Intimate Partner Femicide: Using Foucauldian Analysis to Track an Eight Stage Progression to Homicide

The links between IPF and domestic abuse are powerful, and a history of domestic abuse is a key risk marker in those who are IPF killers (Dawson & Piscitelli, 2017). A history of domestic abuse is also anecdotally increasingly linked to other homicide categories, notably mass homicides (De & Haldevang, 2018; Everytown, 2017), and its association to criminal activity more generally is being recognized. Research suggests that domestic abuse characterized by patterns of coercive control and/or stalking is more likely to end in homicide (Campbell, Glass, Sharps, Laughon, & Bloom, 2007; Campbell, Webster, et al., 2003; Dobash & Dobash, 2015; Mullen, Pathé, & Purcell, 2000; Stark, 2009). The coercive control discourse is a resistance to more traditional and dominant rationalizations for domestic homicide which have been explained with reference to the crime of passion.

Recent legislative changes in the United Kingdom and Ireland have criminalized controlling patterns—see Section 76 of Serious Crime Act 2015 (UK); Domestic Abuse (Scotland) Act 2018; and Domestic Violence Act 2018 (Ireland), also for stalking patterns, see Protection of Freedoms Act 2012 (UK). These political moves, at least in part, reflect the importance and value of control and stalking in predicting serious harm and homicide. For example, M. Johnson (2008, 2011) identifies three typologies of domestic abuse, broadly organized around perpetrator motivation, which are referred to as intimate terrorism, violent resistance, and situational couple violence. The pattern intimate terrorism has the strongest links to controlling patterns and the highest risk for homicide. Stark’s (2009) work described and defined coercive control which he constructs as a liberty crime, arguing that it erodes personal freedoms and choice leading the victim to a state of subjugation. Again, perpetrator motivation is set around a need for control, and challenges to that control produce an environment where homicide is more likely. Research in an Australian study by H. Johnson, Eriksson, Mazerolle, and Wortley (2019) found that controlling patterns were identified in most cases of IPF. Some argue that domestic abuse cannot be split into mutually exclusive categories (Gulliver & Fanslow, 2015), but describing heterogeneity in behavior and motivation through typologies can help organize approaches to risk. For example, there is a precedent in stalking research that motivation can be categorized and has a relationship to risk and threat (Mullen et al., 2000). The notion that IPF is part of a process motivated by control, rather than a response to an incident, is central to considering risk in this article.

Predicting if, and when, homicide may occur dominates U.K. police and professional responses to disclosures of domestic abuse and stalking, as it does in many jurisdictions. Two key risk approaches are taken, especially in triage: an actuarial and a clinical approach. There are problems inherent in any risk assessment model, and the history of assessing risk of violence specifically suggests that it is not a precise science (Shapiro & Noe, 2015). Police assess the imminence of homicide using a quasi-actuarial approach which identifies and measures presence of domestic abuse or stalking using Risk Identification Checklists (RICs). Such RICs are lists of known behavioral markers which have been found to be present in previous research into domestic homicides. Thus, they identify if known high-risk markers are present in any relationship where domestic abuse is suspected. Broadly, in practice, the number of markers identified on the list will influence whether the victim is considered high, medium, or standard risk for homicide. This risk level will then determine the type of resources allocated to that victim (Monckton Smith et al., 2014). However, there is little support for the idea that sheer numbers of risk markers alone equate to imminence of homicide (Shapiro & Noe, 2015). The presence of some high-risk markers on RICs, like violence, fear, and control, may identify domestic abuse, but markers like separation and escalation in control better identify the potential for imminent homicide (Dobash & Dobash, 2015; Gilchrist, 2013; M. Johnson, 2008; Stark, 2009; Websdale & Dedolph, 2000). It is probably more accurate to say that markers interact with each other in quantifying risk, but this is not always reflected in use of RICs. Clusters of risk markers are considered more predictive than numbers of them; for example, where there is control, violence and a separation after living together, there is said to be a 900% increase in the potential for homicide (National Center for Injury Prevention and Control, 2003).

A clinical approach to risk assessment is also argued to be problematic. Domestic abuse, coercive control, and stalking are not clinical conditions, but patterns of behavior. There is a strong backlash noted in feminist scholarship which constructs clinical explanations for domestic abuse as excusing the patterns of behavior. It is argued that social and cultural belief systems and systemic gender inequality are more influential (Dobash & Dobash, 1998; Hall, 2015; M. Johnson, 2011; Yllo & Bograd, 1988). Mental illness and drug and alcohol misuse are thought to exacerbate the predicament the behaviors create, but they are not considered to be causative (Stark, 2009). M. Johnson (2008) identifies two distinct typologies of abuser who are controlling and dangerous, but neither has significant links to mental illness, although there is a relationship to personality disorder. Profiles of rejected stalkers (dominated by intimate partner stalking) also reveal personality disorders to be more of a problem than mental illness (Farnham, 2018; Mullen et al., 2006; Mullen et al., 2000). The emergence of stalking clinics in England and Wales which employ a dual approach to risk assessment, and use both actuarial and clinical models in triage (Farnham, 2018), reveal the importance of both, and the multi-faceted nature of domestic abuse and IPF.

The introduction of domestic homicide reviews (DHRs) in England and Wales in 2011 as part of the Domestic Violence, Crime and Victims Act (2004) was another move toward identifying opportunities to prevent future homicide and forms part of the Government’s strategy to end violence against women and girls (Home Office, 2016c). The model used in the DHR focuses on tracking the antecedent histories in cases of IPF (and other categories of domestic homicide) to produce a chronology which may go back many years. Opportunities for intervention are then identified by looking at the chronology and, especially, interactions with professionals and agencies. Research has revealed that there are many consistent themes in the chronologies of IPF cases in the DHRs (Home Office, 2016a; Standing Together, 2016). Chronologies and temporal sequences are potentially useful in understanding the dynamic nature of risk, and how and when it can escalate.

Temporal sequencing has an established presence in homicide studies, from Luckenbill’s (1977) microlevel situational analysis producing the six stages of male confrontational homicide to Stanton’s (2016) macrolevel structural analysis producing the 10 stages of genocide. Wertham’s (1937) thesis on catathymic crisis as an explanation for apparently motiveless violent crimes produced a five-stage sequence, which has been more recently developed to understand an offender’s psychiatric progression to a decision to kill in cases of serial killing (Garrison, 1996; Schlesinger, 2002). Mullen (2004) and Adams (2007) analyses of mass homicides and IPH, respectively, mapped the consistencies in the antecedent histories of the homicides to understand the progression to fatal violence. This approach, therefore, is potentially useful.

The diversity of methods, perspectives, and categories of homicide in temporal sequencing notwithstanding, there is a unifying idea articulated by Stanton (2016) that the later stages in the sequence are preceded by the earlier stages, and the progression helps understand the motivation to kill and identify opportunities at each stage to prevent the homicide from happening.

The theoretical positions around the motivation to kill in IPF are important in constructing and organizing a temporal sequence. I draw from Stark (2009) and M. Johnson’s (2008) work which situates controlling patterns as driving domestic abuse, and any threats to that control as raising risk for homicide. I also consider the work of Mullen et al. (2000) around motivation in the rejected stalker category to be important as the literature builds a picture of stalking and controlling characteristics interacting together and simultaneously present in cases of IPF (Monckton Smith, Szymanska, & Haile, 2017). The literature also tells us about motivation to kill in IPF, and Dobash and Dobash’s (2015) work, which concludes that “a man may decide to ‘change the project’ from attempting to keep her (the victim) within the relationship to destroying her for leaving it” (p. 39), most clearly articulates that a change in thinking may precede homicide in this context. Changes in thinking are often a specific stage in homicide sequencing, and so-called “last chance” thinking, where an individual may feel there is nothing left for them, is a risk marker for imminent homicide. Mullen (2004) describes this in his description of the change in thinking in autogenic massacre (e.g., mass killings). Mullen argues that autogenic homicide is “self-generated,” that is, the perpetrator generates frustration, anger, and rage in respect of specific issues or challenges and moves increasingly toward considering homicide as the only relief. A decision to kill is made and acted on, rather than the killing being a spontaneous response to a proximal provocation. Schlesinger’s (2002) description of the change in thinking during catathymic crisis is also helpful. Along with Mullen (2004), this establishes thinking change which situates homicide as a relief or solution to challenges. Although I do not argue that IPF is the result of catathymic crisis, the description of the change in thinking is useful:

Changes in thinking, especially a move to last chance thinking, which lead to a decision to kill, are established in homicide temporal sequencing and also in the literature surrounding IPF. Scholars argue that IPF killers respond to a trigger with an attitude captured in the often-used quote “if I can’t have you, no-one can.” It is recognized that some IPF appear to be more spontaneous, especially when there is little temporal distance between a provocation or challenge and the fatal violence. Although this is certainly the case, it is possible that the risk of homicide in such relationships follows a similar, if condensed, pattern. It is also recognized that there is a body of work which looks at psychodynamic stages in such homicides (C. Johnson & Sachman, 2014) and more nuanced and detailed individual psychopathologies, especially in familicide (Websdale, 2010). The argument proposed is that IPF is part of a journey where the motivation to abuse (need for control) is linked to the motivation to kill (loss of, or threat to, control). Breakdown in control can be preceded by a somewhat broad spectrum of triggers, and this often revolves around not only separation but also financial ruin and mental or physical health crises. This means that there will be many more relationships where there is a breakdown in control than there are homicides. However, if the response shows signs of last chance thinking, or determined revenge, then risk of homicide escalates.

Foucault (1972) characterized discourse as more than merely a linguistic concept; he saw it as a way of knowing about something. Domestic abuse discourses, for example, tell us, among other things, what domestic abuse is, who its victims and perpetrators are, what motivates it, and how we should respond. There will be multiple discourses in circulation at any point in time, but some become dominant. Foucault saw no linear progression or incremental building of knowledge within discourse, merely various constructions of it with varying degrees of power to assert their position as the truth. His genealogical approach is characterized by “digging” through discourses across time, and producing a history of the present. In this article, I specifically focus on how discourse impacts on assessment of, and response to, risk of homicide in domestic abuse. I identified two discourses, which I call the crime of passion discourse and the coercive control discourse, which are key. The crime of passion discourse shares knowledge with dominant discourses of domestic violence and romantic heterosexual love. There is heavy focus on gender, gendered roles, and perceived differences in male and female behaviors, responsibilities, and expectations (Dobash & Dobash, 2002; Monckton Smith, 2010; Websdale, 1999). More importantly, this discursive position prescribes the gendered order where women take a subservient position, as normal or natural, providing justifications for violence and homicide against women should the “natural order” of things be broken (Monckton Smith, 2012). These justifications have even formed legal defenses to murder and excused violence against women (Lees, 1997). A crime of passion homicide is characterized by spontaneous and unpredictable violence motivated by something the victim has done, and which is perceived by the perpetrator to be intolerable. What can be considered intolerable is also discursively constructed, and Lees’s (1997) research of Old Bailey trials identified three key behaviors sanctioned as bad enough to potentially produce justifiable fatal violence. They are nagging, infidelity, and rejecting the gendered order. The coercive control discourse, which has its theoretical roots in feminism, is a resistance to this knowledge and constructs IPF as preceded and motivated by a breakdown in the control the perpetrator has of the relationship (and which they may feel entitled to) and their perceived loss of status or stability as a result. Traditional gendered roles are represented as socially constructed and discriminatory, and not natural. These two discourses, which both have a footprint in professional responses to domestic abuse victims, represent risk very differently. Risk of homicide in the crime of passion discourse is seen as partly the product of victim behaviors and provocations. The homicides are only seen as predictable when the perpetrator displays routine serious violence. In this article, I look at the way risk escalates when considering control rather than passion, and violence as preceding homicide, taking account of knowledge produced in coercive control discourse. My specific aim was to identify a typical risk escalation in cases of IPF.

I followed Carabine’s (2001) process for performing a Foucauldian discourse analysis. The first stage was to find all possible sources of information for the chosen topic, which is IPF chronologies. It is necessary for the researcher to know these data, to keep rereading them, and become immersed in them (Carabine, 2001). Foucauldian analytical tools which focus on concepts of power, truth, and knowledge are particularly useful in exploring social issues. Knowledge is seen to be produced within discourse in this approach, so it is important to consider who has the power to present their perspective as the truth. This may be through medical, legal, political, cultural, or social institutions, and where a number of these institutions share and agree on a perspective as the truth, what is called a discursive formation is built. Where this is the case, contrasting opinion can become a mere resistance to what is accepted. In many ways, the coercive control discourse is a resistance to traditional truths, having its roots in feminist scholarship, which has been a resistance to most institutional knowledge around gender. However, it is gaining dominance, and institutional endorsement, and this makes it an important discourse to consider. The truths presented in the coercive control discourse direct this analysis, and I set them in contrast to the dominant crime of passion discursive traditions.

To begin, 575 homicide cases involving women killed by men were identified using the Counting Dead Women database (Ingala Smith, 2018). This was all cases from 2012-2015. There were 372 IPF cases from the total of 575 homicides. Every case was reviewed drawing from published media and homicide reviews to establish the broad history and circumstances of the homicide and to identify common and consistent themes. Those themes were placed into a simple temporal sequence. The sequence was initially formed of three stages: prerelationship themes, relationship themes, and postrelationship themes. Dominant prerelationship themes were identified as: perpetrator has history of domestic abuse, controlling patterns, or stalking. Relationship themes were identified as: presence of domestic abuse, controlling patterns, and monitoring/stalking behaviors. Postrelationship themes were identified as: separation instigated by the victim, stalking, threats to kill, or a threat to control or status (like illness or financial ruin). Then, risk markers appearing on standard RICs used by police (Domestic Abuse, Stalking and Harassment and Honour [DASH], 2009) which were found to be present in the cases, were placed in the three stage-sequence, and any marker could potentially appear in more than one stage. These markers are consistent in research and histories of IPF cases and are considered to identify domestic abuse and predict risk of homicide (DASH, 2009). The markers were history of abuse or stalking, possessiveness, control, monitoring, violence, sexual violence, isolation, threats to kill, threats to suicide, stalking, separation, escalating control, or violence. Each marker had a working definition for consistency. Then, additional recurring themes identified through the analysis were included, and the combined data produced five additional stages to the sequence, creating eight stages which are named: prerelationship, early relationship, relationship, trigger event, escalation, change in thinking, planning, and homicide.

The coercive control discourse gives meaning to the sequence, and when control is considered as influencing every stage, this more nuanced sequence emerges.

This sequence then formed the basis for a deeper analysis, which involved focusing on a convenience sample of 25 cases of IPF drawn from the researcher’s professional work. These cases provided much more detailed information about the progression of the relationships. All cases were of women killed by a male intimate partner within the last 15 years, and all cases provided detailed data beyond what could be obtained through a media search. All possible sources of information about the cases were gathered, and there were different types of data available for each case, but information came from DHRs; news and media reports; documentaries; court reporting; Hansard reports; and professional, perpetrator, and family reflections and perspectives. Because of the inclusion of personal comment in the cases, all remain anonymized, and no identifying information is included in this article. The sample, with its diverse sources of data, provided detailed information which allowed a concentrated examination of each identified stage. The following analysis explains how each of the eight stages was identified and developed.

Drawing from coercive control discourse, the eight stages which show a risk escalation can be explained through a narrative of controlling patterns being central to a perpetrator’s motivation and behaviors, and forming the homicide trigger. However, dominant discourses are in conflict in many respects and construct risk and culpability as lying more equally between perpetrator and victim, and control as an abusive “action” rather than the core motivation and driver of domestic abuse. This conflict could potentially have significant impact in risk assessment.

Another point of conflict is that in some cases the temporal distance between the homicide trigger and the homicide was represented as very short, just a matter of hours. It could be argued that this short time lag may indicate that some stages were missed, specifically the stages between a trigger (Stage 4) and the homicide (Stage 8). To travel from homicide trigger to homicide in less than 24 hr may suggest that there was an emotional and spontaneous response to the trigger, and this is part of the thinking in the crime of passion discourse. For example, in one case, a temporal gap of around 3-4 hr was accepted in court as an immediate response. However, that presupposes there was no other trigger preceding the identified provocation. It could also be argued that 3 hr could be time enough to consider options; this depends on the discursive position taken. Further complicating these short gap cases was another consistency: there had been use of serious domestic violence in the history of most of these cases. This may suggest that the perpetrator resorted to violence routinely, and this was a step too far. This again would be part of the crime of passion discourse. However, it is equally possible that there was a rapid progression through the final stages, with violent perpetrators more likely to proceed rapidly. A key consistent theme supporting the possibility that all eight stages were present is the difference in the quality and type of violence used in the domestic violence history, and that used in the homicide, which is also found in previous research (Monckton Smith, 2012). These short-gap IPFs are more likely to be constructed in defense narratives and media reports as routine violence taken “a step too far” or even as accidental, with the homicide seen as a natural and inevitable progression of the history of violence. However, if the quality of the violence changes, and homicide is a direct result of the type of violence used, there may have been a change in thinking. In the initial sweep of 372 cases, these short-gap homicides made up around 30% of the sample. It was considered at first that there may be two homicide progressions, but rapid travel through the final stages may be just as likely.

This eight-stage progression, as a discourse, is a resistance to dominant discourses of domestic abuse and IPF at many levels. It subverts discursive knowledge in the crime of passion discourse that suggests that IPF is preceded by a proximal provocation, is characterized by a loss of control, and is passionate. This coercive control discursive position, in contrast, suggests that perpetrators are not representative of all men but are representative of each other in many ways. It may be that perpetrators feel the need for control as a complex amalgam of individual personal histories and experiences, and cultural authorization of entitlement, but their route to feeling this way is less important in this discourse than the fact that they feel it.

The hypothesis that IPF is part of a journey where the motivation to abuse (need for control) is linked to the motivation to kill (loss of, or threat to, control) drives the interpretation of the data and the findings around risk. The more positive message might be that the coercive control discourse suggests these homicides are predictable and, therefore, there may be opportunities for prevention. In the eight-stage progression, travel through all eight stages is not inevitable. However, what was found was that perpetrators may get to Stage 5, regain control, and return to Stage 3. They may get to Stage 5 and move on to their next victim. These circular journeys within the eight-stage sequence were common. Even where a change in thinking has happened, opportunity and circumstances play a role, and this is important in terms of risk. Interventions at any stage can be qualitatively different but may be effective in stopping the progression.

Where Stages 1-2 are positively identified, there is much higher likelihood that attempts at separation, should the relationship continue, will be met with significant resistance. It is important to note that a relationship can be established in the mind of the perpetrator through as little as one date or one sexual coupling, and this is supported in stalking research (Mullen et al., 2000). Where there is progression through Stage 3, irrespective of the length of that stage, there is much higher likelihood that separation will be very difficult, or even dangerous. Travel through Stages 4 and 5 is clearest indication of the increased potential for homicide. Travel through Stages 6-8 may be very rapid where there has been routine and serious domestic violence. In most cases of IPF, travel through Stages 6-8 takes longer, often around 1-2 months, and thus interventions may be useful.

Finally, the sequence and coercive control discourse give opportunities to think about possible interventions at each and every stage. It also has value in identifying high-threat individuals, and where in the temporal journey any relationship may be.

The coercive control discourse provides a plausible and logical interpretation of the common characteristics and behaviors noted in the extant research that precede IPF. It is resistant to the idea that violence alone is the most significant predictive risk marker and also that IPF is spontaneous and situational. As in stalking research, the sequence suggests that motivation is central to understanding risk of homicide. In terms of assessing risk, the temporal sequence has utility in tracking an escalation and understanding the dynamics of increasing and decreasing risk. There are some interesting areas for further research which are suggested and include exploring the benefits of identifying control issues by clinicians, health professionals, social workers or others, and interventions which may address their impact. Perpetrator programs that focus on control may also be an interesting area for development. More research into ways in which victims may exit controlling relationships safely is urgently indicated, as is research into what may cause men to become controlling or need such control in their lives, and the way they can safely exit intimate relationships.

The eight-stage sequence has utility when used in tandem with current risk processes and may add an extra dimension in decision making for victims and professionals.