Risk Factors for Intimate Partner Homicide

Margo Wilson and Martin Daly—Scholarship Heroes

My first research study was an investigation in 1980 of homicide involving women either as perpetrators or victims, as my master’s thesis at Wright State University School of Nursing. I came to the topic as a nurse who was trying to identify ways to prevent the leading cause of death among young (15-34) African American women at that time (Campbell, 1986). It is now the second leading cause of death of Black women in that age group, with HIV/AIDS currently the leading cause (Black, Bsile, Breiding et al., 2010).

The literature I needed to review in 1980 on the topic of homicide of women was extremely sparse. I kept finding only aggregated homicide reports rather than studies disaggregated by gender, which obscured the profound differences that I had found in my study. Women were most likely to be killed by a husband, boyfriend or ex, and prior violence against the female by the male that killed her was the most frequent underlying dynamic. The perpetrators and dynamics were totally different for males. As I continued to pursue the literature in the field, I remember being ecstatic to see the 1982 publication by Margo (Wilson) and Martin (Daly; 1982) on male sexual jealousy. I immediately became a fan of their scholarship. I read as much as I could about evolutionary psychology, but was never totally convinced that this was the underlying mechanism of the factors associated with domestic violence and intimate partner homicide that Wilson and Daly and I and others were identifying. Still, as I was writing about the phenomena from a nursing and health care perspective, I was finding increasing evidence from Wilson and Daly and was citing their work.

Support and Encouragement

In 1987 Margo asked me to present with her and Martin, along with Rebecca and Russell Dobash (two other of my heroes in scholarship), at a symposium at the American Society of Criminology (ASC). As a newly minted PhD, I remember being totally in awe of being there but entirely welcomed by Wilson and Daly and the Dobashes, as well as by the audience and membership of ASC. I was thoroughly intimidated by what I felt was their superior scholarship, but encouraged by their support to not only continue in domestic violence homicide research but also their validation that what I had found so far was important in advancing the knowledge in the field. I also was thrilled at the spirit of collaboration, the notion that we could continue to develop this knowledge together, not necessarily in collaborative studies but in parallel and mutually reinforcing strategies, rather than in the competitive research trajectories I had observed in medicine.

It was at that ASC conference that I met Rebecca and Richard Block and that, shortly thereafter (1991), the Homicide Research Working Group was formed. I was proud to become part of the interdisciplinary (health care and nursing) component of the organization. The chapters Margo (and Martin) and I were asked to author in Diana Russell’s (1992) groundbreaking text Femicide: The Politics of Woman Killing gave both Margo’s work and mine visibility in the emerging feminist research on domestic violence. I began to feel like a legitimate scholar in the field as my work was welcomed and interrogated and challenged, as was anyone’s in the field of homicide research. I was incredibly impressed with Margo’s combination of collegiality and scholarly challenge, her careful thinking about the quality of evidence and the rigor of the research methods, her welcoming of various combinations of evidence (animal observations, mortality and/or homicide records, surveys, in-depth interview data) before “mixed methods” became an accepted research strategy, her ability to see the legitimacy of scholarship that came from frameworks other than evolutionary psychology, her willingness to continue thinking long after the working day was done and the rest of us just wanted to socialize, and most of all her welcoming of young (in experience if not years!) researchers like me into the scholarly exchange as if we were all equal.

Influence of Margo Wilson’s Work on the Danger Assessment

As I read Margo’s work through the 1990s, I was also developing the Danger Assessment (DA; Campbell, 1986) as a way for abused women to more accurately identify their own risk for homicide or near homicide. I was pleased that some of her research supported some of the factors I had on the original DA, and some of her other findings indicated additional risk factors I should consider. As I read her work I was always impressed by the clarity of her writing, especially the precision of research language, her original use of various existing data in Canada, the United Kingdom, and the United States to support theoretical propositions and her ability to tell a coherent story with research. I try to pass these lessons on to my doctoral and postdoctoral students. I was also impressed with her use of a good title in articles, another example of her precise use of language and ability to tell a story. For instance, the chapter titled The man who mistook his wife for a chattel (Wilson & Daly, 1992a) is such a coherent exposition of the male jealousy and control so often exhibited by wife beaters all around the globe that one comes away convinced that evolutionary psychology must have something to do with it, even if it is not the whole story. Till death do us part (Wilson & Daly, 1992b) is another incredibly evocative title about the dynamics of spousal homicide. Margo’s insistence on using the correct and precise term “uxoricide” for the murder of one’s wife (Wilson & Daly, 1995) was influential not only on me but also on the field. Rick Steeves and Barbara Parker (2007) went on to use that terminology as well as the legacy of Margo’s work in their important National Institutes of Health funded study of children of families where uxoricide has occurred.

I was also using the DA in my first funded research studies (e.g. Campbell, Miller, Cardwell, & Belknap, 1994), mainly because I was interviewing women outside of the safety of a shelter system and was worried that they might be underestimating their risk. I equated the need for them to have information about their level of risk with a patient in the health care system needing to have information about their diagnosis or risk factors for disease, so that they could choose their options with adequate information (Campbell, 1986). As I used the DA, I became increasingly convinced that it had clinical utility for helping abused women accurately appraise their level of danger. I talked to women about what made them afraid for their lives and how best to ask them about the risk factors on the DA. I added the “choking” item that is now known as “nonfatal strangulation” and a field of study in its own right, because of the advice of one of the women in my study and a master’s student I was advising (Glass et al., 2008; Stuart & Campbell, 1989). I was listening to my students as Margo had taught me to do.

As I continued to use the DA and to be part of the development of knowledge in the domestic violence field, I became convinced that I needed to conduct a data-based exploration of the risk factors for intimate partner homicide to validate the DA, determine a weighted scoring, and identify which other risk factors for intimate partner homicide might be important. Margo, along with Martin Daly, Becky Block, and the Dobashes, encouraged me to find the funding to conduct that kind of study using national data. We talked about the serious limitations of such a study using information from the Supplemental Homicide Reports (SHR), which was the usual practice in homicide research. We discussed how the SHR, which are based on police homicide files, carry important limitations for the study of risk factors for intimate partner homicide; information on prior arrests for domestic violence vastly underestimates actual prior domestic violence, and homicide suicides (approximately 25%-30% of intimate partner homicides of women) do not have much information in the SHR (Koziol-McLain et al., 2006). I wrote at least four major proposals to various funding agencies that were all rejected, improving the proposal each time with the help of these colleagues and tremendously influenced by Margo’s research and publications during that time. Finally, my interdisciplinary collaborative proposal “Risk factors for homicide in violent intimate relationships” (R01 DA/AA11156-01) was funded in 1996 by the joint (National Institutes of Health [NIH], National Institute of Justice [NIJ], Centers for Disease Control and Prevention [CDC]) research funding written into the first Violence Against Women Act passed in 1995. Margo was an enormous influence in the proposal and would have been a coinvestigator except for the need to keep the data collection in the United States and our budget constraints.

The study was designed as a case control investigation of risk factors for intimate partner femicide (homicide of women) with the purpose of determining risk factors over and above prior intimate partner violence, which we already knew to be the major risk factor for such murders. Therefore, our cases were women who had been killed by an intimate partner or ex-partner in 12 geographically disbursed cities, while the controls were women who had been abused by a partner in the same cities identified through a random digit dial interview (with careful attention to safety and privacy). We investigated the original DA items as potential risk factors, but also carefully chose additional potential factors to investigate. Directly attributable to Margo’s (and Martin’s) work were the testing of three new factors.

The first was a stepchild in the home (the woman’s biological child, but not the abusive partner’s), first identified as a risk factor in the article “Women With Children Sired by Previous Partners Incur Excess Risk of Uxoricide” (Daly, Wiseman, & Wilson, 1997). In our study the stepchild variable increased the odds (multivariate logistic regression) of a femicide by a male intimate partner by an adjusted odds ratio of 2.4 (Campbell, Webster, Koziol-McLain, et al., 2003). The second major factor we tested specifically was estrangement from the partner within the past year that had been substantiated in Margo’s studies conducted in Chicago, Detroit, and Canada (Wilson & Daly, 1992a). In the same multivariate logistic regression, the leaving or estrangement variable increased the odds of a femicide by 3.5 (Campbell, Webster, Koziol-McLain, et al., 2003). The third factor from Margo’s work was the highly controlling aspect of jealousy, which interacted with estrangement in our study so that the two together increased the odds of a femicide by 5.5 (Campbell, Webster, Koziol-McLain, et al., 2003). This extreme jealousy dynamic is evocatively stated by the phrase “If I can’t have you, no one can,” which Margo used as a heading in her chapter in the Diana Russell’s Femicide (1992) and I started to hear in variations over and over again from severely abused women and in newspaper and magazine reports about intimate partner femicide. I used those words as my chapter title in the same book (Radford & Russell, 1992) with Margo’s permission and also to illustrate the item in the DA on severe jealously. That item now reads “Is he (husband, partner or ex) constantly and violently jealous of you? (For instance does he say something like “If I can’t have you, no one can?”).

With the addition of these factors, the DA became far more accurate in identifying abused women at increased risk for being killed. Prior to the addition of these factors, there were no “cutoff points” or levels of danger scores for the DA, and the best accuracy determination was done by Heckert and Gondolf (2004), who found that the DA predicted risk of reassault with an accuracy of AUC = .73. With the addition of these factors, we have been able to achieve an AUC of .92 for near lethal intimate partner assaults (Campbell, Webster, & Glass, 2009). The DA is now being used in many domestic violence services agencies across the United States and Canada, as well as Portugal, Spain, the United Kingdom, New Zealand, and Taiwan. The DA also has been adapted for use by first responders in the Lethality Assessment Program (LAP) being used in most police jurisdictions in Maryland and at least nine other states. The LAP was developed in a collaborative process among law enforcement, DV advocates, and researchers under the direction of Dave Sargent of the Maryland Network Against Domestic Violence (www.mnadv.org).

Summary

In sum, Margo Wilson’s work is living on in very tangible ways. Police officers all over the United States are asking about several of the risk factors she first identified as they try to determine which domestic violence cases they are called to investigate are most likely to end in a homicide. Thousands of women in domestic violence shelters and other agencies around in the United States and around the world are doing safety planning based on the DA and the risk factors for homicide that Margo first wrote about. Margo’s foundational research in identifying the risk factors for femicide in intimate partner relationships as well as the way she taught us to conduct research has resulted in increased collaborative research on the topic. We all owe her so very much.