T+ (and other) victims/survivors are being indefensibly negated in spite of their very real experiences of harm. We would therefore urge the mainstream IPV research community to become more inclusive, theoretically and methodologically, in order to stop reproducing both the invisibility of LGB and/or T+ victims/survivors and to stop perpetuating the heterosexual, cisgender assumption. This is vital if the overlapping risks of gender and sexuality are to be identified in order to better make sense of IPV as well as to support adequate interventions for all survivors. This is not a plea for simple inclusion but, more broadly, an argument against a methodology, seemingly driven by ideological concerns, that is in danger of rendering invisible many of the lived experiences of all victim/survivors across different sexualities and gender identities.