Abstract
This research investigated cases of femicide in Great Britain where perpetrators claimed that victims died in “sex games gone wrong.” Forty-three femicides that resulted in a conviction for murder, manslaughter, or culpable homicide were examined in relation to victim–perpetrator sociodemographic characteristics, victim–perpetrator relationship, homicide details, and criminal justice outcomes. Key patterns within the data were identified. The research highlights that the “sex game gone wrong” narrative has gained traction against a cultural backdrop of normalized bondage, domination, and sadomasochism (BDSM) within neoliberal political economy.
Introduction
In December 2016, John Broadhurst killed his girlfriend Natalie Connolly. Natalie’s body was found in the hallway of the home they shared in Worcestershire, England. Natalie had a blood alcohol level of 389 mg per 100 mL—5 times the drink drive limit (R v. Broadhurst, 2018). She had sustained more than 40 separate injuries, the majority of which had been caused by Broadhurst striking her with his hand and a boot and inserting a bottle of spray carpet cleaner—a large container with a trigger attachment—into her vagina (R v. Broadhurst, 2018). This latter act caused severe lacerations which resulted in arterial and venous hemorrhage (R v. Broadhurst, 2018). Broadhurst claimed that she had died in a sex game gone wrong, that they regularly engaged in sadomasochistic sex and that she had consented to the acts that preceded her death. At his trial for her murder in 2018, Broadhurst’s guilty plea to the lesser charge of manslaughter was accepted. He was sentenced to 3 years and 8 months in prison. The outcome caused outrage among those campaigning to end violence against women. Members of the public wrote to the Attorney General’s Office, challenging the sentence but despite this, Broadhurst’s conviction and sentence were upheld (BBC News, 2019). Broadhurst later appealed unsuccessfully against his conviction.
The Natalie Connolly case drew national and international attention to claims of a “sex game gone wrong” (hereafter referred to as SGGW) in cases where men stood accused of killing women. Defendants attempted to claim consent, stating that the women they killed had wanted to engage in “rough sex” and, in many cases, that they had initiated the activities that led to their deaths. Such cases are prominent in mainstream media and popular cultural discussion, given their embodiment of sex and violence news values (Jewkes, 2015). However, there is little academic research into this phenomenon. This article intends to fill this gap by identifying and analyzing femicides in Great Britain where perpetrators claimed that women’s deaths were the unintended outcomes of alleged SGGWs. Such empirical research is necessary to explore the nature and scale of these cases and highlight unique and interesting features of the data. The article begins by situating SGGW in the wider literature around femicide, before describing the approach taken in gathering data about British cases. The findings are then considered in light of the existing literature, outlining the key points to emerge from this research and identifying avenues for future inquiry.
Femicide and the “SGGW”
In the year ending March 2019, there were 241 female victims of homicide in England and Wales (Office for National Statistics [ONS], 2020). In cases where the suspect was acquainted with the victim (N = 135), 59.3% (N = 80) had been killed by partners or ex-partners (ONS, 2020). Within the period year ending March 2001 to year ending March 2019—the period of official statistics of closest correspondence to the 18-year timeframe covered by this research—the total number of female homicide victims who were acquainted with the suspect was 2,682 (ONS, 2013, 2020). Of these female victims, 60.1% (N = 1,611) had been killed by a partner or ex-partner (ONS, 2013, 2020). By contrast, male victims of homicide are highly unlikely to be killed by their female intimate partners. The figure for year ending March 2019 stood at just 8.1% (N = 16) in cases where victims were acquainted with suspects (N = 197). For the period year ending March 2001 to year ending March 2019, of the 3,948 male victims who were acquainted with the suspect, only 9.1% (N = 358) were killed by a partner or ex-partner (ONS, 2013, 2020). As such, when examining cases of victims killed by partners or ex-partners over the 18-year period of official statistics, the proportion of women killed by this suspect group was more than 6 times that for men. It is not only the victim–perpetrator relationship that is gendered; so, too, is the homicide method. While a sharp instrument was the most common method of homicides involving male (50.1%, N = 198) and female (29.6%, N = 61) victims in England and Wales in the year ending March 2019, the second most common method among female victims was strangulation/asphyxiation (ONS, 2020). Of the 206 female victims for which the homicide method was known in year ending March 2019, just below one fifth (19.9%, N = 41) were strangled or asphyxiated, compared with less than one in 20 (3.8%, N = 15) male victims (ONS, 2020). When considering all victims who died from strangulation or asphyxiation in the year ending March 2019 (N = 56), nearly three quarters (73.2%, N = 41) were women and just more than one quarter (26.8%, N = 15) were men. Despite the fact that men are generally overrepresented as homicide victims, making up 63.9% (N = 429) of victims killed in the year ending March 2019, women were overrepresented as victims of homicide in which the method was recorded as strangulation/asphyxiation. As such, domestic homicide in general—and homicide by strangulation/asphyxiation in particular—is inherently gendered.
This research defines femicide as the killing of women by men. The explicit use of the term “femicide” is in response to the tendency of terms such homicide, murder, and manslaughter to deny the gendered nature of these women’s deaths. Femicide is distinct in drawing attention to female victimization and encouraging a focus on the broader social, cultural, political, and economic forces in which misogyny thrives and violence against women is denied and minimized (Ingala-Smith, 2018). Current interpretations of femicide are heavily gendered. Monckton-Smith (2019) notes the considerable force of the “crime of passion” discourse, which emphasizes traditional gender roles, norms of heterosexual romantic love, and differences in expectations, responsibilities and behaviors of men and women. In this frame of reference, femicide perpetrators “snap” and the violence comes “out of the blue” in a moment of madness, evoking terms such as “red mist,” which serve to deny their responsibility (Adams, 2007; Dobash & Dobash, 2009; Monckton-Smith, 2012; Monckton-Smith et al., 2014). This approach interprets femicides as spontaneous incidents, in which an otherwise “normal” man has been compelled to kill by forces beyond his control. Perpetrator past histories of abuse of women, common in femicide, are absent from the crime of passion discourse (Campbell et al., 2007; Dobash & Dobash, 2015; Polk, 1994; Stark, 2007). There is a focus on the victim’s behavior—their alleged nagging, infidelity, and/or failure to live up to expectations of who women are and how they should behave (Lees, 1997; Monckton-Smith, 2012). The provocation—or “nagging and shagging”—defense was often a successful one for men accused of killing their partner, resulting in convictions for manslaughter as opposed to murder (HC Deb, 2019). The provocation defense, however, was replaced with a “loss of self-control” defense in 2009 under the Coroners and Justice Act, which removed infidelity as an accepted factor in provocation for violence (Horder & Fitz-Gibbon, 2015). The focus on victim behavior means that those killed by their current or former partners are accorded the lowest status in a victim hierarchy where levels of sympathy are contingent upon the nature of the woman’s relationship with the perpetrator (Greer, 2007; Monckton-Smith et al., 2014). The “popular imagery” (Dobash & Dobash, 2015, p. 123) of sexual murder, where a woman is attacked by a stranger, often late at night in a dark area, causes outrage. The tendency of mainstream media to sympathetically cover “ideal victims” (Christie, 1986; Van Wijk, 2013) means that women killed by men with some degree of legitimate access to them—particularly relevant in alleged cases of SGGW—are increasingly held accountable for their own deaths. This victim blaming draws upon neoliberal tropes of the sovereign individual, responsibilized to protect themselves from harm (Bumiller, 2008; Tudor, 2018).
The assumption of the sovereign individual obscures the fact that men’s legitimate access to women is sometimes secured through abusive and manipulative courses of conduct. Particularly prominent within such behavioral patterns is coercive control, a liberty crime (Stark, 2007) in which a perpetrator gradually strips a victim of their autonomy until “who and what she is, is who and what she is for him” (Stark, 2007, p. 262, emphasis in original). Perpetrators engage in acts of everyday terrorism (Pain, 2014) intended to deprive victims of their liberty. These acts include behaviors relating to isolation and deprivation, destruction of self-worth, threats and acts of abuse, and monitoring and surveillance. Examples include dictating what a victim can wear; where they can go and whom they can see; directly and indirectly putting them down; telling them they are worthless; enforcing ever-changing rules that humiliate and degrade them; threatening to harm or kill the victim, their children, and/or pets; and placing tracking devices on the victim’s vehicle or property. Perpetrators target specific vulnerabilities, exploiting a victim’s insecurities and anxieties. Coercive control is highly personalized but factors such as a victim’s younger age and lesser social status can be significant in establishing a foundation upon which to accumulate power and control (Stark, 2007). The gradual introduction of these behaviors over time slowly chips away at the victim’s sense of personhood. It effectively removes women’s capacity to self-safeguard, challenging the assumption that all women are equally free to exit an intimate relationship on which common tropes like “why doesn’t she just leave?” are based. Despite the considerable academic evidence around coercive control and its criminalization in England and Wales in 2015, cultural and social reactions to women killed by men continue to place an emphasis not upon the choices of the perpetrator to engage in such conduct but upon the behavior of the victim in failing to prevent her own homicide.
Through a lens of coercive control, femicide is not a loss of control but an attempt to maintain control (Monckton-Smith, 2019; Stark, 2007). If the perpetrator perceives that the victim has withdrawn her commitment to him via a decision to leave, a relationship with another man, or some other perceived disloyalty or rejection, he will “change the project” (Dobash & Dobash, 2015, p. 39) from controlling her through keeping her in the “relationship” to destroying her for leaving it. The perpetrator’s feelings of entitlement to own and possess the victim preclude her assertion of independence and he begins planning, creating opportunities for the killing to happen— e.g., obtaining weapons, carrying out searches for information online, organizing finances, stalking the victim, and beginning to shape the narrative around the planned homicide (Monckton-Smith, 2019). This planning is evident in cases where the perpetrator later claims that the victim’s death was an SGGW. Monckton-Smith (2019) describes one such case: In one of the cases a woman’s body was dressed after death in “sexy” clothes to make it appear as if she had consented to the strangulation she died from. There was also evidence of forced ingestion of illicit drugs with claims of suicide or misadventure. (p. 14)
Drawing on knowledge of the victim’s habits, routines, strengths and needs—within a cultural context of victim responsibilization—abusers are able construct a compelling SGGW story. The silenced victim is presented as a consenting, free-willed, and rational party, and the perpetrator’s version of events becomes the dominant one (Busby, 2012), obscuring the context of the femicide. Harriet Harman, MP, recently argued this point in a House of Commons debate on the proposed Domestic Abuse Bill for England and Wales: She, of course, is not there to say otherwise. In the witness box, he gives lurid, unchallengeable accounts of her addiction to violent sex, and explains that the bruises that cover her body were what she wanted. The grieving relatives have to listen to his version of her sexual proclivities and see them splashed all over social media and the newspapers. He has killed her, and then he defines her. She is dead, so only he gets to tell the story. (HC Deb, 2019)
The SGGW appears to be emerging as a contemporary variant of the crime of passion discourse, a new version of the old provocation defense which distracts from the reality of coercive control. Rather than presenting the homicide as a heteronormative “natural” male response to a woman’s failure to conform to gendered expectations, it becomes an outcome of the perpetrator’s compliance with the victim’s demands for “rough sex,” which in turn is interpreted as her expression of women’s sexual liberation. The normalization of bondage, domination, and sadomasochism (BDSM) via mainstream cultural artifacts such as Fifty Shades of Gray have reshaped patriarchal narratives of women’s subjugation under the guise of empowerment and sexual agency, reinforcing traditional gendered assumptions (Busby, 2012; Carter, 1979; Downing, 2013; Moore & Khan, 2019; Weiss, 2006). Moore and Khan (2019) argue that pornography and women’s magazines have played a significant role in the mainstreaming of BDSM. The portrayal of the Fifty Shades of Gray male lead character’s sexual conduct as “safe, sane and consensual”—the widely accepted BDSM credo—obscures the “everyday terrorism” (Pain, 2014) evident in his relationship with the lead female character, a focus on the prurient serving to minimize the coercive and controlling elements of his behavior (Downing, 2013). In cases of alleged SGGWs, the mainstreaming of BDSM ensures that the emphasis remains firmly upon the female victim’s agency and behavior rather than the male perpetrator’s decisions and actions.
Perhaps the best-known case in which consent was proposed as a defense for physical harm in the context of sexual activity was the infamous “Spanner Trials,” named after the police investigation into a group of homosexual men who had engaged in sadomasochistic sex acts resulting in physical injuries. In December 1990, 16 men were convicted of offenses including assault occasioning actual bodily harm. The men unsuccessfully appealed their convictions, the Law Lords ruling that sadomasochistic acts are not an exception to the rule that consent cannot be used as a defense against charges relating to bodily harm (Edwards, 2017). However, despite this legal precedent, consent to physical harm is increasingly being presented by defendants under the guise of alleged SGGWs in criminal proceedings relating to femicides (Edwards, 2017).
The SGGW argument assumes a level playing field of power in a consensual, healthy relationship, holding women accountable for the harms they experience (Edwards, 2017). This makes it challenging to see or respond appropriately to cases where the perpetrator has engaged in abusive or controlling behavior toward the victim, removing her liberty and independence to such an extent that the playing field is anything but level. One act that has been appropriated by abusers in the context of alleged SGGWs is strangulation, which has been subsumed within a perpetrator-led BDSM narrative and is not treated with the gravity it deserves in criminal proceedings (Busby, 2012; Edwards, 2015). This is despite a lack of evidence that it heightens women’s sexual gratification, expert opinion that there is no safe way to practice this act (Edwards, 2017), and that the risk of homicide increases significantly for women who have experienced non-fatal strangulation at the hands of their abusers (Glass et al., 2008; Edwards, 2015). Busby (2012) argues that strangulation, suffocation, and choking are not synonymous but conceptual confusion has diluted the unique and serious risks strangulation poses—the judiciary using “choking” while journalists write about “erotic asphyxiation.” It is also worth noting that official homicide statistics from the Office for National Statistics in England and Wales quoted earlier (ONS, 2020) conflate strangulation and asphyxiation within the method category termed “strangulation/asphyxiation.” Strangulation has been criminalized in Australia, Canada, New Zealand, and several U.S. states, and the act is unlawful within England and Wales (s21 Offences Against the Person Act) but is rarely cited on indictments in this jurisdiction (Edwards, 2017). So too, the 2008 criminalization of extreme pornography depicting sexual violence and strangulation has not resulted in significant numbers of prosecutions for those possessing such imagery. The law has instead been applied to possession of child sexual abuse imagery or bestiality (Dymock, 2018; Edwards, 2017). In addition, strangulation is not listed as a specific aggravating factor in sentencing for murder in England and Wales (Edwards, 2015). Judges may consider it within broader aggravating factors such as “seriousness” or “sadistic conduct” but a lack of knowledge about its gravity often prevents it being considered in this way. This becomes particularly problematic in manslaughter cases, where judges have considerably more discretion around sentencing.
Judicial ignorance around BDSM is compounded by a reluctance to call evidence from appropriate experts—for example, with regard to the “safe, sane, and consensual” credo among BDSM practitioners (Busby, 2012). This in turn leads to a heavy emphasis on a perpetrator’s version of events. The lack of scrutiny of their claims about the victim’s sexual behavior runs the risk of painting a misleading picture for juries, particularly when similar evidence about their prior (and often serial) abuse of women is not deemed admissible (Busby, 2012). This prevents juries from being able to differentiate abuse from consensual activity. This has significant implications in relation to understanding consent; as Busby explains, “unconsciousness is not the desired result of erotic asphyxiation. Rather it is the light headedness that comes when one is taken to the point of unconsciousness but not past it” (2012, p. 339, emphasis in original). Abusers do not see their victims as autonomous individuals to please via sexual gratification. They seek power and control. Rendering them unconscious is a reliable means of achieving this control, especially when their version of events is not sufficiently scrutinized in court and their histories of coercive and controlling behavior are excluded from evidence. Edwards (2017) cites the 1994 case of Williamson, where the defendant was convicted of the manslaughter of his female partner. Williamson had claimed that consensual asphyxiation was part of their usual pseudo-masochistic sexual activity. An appeal court heard that the defendant and the deceased had been “deeply attached to each other” (Edwards, 2017, p. 90), evoking the narratives of romantic love commonly associated with the crime of passion discourse (Monckton-Smith, 2019). His sentence was reduced from 4 to 3 years. After his release, Williamson went on to abuse other women and kill his mother. It emerged that he had multiple previous convictions for violence. When considered within the coercive control discourse, “sex games” in this context are exposed for what they really are—attempts to manipulate trial and sentencing outcomes (Edwards, 2017)—and the sex acts within the homicides are revealed as an abuser’s way of humiliating and possessing the victim (Dobash & Dobash, 2015).
The literature establishes that women are disproportionately likely to be killed by male partners and are disproportionately likely to be killed by strangulation. The legitimate access that men have to female partners accords femicide victims a low status. Following the removal of the old provocation defense, a new one has emerged in its place—the SGGW, which is reinforced by four key social and cultural factors. First is the continued dominance of the crime of passion discourse, which interprets acts of fatal violence as expressions of romantic love and precludes acknowledgment of coercive control as a contextual factor, diverting attention away from the planned nature of femicide. Second is the fallout from formal sex equality, which carries the assumptions that all women are equally free to choose to remain in an intimate relationship and that all women are equally able to choose and consent to specific types of sexual activity with men. Such assumptions facilitate the denial of coercive control and responsibilize women for men’s behavior. Third, linked to formal sex equality, the cultural normalization of BDSM has reframed acts such as non-fatal strangulation as “play,” obscuring its strong association with domestic abuse and homicide. Finally, forensic misunderstandings of BDSM and the failure to call expert evidence in court mean that BDSM narratives can be easily called upon in defenses. In turn, these four factors enable perpetrators to maintain their monopoly on the narrative of femicide.
Despite these insights from the literature, little is known about the scale and nature of femicides presented as SGGWs. In addition, the existing literature tends to focus upon cases where victim and perpetrator were in—or had been in—a substantive intimate relationship. However, cases were emerging in the mainstream media of the SGGW defense being used in cases that did not feature such relationships—for example, where the victim and perpetrator had just met or were on a first date. As such, there was a need for empirical research into all femicides in which the SGGW defense had been used. The following section describes the approach to collecting and analyzing data about such cases in the British context.
Approach to the Research
It was decided that the focus for this specific research would be cases where a conviction for a homicide offense had been secured. Such cases represented an important starting point for enquiry into SGGWs given the likelihood of more information being available about these cases than cases where there had been no prosecution or conviction. In addition, the cases provided an opportunity to explore instances where the SGGW defense had not been successful and identify key variables associated with this. There is no publicly available, official criminal justice database in Great Britain identifying homicide cases in which charges are brought. As such, in gathering information about British cases of women killed in so-called SGGWs, the researcher began by consulting the website of the organization We Can’t Consent to This. This organization collects information about women killed by men in alleged SGGWs and publishes details of the cases on their website. WCCT was the first source documenting cases of women killed in the UK. Contact was made with WCCT, who shared with the researcher a spreadsheet containing details of the cases they had recorded. The organization also campaigns for legislative change to end the use of consent as a defense in cases of women killed by men (We Can’t Consent to This [WCCTT], 2019). The details of victim and offender names in cases in which a conviction for a homicide offense (murder, manslaughter, or culpable homicide) had been secured were noted. The researcher then used the Nexis news database to source mainstream media coverage of these femicides. In addition, searches were conducted for other femicides not included in the WCCTT data set that fit the inclusion criteria. A combination of different search terms were used including “murder,” “manslaughter,” “rough sex,” and “sex game gone wrong.” Given that cases of alleged SGGWs attract considerable media attention because of their newsworthy (Jewkes, 2015) elements, this method of data collection was considered sufficiently robust for the purposes of this research. However, it should also be noted that there may be other cases of alleged SGGWs not captured by this data collection method given the subjective thresholds of importance attached to news media content, particularly when such stories competed with other topics of the day. This serves to highlight the low priority attached to violence against women (Monckton-Smith et al., 2014; Stark, 2007).
Thereafter, an Excel spreadsheet was set up for the purposes of this research project, and the researcher began to collect descriptive information relating to each case, including victim and perpetrator ages and occupations, the nature and length of the relationship between victim and perpetrator, the homicide location and method, perpetrator’s criminal record and history of violence against women, type of conviction and length of sentence. This information was obtained from the articles retrieved via the Nexis news database. It was supplemented and cross-checked with information from sentencing remarks (where available) and from records available via Lexis—a legal database holding official court documentation, for example, documents of proceedings in cases where perpetrators have appealed their conviction or sentence. Thereafter, the descriptive information was entered into the spreadsheet and imported into the IBM SPSS Statistics package. Frequency tables were generated and the chi-square test was applied where appropriate to establish whether differences between observed and expected frequencies were statistically significant. The findings are detailed in the following section.
Findings
The data collection exercise identified cases of 43 women who had been killed by men in alleged SGGWs between 2000 and 2018 in Great Britain where the perpetrator had been convicted of a homicide offense in relation to their deaths. The homicide offenses covered were murder, manslaughter, and culpable homicide; manslaughter in the English and Welsh context is considered equivalent to culpable homicide in the Scottish context. There were a total of 40 perpetrators, as two perpetrators had killed more than one victim. The majority of cases were located in England (86.0%, n = 37), with three in Wales (7.0%, n = 3), and three in Scotland (7.0%, n = 3). The search did not locate any cases fitting the criteria in Northern Ireland. In this section, the findings are presented under four headings: number of convictions 2000–2018; characteristics of victims, perpetrators, and relationships; the nature of the homicides; and criminal justice outcomes.
Numbers of SGGW Convictions
The number of cases per year ranged from zero cases in the years 2004, 2005, 2008, and 2014 to six cases in 2018, see Figure 1. The differences observed in the number of cases pe year were statistically significant, χ2(18, N = 43) = 57.68, p < .001.
Number of femicides per year 2000–2018 where perpetrator claimed SGGW.
Characteristics of Victims, Perpetrators, and Relationships
Victim age ranged from 16–66 years with a mean of 31.7 years (SD = 11.14 years) and a median of 31.0 years. With regard to age groups (see Table 1), there was an overrepresentation of victims in the younger age groups, notably 25–34 years (41.9%, N = 18) and 16–24 years (30.2%, N = 13). The unequal distribution of victims across different age groups was statistically significant, χ2(5, N = 43) = 236.83, p < .001. Perpetrator age ranged from 18–63 years with a mean of 37.4 years (SD = 11.37 years) and median of 38 years. As such, the difference in mean age of victims and perpetrators was 5.7 years and median age 7.0 years. Compared with general age differences between male and female heterosexual couples—the average age difference between couples marrying for the first time in England and Wales is 1.9 years (ONS, 2019)—the age differences observed in this sample are markedly higher. In terms of perpetrator age groups (see Table 1), there was an overrepresentation of perpetrators in the 35–44 years age group (37.5%, N = 15). The unequal distribution of perpetrators across different age groups was statistically significant, χ2(4, N = 40) = 90.00, p < .001. The age difference between victims and perpetrators ranged from a victim who was 31 years younger than the perpetrator to a victim who was 23 years older than the perpetrator. The averages in terms of the number of years that the perpetrator was older than the victim was a mean of 6.1 years (SD = 9.85 years) and a median of 4.0 years. When exploring the combination of victim and perpetrator age, the most frequent combination was perpetrators in the age group 35–44 years with victims in the age group 25–34 years (23.3%, N = 10). The second most frequent combination was perpetrators aged 16–24 years and victims aged 16–24 years (14.0%, N = 6).
Victim and Perpetrator Age Groups.
Occupations were coded using the Standard Occupational Classification as a guide (ONS, 2010), and victims and perpetrators were assigned to occupational groups (see Table 2). Where information about victim occupation was available (N = 34), the most common occupational group for victims was “Caring, leisure, and other service occupations” (26.5%, N = 9). This included occupations such as care assistants, classroom assistants, and hairdressers. The second most common groups were “Students” (14.7%, N = 5) and “Sex workers” (14.7%, N = 5). The unequal distribution of victims across occupational groups was statistically significant, χ2(11, N = 34) = 63.64, p < .001. In relation to perpetrators, where this information was available (N = 29), the most common occupational group was “Unemployed” (27.6%, N = 8). The second most common occupational groups were “Skilled trades occupations” (17.2%, N = 5), which included positions such as carpenter and builder, and “Process, plant, and machine operatives” (17.2%, N = 5), which encompassed jobs such as forklift truck drivers and machine operators. The unequal distribution of perpetrators across occupational groups was statistically significant, χ2(9, N = 29) = 56.90, p < .001.
Victim and Perpetrator Occupational Groups.
Regarding the nature of the relationship between the victim and the perpetrator (see Table 3), the most frequent relationship was “Partner” (37.2%, N = 16). This was followed by “Just met” (16.3%, N = 7), where the victim and perpetrator had encountered each other for the first time prior to the homicide, and there was no significant time break between this meeting and the homicide. Six relationships (14.0%) were classed as “Client,” where the victim was engaged in sex work and this was the context in which she had encountered the perpetrator. In five cases termed “Ex-partner” (11.6%), the perpetrator and victim had been in a partner relationship but had separated prior to the homicide. In four cases (9.3%), the relationship between victim and perpetrator was “Friend”; and in three cases, they were acquaintances (7.0%). Two cases were classed as “First date” (4.7%), which differed from “Just met” in that there had been prior contact between the victim and the perpetrator—notably online—prior to the first in-person meeting which preceded the homicide. The differences observed in types of relationship between victim and perpetrator were statistically significant, χ2(6, N = 43) = 134.98, p < .001. Where the relationship was “Partner” (N = 16), information was sought in the source documentation that identified the length of the relationship in months. This information was obtained for 13 cases. The relationship length ranged from 4 months to 31 years, with a mean of 113 months (9.4 years, SD = 117.9 months) and a median of 72 months (6 years). When grouped, the highest frequency was observed in the “Up to 5 years” category (46.2%, N = 6), followed by “Between 10 and 15 years” (23.1%, N = 3), “Between 5 and 10 years,” and “More than 15 years” (both 15.4%, N = 2). In most “Partner” cases, victim and perpetrator lived together at the time of the homicide (87.5%, N = 14); in only two cases (12.5%), they did not. In just below two thirds of “Partner” cases (62.5%, N = 10), the victim and the perpetrator were not married, and in just more than one third (37.5%, N = 6), they were. Of the 21 cases where the relationship was “Partner” or “Ex-partner” (N = 21) and the information was available (N = 16), in five of these cases (31.3%), the victim and perpetrator had children together and in 11 cases (68.8%) they did not. Also of interest to this research were claims made by the perpetrator about the nature of the sexual relationship between himself and the victim in cases where the relationship was “Partner” or “Ex-partner” (N = 21). In two thirds of these cases (66.7%, N = 14), the perpetrator had made claims that he and the victim regularly engaged in consensual BDSM sexual activity, which included acts of the sort allegedly performed immediately prior to the homicide, and in one third of cases (33.3%, N = 7), such claims had not been made.
Relationship Between Victim and Perpetrator.
Exploring perpetrators’ histories of criminal and abusive behavior, in cases where this information was available (N = 26), more than half (57.7%, N = 15) of perpetrators had past criminal convictions for violent offenses. The differences observed in whether or not there were past criminal convictions for violence were statistically significant, χ2(1, N = 26) = 8.00, p < .05. In relation to property offenses, where this information was available (N = 23), nearly two thirds (65.2%, N = 15) of perpetrators had such convictions. The differences observed in whether or not there were past criminal convictions for property offenses were statistically significant, χ2(1, N = 23) = 24.5, p < .001. In relation to a history of abuse of women (N = 40), this was identified in the source material in just more than half of the cases (52.5%, N = 21), but differences were not statistically significant. In relation to the perpetrator engaging in behaviors toward the victim that could be described as domestic abuse, coercive control, or stalking, this was identified in the source material in three quarters of cases (75.0%, N = 30). Differences in whether or not these behaviors were present were statistically significant, χ2(1, N = 43) = 144.5, p < .001.
Characteristics of the Homicides
With regard to homicide location (see Table 4), where this information was known (N = 42), the perpetrator’s home was the most common homicide location (47.6%, N = 20), followed by a home shared by the victim and perpetrator (23.8%, N = 10), the victim’s home (11.9%, N = 5), an outdoor location (including cars; 11.9%, N = 5), and a hotel (4.8%, N = 2). Differences observed between the homicide locations were statistically significant, χ2(4, N = 42) = 201.2, p < .001. Homicide location was also explored alongside the nature of the relationship between victim and perpetrator where this information was available (N = 42). In homicides in the context of a partner relationship (N = 16), victims were most likely to be killed at a shared home (62.5%, N = 10). In ex-partner relationships (N = 5), victims were most likely to be killed at the perpetrator’s home (80%, N = 4). In homicides where victims and perpetrators had just met (N = 7), the perpetrator’s home was also the most likely location (85.7%, N = 6). The perpetrator’s home was also the location in all cases where the victim–perpetrator relationship was “Client” (N = 5). In both the “First date” femicides, the location was a hotel (N = 2). In the three “Acquaintance” cases, one victim (33.3%) was killed at the perpetrator’s home and the other two (66.7%) were killed at the victim’s home. “Friend” cases (N = 4) were characterized by a diversity of homicide locations, consisting of outdoor locations such as cars or public parks (50%, N = 2), the perpetrator’s home (25%, N = 1), and the victim’s home (25%, N = 1).
Homicide Location.
With regard to the method used to kill the victim, where this information was known (N = 36; see Table 5), the count of methods used totaled 45 because, in six cases, perpetrators had used more than one method. The most common method used was strangulation (60%, N = 27), followed by blunt instrument (20.0%, N = 9), asphyxiation (13.3%, N = 6), and sharp instrument (6.7%, N = 3). Differences observed between the homicide methods were statistically significant, χ2(3, N = 45) = 2,205.0, p < .001. Within “strangulation” (N = 27), three categories of strangulation were identified: manual strangulation (44.4%, N = 12), ligature strangulation (37.0%, N = 10), and strangulation where manual or ligature was not specified (18.5%, N = 5). Differences observed between the strangulation types were statistically significant, χ2(2, N = 27) = 26.00, p < .001. Exploring the strangulation (grouped) method alongside the nature of the relationship between the victim and perpetrator, it was used in 75% (N = 3) of “Friend” cases, 71.4% (N = 5) of “Just met” cases, 66.7% (N = 2) of “Acquaintance” cases, 60% (N = 3) of “Ex-partner,” 50% (N = 8) of “Partner” cases, 50% (N = 1) of “First date” cases, and 50% (N = 3) of “Client” cases.
Homicide Method.
With regard to the perpetrator claiming that the victim had initiated or asked him to perform the specific act that led to her death, where this information was known (N = 41), the perpetrator did so in 58.5% of the femicides (N = 24) and did not in 41.5% (N = 17) of femicides. Differences in whether or not the perpetrator made this claim were statistically significant, χ2(1, N = 41) = 24.5, p < .001. Exploring this variable alongside the nature of the relationship between victim and perpetrator, the perpetrator claimed that the victim asked for or initiated the activity that led to death in 100% (N = 3) of “Acquaintance” cases, 80% (N = 4) of “Ex-partner” cases, 71.4% (N = 5) of “Just met” cases, 64.3% (N = 9) of “Partner” cases, 50% (N = 1) of “First date” cases, 25% (N = 1) of “Friend” cases, and 16.7% (N = 1) of “Client” cases.
With regard to whether or not the perpetrator contacted emergency services in relation to the victim’s death, where this information was available (N = 42), nearly 6 out of 10 (59.5%, N = 25) perpetrators did not contact emergency services, leaving the victim’s body to be discovered by others, and the remaining 40.5% (N = 17) did contact emergency services. Differences in whether or not the perpetrator contacted emergency services were statistically significant, χ2(1, N = 42) = 32.0, p < .001. Considering this alongside the nature of the relationship with the victim, the relationships where perpetrators were more likely than not to report the incident to emergency services were “Partner” (56.3%, N = 9) and “Just met” (57.1%, N = 4). Relationships where not contacting emergency services was more common were “First date” (100%, N = 2), “Client” (100%, N = 6), “Friend” (75%, N = 3), and “Acquaintance” (66.7%, N = 2). In “Ex-partner” cases, perpetrators were just as likely to report to emergency services as they were not to report (50%, N = 2).
Criminal Justice Outcomes
In relation to the criminal justice outcomes for perpetrators (N = 40), more than three quarters (77.5%, N = 31) of perpetrators were convicted of murder and just more than one fifth (22.5%, N = 9) were convicted of manslaughter or culpable homicide. Observed differences in conviction outcomes were statistically significant, χ2(1, N = 40) = 242.0, p < .001. Conviction types were explored over years (see Figure 2) and grouped years. Considering convictions for murder (N = 34), the greatest number of murder convictions were observed in 2000–2004 (N = 11) and 2015–2018 (N = 11), followed by 2010–2014 (N = 7) and 2005–2009 (N = 5). Differences between the numbers of murder convictions per grouped year category were statistically significant, χ2(3, N = 34) = 27.0, p < .001. In relation to manslaughter convictions (N = 9), the highest concentration was in the most recent time period 2015–2018 (N = 5), followed by 2010–2014 (N = 3) and 2000–2004 (N = 1). Differences in numbers of manslaughter convictions per grouped year category were statistically significant, χ2(3, N = 9) = 14.75, p < .01.
Number of homicide convictions per year—all, murder, and manslaughter/culpable homicide.
Exploring type of homicide conviction alongside victim–perpetrator relationship (N = 40), murder convictions were the outcome in 100% of cases where the relationship between victim and perpetrator was “Ex-partner” (N = 5), “Friend” (N = 3), or “Client” (N = 4). Murder convictions were also secured in 75% (N = 12) of “Partner” cases, 66.7% (N = 2) of “Acquaintance” cases, 57.1% (N = 4) of “Just met” cases, and 50% (N = 1) of “First date” cases. Considering “Partner” cases in relation to relationship length, where this information was known (N = 12), perpetrators who had been in a relationship with the victim for less than 12 months were more likely to be convicted of manslaughter or culpable homicide than those who had been in a relationship with the victim for more than 12 months. All of those who had been in a relationship with the victim for more than 12 months (N = 9) were convicted of murder. In relation to those convicted of manslaughter or culpable homicide, 66.7% (N = 2) had been in a relationship with the victim for 12 months or less, and 33.3% (N = 1) had been in a relationship with the victim for more than 12 months.
In relation to sentence lengths in murder convictions, excluding the two whole life orders, which are not quantifiable (N = 29), sentences for murder were between 12 and 30 years—a range of 18 years. The mean sentence length for murder was 19.8 years (SD = 4.96 years) and the median was 19 years. When sentence lengths for murder were grouped, the most common group was 15–19 years, with 41.4% (N = 12) of murder sentences falling into this category, followed by 25–29 years (20.7%, N = 6), 10–14 years and 20–24 years (both 17.2%, N = 5), and 30–34 years (3.4%, N = 1). Differences observed between the distribution of sentences across groups were statistically significant, χ2(4, N = 29) = 62.8, p < .001. Grouped sentence lengths for murder were further considered in relation to victim–perpetrator relationship (N = 30). “Partner” (N = 12) is associated with shorter sentence lengths of 10–14 years (33.3%, N = 4) and 15–19 years (66.7%, N = 8). Sentence lengths in “Partner” murder cases had a range of 7 years with a minimum of 12 years and a maximum of 19 years, a mean of 15.7 years (SD = 2.36 years), and a median of 16.0 years. Sentence lengths in murder cases involving the “Ex-partner” relationship (N = 5) span a broader range of categories than partner relationships: 10–14 years (20.0%, N = 1), 15–19 years (40.0%, N = 2), and 20–24 years (40%, N = 2). “Ex-partner” murder cases had a range of 8 years with a minimum of 14 years, maximum of 22 years, mean of 19.2 years (SD = 3.27 years), and median of 19.0 years. Sentence lengths in “Just met” murder cases (N = 4) tended to be concentrated in longer sentence groups than “Partner” and “Ex-partner” cases: 20–24 years (25%, N = 1), 25–29 years (50.0%, N = 2), and 30–34 years (25%, N = 1). Sentence lengths in “Just met” murder cases had a range of 6 years with a minimum of 24 years, maximum of 30 years, mean of 27.3 years (SD = 2.75 years), and median of 27.5 years. Murder cases where the relationship between victim and perpetrator was “Friend” (N = 4) were distributed across three sentence length categories: 15–19 years (25.0%, N = 1), 20–24 years (25.0%, N = 1), and 25–29 years (50.0%, N = 2). Sentence lengths in “Friend” murder cases had a range of 7 years with a minimum of 18 years, maximum of 25 years, mean of 22.0 years (SD = 3.56 years), and median of 22.5 years. Sentence lengths in “Client” murder cases (N = 2) fell into two categories: 15–19 years (50%, N = 1) and 25–29 years (50%, N = 1). Sentence lengths in “Acquaintance” murder cases (N = 2) fell into two categories: 20–24 years (50%, N = 1) and 25–29 years (50%, N = 1). The sentence length in the only “First date” murder case (N = 1) was 25–29 years.
Turning to explore manslaughter convictions (N = 9), given lower numbers than murder convictions (N = 31), the same depth of analysis could not be conducted. However, the following key points emerged. The shortest sentence was 4 years and the longest was 20 years—a range of 16 years. The mean sentence length was 8.9 years (SD = 5.96 years) and the median was 6.0 years. When sentence lengths for manslaughter or culpable homicide were grouped, the most common groups were 0–4 years 33.3% (N = 3) and 5–9 years (33.3%, N = 3).
Discussion
This research supports claims in the literature that the SGGW defense is on the increase and is becoming a more prominent feature in criminal proceedings (Edwards, 2017; Monckton-Smith, 2019). The year 2018 saw the highest number of convictions, and there were statistically significant differences between year-on-year conviction counts, suggesting that such a distribution cannot be attributed simply to chance. That the vast majority of convictions were for murder may suggest that cases where prosecutions are pursued are those in which the evidence is considered sufficiently compelling by those tasked with these decisions. Such decisions will in turn draw upon social and cultural values around femicide, which tend to prioritize ideal victimhood and accord a low status to women killed by men with a degree of legitimate access to them (Dobash & Dobash, 2015; Monckton-Smith et al., 2014).
Given the focus of this research upon convictions, cases in which prosecutions did not commence, were discontinued, or where a not guilty verdict was delivered were excluded. Of potential relevance here were findings relating to perpetrator occupational group, the most frequent of which was “Unemployed.” This could be suggestive of two possibilities: that men from higher occupational groups are simply not killing their partners and using the SGGW defense, or that they are successfully using this narrative to avoid prosecution and conviction. It could be argued that the perpetrators in this sample had less cultural capital (Bourdieu, 1973) upon which to draw when constructing the narrative of the SGGW than perpetrators whose occupations were indicative of a higher social class. The perpetrators in this sample were convicted of homicide offenses, but those able to construct a more compelling narrative may be absent from the data set because they were less likely to have faced prosecution in the first place or be convicted of homicide offenses in the event that they did. More empirical research is needed which spans a wider range of criminal justice outcomes to further investigate the new questions emerging in relation to these points. In addition to this, the political-economic context of men’s violence against women deserves considerably more scrutiny than has been the case to date. Specifically, future research should examine how the seismic systemic changes brought about by neoliberalism—for example, widespread deindustrialization—have served to facilitate and enable the development of harmful subjectivities among men (Ellis, 2016). Patriarchy’s contemporary neoliberal manifestation has thus far escaped extensive feminist scrutiny, with a handful of exceptions (see, for example, Bumiller, 2008; Fraser, 2013). However, the identification of structural drivers of misogynistic behavior is crucial in tackling phenomena such as SGGWs.
Consistent with established knowledge that femicide perpetrators do not kill “out of the blue” (Adams, 2007; Dobash & Dobash, 2009; Monckton-Smith, 2012; Monckton-Smith et al., 2014), this research found perpetrators in SGGW femicides were more likely than not to have previous convictions for violence and property crimes, a history of abusing women, and a pattern of behavior toward the victim characterized by abuse, coercive control, or stalking. These findings are important in challenging claims that SGGW femicides are unfortunate accidents in which an otherwise “normal” man of good character kills a female partner. The femicides explored in this research are not one-off, isolated incidents, but they were the culmination of entrenched and well-evidenced patterns of abusive and coercively controlling behavior toward women. The literature exploring dynamics in coercive control and abuse suggested that perpetrators draw upon specific victim vulnerabilities and general heteronormative conventions around romantic love to establish a basis for personalized abuse (Monckton-Smith et al., 2014; Stark, 2007). The findings of this research appear wholly consistent with these patterns. There were statistically significant age differences between victims and perpetrators, characterized by the perpetrator’s seniority in age. There was an overrepresentation of victims employed in “Caring, leisure, and other service occupations.” The values and traits associated with such occupations—e.g., empathy, selflessness, care for the well-being of others—and the relatively low economic reward associated with these careers are precisely the type of characteristics and vulnerabilities targeted by perpetrators to establish the foundations for abuse and coercive control. Strangulation was overrepresented in this sample, a heavily gendered method of homicide strongly associated with prior histories of abuse toward women (Edwards, 2015; Glass et al., 2008; Monckton-Smith, 2019; Stark, 2007). The most frequent location for the homicides was the perpetrator’s home, the home being an important locus of control (Stark, 2007). Edwards (2017) argued that the perpetrator’s claim of victim consent in cases of alleged SGGWs is “an allegation easily made and difficult to refute” (p. 97). However, this research has established that when SGGW femicides are located within the broader context of a coercive control discourse—as opposed to a crime of passion discourse, which capitalizes upon the cultural mainstreaming of “rough sex”—such claims become less challenging to refute.
Despite the prevalence of murder convictions in the sample—which may at first sight appear to be an encouraging finding—when criminal justice outcomes were examined in more detail, disparities emerged. The lesser the degree of legitimate access to a victim (e.g., in “Just met” or “First date” dynamics), the more severe the conviction and sentence appeared to be. “Partner” cases where the victim and perpetrator had been in an intimate relationship for less than 12 months were more likely to result in a manslaughter or culpable homicide conviction than a murder conviction. Sentences for murder convictions in partner cases were at the lower end of the scale when compared with sentence lengths in cases featuring other types of victim–perpetrator relationship. The SGGW narrative appears to be one that is most effective when situated against the backdrop of an intimate relationship; in two thirds of partner and ex-partner cases, the perpetrator claimed that he regularly engaged in consensual BDSM with the victim. Against a backdrop of formal sex equality, the intimate relationship, especially when it is long-standing, is imbued with notions of mutuality, respect, and consent. Such assumptions reinforce the myth of a level playing field where all women are free to leave relationships and obscure the realities of abuse, coercive control, and strategic planning of femicides (Monckton-Smith, 2019). While this did not prevent perpetrators in long-standing intimate relationships with victims being convicted of murder in this sample, it can be argued that these cultural factors may well have influenced the shorter sentences they received. This supports the argument in the existing literature that women killed by men with whom they have had an intimate relationship are accorded a lower status than those killed by strangers (Dobash & Dobash, 2015; Monckton-Smith et al., 2014) and adds to this literature in highlighting an association between closer victim–perpetrator proximity and less severe criminal justice outcomes in cases of alleged SGGWs. The SGGW defense may be becoming more effective in securing more favorable criminal justice outcomes for perpetrators with a higher degree of legitimate access to a victim via a current intimate partner relationship.
This research highlights the use of the SGGW defense in cases where there are a wider range of victim–perpetrator relationships than partner or ex-partner. It is emerging in femicides where the victim and perpetrator have just met, are on a first date, are friends, acquaintances, or have met through the perpetrator’s procurement of sex workers. More than half of the 43 victims in this research did not have a partner or ex-partner connection with the perpetrator. The normalization of BDSM noted in the literature (Busby, 2012; Downing, 2013; Moore & Khan, 2019; Weiss, 2006) and the subsuming of strangulation within this discourse create a culturally approved script which all perpetrators of fatal violence against women are able to use, regardless of whether or not they are or have ever been in an intimate relationship with the victim. This is clear in the findings of this research in relation to claims made by well over half of the perpetrators that the victim initiated or asked for the activity that led to their death, whether that was strangulation, impact through hitting or kicking, or the use of a knife or sharp instrument during sex. The ubiquity of BDSM makes it easier for perpetrators to make such claims about victims, who may indeed have discussed this topic with friends or partners, not because they wanted to experiment or consented to such acts, but because BDSM is so prominent in popular culture that it is highly likely to be discussed in everyday social interactions. However, testimony about such conversations can be presented as evidence for the defense in cases of alleged sex games gone wrong, which serves to further sully the victim’s reputation and responsibilize her for her own death in much the same way that a victim’s sexual history is often used against her in rape and sexual assault trials. In addition, the disproportionate concentration of victims and perpetrators observed in the 16–24 years age group may be indicative of the fact that the cultural normalization of BDSM has occurred during the formative years of people in this age group. As such, women in this age group may be uniquely vulnerable to femicides and non-fatal harm being presented by their abusers as SGGWs given the strength of this cultural reference point among their peer group. Qualitative research exploring cases affecting victims and perpetrators in this age group is required to investigate this further. Finally, this research further reinforces the need for centralized, publicly available official data sources for femicide and violence against women. While official statistics often obscure the “dark figure” homicides that are missed or wrongly classified under other manners of death (Brookman, 2005), the continued reliance of feminist criminologists upon media sources and unofficial repositories to identify cases limits the scope, depth, and impact of research in this area. If we are to tackle violence against women, we need access to comprehensive data sets to do so; the Violent Death Reporting System in the United States is one such example that the British government could build upon.
Conclusion
Despite consent to bodily harm not being considered a legitimate defense in the context of sadomasochistic sexual activity, femicide perpetrators are increasingly drawing upon this premise in claiming that women’s deaths are the outcomes of SGGWs. Most men convicted of homicide offenses in relation to such cases have a history of harmful behavior in general and abuse of women in particular. As such, femicides framed as SGGWs are not the accidents or one-offs that they are often presented as. Rather, they represent consistency and continuity of entrenched misogynistic and abusive courses of conduct. The cultural normalization of BDSM in the context of neoliberal political-economic ideology has enabled abusers to justify and excuse fatal violence against women, using formal sex equality and women’s sexual liberation against them. The term “game” implies rules, fairness, transparency, and a level playing field. The playing field is anything but level. The so-called sex “games” within the cases explored in this research are the very antithesis of such principles. It is important to develop and expand criminological inquiry into so-called SGGWs to expose and challenge both individual perpetrators’ use of such claims and to interrogate the cultural and social context in which such narratives take root and thrive.
Footnotes
Acknowledgements
The author would like to thank We Can’t Consent to This for their help in identifying the cases in this work and making their dataset available. The analysis of this dataset is that of the author and not that of WCCTT or its members. The author would also like to thank the anonymous reviewers for their thoughtful and helpful feedback on the manuscript.
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) received no financial support for the research, authorship, and/or publication of this article.
