Neil Websdale
With contributions from Bahney Dedolph

Violence against women takes many forms and is much more pervasive, injurious, and lethal than official statistics report. Nevertheless, it is safe to say that far fewer than one percent of battered women are killed by their intimate male partners (Bureau of Justice Statistics, 1998). The question then arises: Is there something about these relationships in which women are killed that distinguish them from the vast majority of non-lethal but nevertheless abusive intimate relationships? If these lethal relationships are discernibly different, can we use these distinguishing characteristics as a means of identifying and screening out other high risk domestic violence relationships with a view to preventing their escalation to lethal outcomes? The simple answer to both these questions is no. Research into domestic homicides typically reveals these to be crimes of cumulation in which men's violence and women's entrapment seem to intensify over time. The absolute distinction between lethal and non-lethal cases is a false dichotomy; rather there is a range or continuum of violence and entrapment that underpins abusive intimate relationships. Indeed, it would be far more appropriate and useful to employ the term "dangerousness" rather than "lethality" assessment.

The research into or evaluation of lethality assessments in domestic violence cases is practically non-existent. There is little research on how lethality or dangerousness assessment tools are used, what agencies do with the scores, and how battered women are affected by the instruments. In this article, I review lethality assessment tools and the pertinent research into domestic homicide. I then critique this information and suggest that these instruments are more useful as a means of identifying future dangerousness rather than precisely predicting lethal outcomes.

PREDICTING LETHAL DOMESTIC VIOLENCE: RESEARCH REVIEW

Reviewing the Content of Lethality Assessment Tools

For the purposes of this paper, I reviewed the following assessment tools ranging from checklists to more detailed instruments that ask more searching and complex questions. Not all of these tools are widely available although in my review I consider those that seem to be more common. The selection includes: Domestic Violence Inventory (Risk and Needs Assessment Inc.); Salt Lake City Victim Advocate Program Lethality Assessment; Lethality Assessments (Arizona Coalition Against Domestic Violence); Domestic Violence Risk Assessment (Dane County, Wisconsin); Assessing Risk (Orange/Durham County Coalition for Battered Women, Inc., North Carolina); Enrollment form ADT Aware Program (origin unknown); Lethality Assessment, Durham, N.C. Police Department; Mosaic 20; Danger Assessment Instrument (see Campbell, J. ed. 1995; Chapter 5); Assessing the Lethality of Batterers (Barbara Hart, 1990: p 240-243).

The majority of assessment tools are designed to ask questions of victims rather than perpetrators, although questions could obviously be asked of both parties. Most instruments take the form of checklists, although some present a range of possible answers to questions rather than simple yes/no type responses. The more comprehensive instruments ask more searching questions about the abusive relationship, inviting discussion of how victimization patterns have changed over time and what twists there may have been in the relationship. One way of examining the tools is to see which questions or clusters of questions appear most commonly. With some notable exceptions, the most common clusters of questions are concerned with: prior victimization; batterer's drug and alcohol problems; batterer's obsessive-possessive behavior and excessive jealousy; batterer's threats to kill the victim or her children; batterer's possession of, access to, familiarity with, and degree of fascination with weaponry, especially guns; batterer's use of violence in settings outside the home (e.g. bar fights); stalking behavior; batterer's suicidal ideations, plans, threats, and past attempts; the status of the relationship in terms of whether the parties are separated, separating, estranged, or whether she is in the process of fleeing. Fewer tools seek information about: the batterer's depressive state; any history of domestic violence in the victim/batterer's family of origin; whether he has physical access to her or knows of her whereabouts, her lifestyle, and movements; whether he has threatened or harmed family pets; whether there is any history of hostage taking; and, whether she believes he is capable of killing her.

Various researchers, relying on official data and/or more qualitative interpretations drawing upon many years of field experience, have noted the presence of the following antecedents in cases of intimate partner killings: escalating domestic violence and the increasing entrapment of battered women (see Stark and Flitcraft, 1996); the separation/estrangement/divorce of the parties (Wilson and Daly, 1993); obsessive possessiveness or morbid jealousy on the part of the abusive partner (Daly and Wilson, 1988: 295; Easteal, 1993); threats to commit intimate partner homicide, suicide, or both (Hart 1988: 242); prior agency involvement, particularly with the police (Browne, 1987: 10); the issuance of protection or restraining orders against one of the parties, nearly always the male; depression on the part of the abuser (West, 1967; Lester, 1992; Buteau et al, 1993; all in regard to homicide-suicide); and, a prior criminal history of violent behavior on the part of the abusive man (Klein, 1993; Fagan, Stewart, and Jansen, 1983; Bureau of Justice Statistics, 1998). Campbell's research (1995, 1986) usefully summarizes key risk factors identified by the majority of experts in the field; the original Dangerous Assessment being intended to help battered women ascertain their own levels of risk, rather than provide absolute cutoffs. These risk factors include:

* Access to/ownership of guns

* Use of weapon in prior abusive incidents

* Threats with weapons

* Serious injury in prior abusive incidents

* Threats of suicide

* Drug or alcohol abuse

* Forced sex of female partner

* Obsessiveness/extreme jealousy/extreme dominance (from Campbell, 1995: Table 5.2)

Hart (1988) identifies attempts/threats/fantasies of homicide or suicide as key indicators of a risk of possible serious or lethal assaults. She notes that when these factors are present alongside a number of others (availability/access to/willingness to use or history of using weapons; obsessiveness; isolation of the batterer and his degree of dependence on the battered woman; rage; depression; drug and alcohol consumption; access to the battered woman) the risk is elevated.

In my recent analysis of male perpetrated intimate partner killings in Florida, I quantify some of these above antecedents by examining documents from different agencies and conducting follow-up interviews with various system players. I distinguish between the antecedents in multiple killings (47 cases, 104 victims, including children; essentially homicide-suicides and familicides) and single killings (67 adult female victims). In single killings, perpetrators are more likely to have a criminal history of violence, to have had prior contact with the police regarding domestic violence, and to be poorer. Essentially, the antecedents that emerged most prominently in both multiple and single killings are, in order of importance:

* A prior history of domestic violence.

* An estrangement, separation, or an attempt at separation nearly always by the female party.

* A display of obsessive-possessiveness or morbid jealousy on the part of the eventual perpetrator; often accompanied by suicidal ideations, plans, or attempts; depression (clinical or more rarely, psychotic); sleep disturbances (sometimes under treatment medically), and stalking of the victim.

* Prior police contact with the parties, more so in cases of single killings; often accompanied by perpetrators failing to be deterred by police intervention or other criminal justice initiatives.

* Perpetrator makes threats to kill victim; often providing details of intended modus operandi and communicating those details in some form or other, however subtle, to the victim herself, family members, friends, colleagues at work, or others.

* Perpetrator is familiar with the use of violence and sometimes has a prior criminal history of violence. Included in this group is a small but significant number of killers who have both access to and a morbid fascination with firearms.

* Perpetrator consumes large amounts of alcohol and/or drugs immediately preceding the fatality; especially in cases of single killings.

* Victim has a restraining order or order of protection against perpetrator at time of killing.

Informal and semi-formal adult domestic violence death reviews have been conducted in a number of states for the past decade (For a review, see Websdale, Sheeran, and Johnson, 1998). These analyses by various review bodies reach similar conclusions to the aforementioned research; although they employ varying methodologies and do not necessarily select random samples of deaths or examine all domestic homicides within a particular time frame. Examples of these reviews include: the Charan Investigation into a homicide-suicide in San Francisco in 1990; the Philadelphia Women's Death Review, which is a multiagency, multidisciplinary group convened as a public-private collaboration; the Santa Clara County Death Review Committee Report, published in October 1997, containing information on 51 domestic violence homicides; the Kentucky Attorney General's Task Force on Domestic Violence Crime: Domestic Violence Homicides and Suicides, October 1993, summarizes the findings on domestic violence homicides and suicides which occurred in 1991, 1992, and the first quarter of 1993; and The New York Commission appointed by Executive Order of Governor George Pataki on October 1, 1996. The New York Commission examined 57 domestic violence related deaths, and was appointed in response to a number of high-profile domestic homicides that occurred when other forms of violent crime (e.g. murder, robbery, aggravated assault) in New York were declining. (For another analysis of domestic homicides in New York, see Wilt, Illman and Brodyfield, 1997).

A Critical Analysis of the Research into Domestic Homicide and the Lethality Assessment Instruments.

The research into domestic homicide is limited because it is impossible to know precisely the characteristics of domestic violence relationships that end in death. Obviously, we can no longer ask the victim. In cases of homicide-suicide we can also no longer ask the perpetrator. Indeed, domestic violence research in general is plagued with disagreements about what is the best way to learn about tumultuous relationships, who are the best people to ask about victimization, and what are the best instruments available to measure victimization. Take for instance the antecedent "male perpetrator makes threats to kill victim prior to doing so." From my research in Florida, it is likely that such threats were present in most cases where men killed women. However, with the data available, I was only able to document the presence of this antecedent in 29.8 percent of the 47 male perpetrated multiple killings and 47.8 percent of the 67 cases where men killed intimate female partners without killing themselves or others in the process. In the final analysis, our knowledge is limited by the information reported by the involved parties prior to the homicide, and how much of that knowledge finds its way into the official record.

As far as I can see, only one prediction instrument is largely based upon a domestic homicide dataset; this is Jacquelyn Campbell's (1995) "Danger Assessment Instrument." Most of the other instruments derive from a generalized appreciation or commonsense analysis of what questionnaire writers have gleaned from the research literature on domestic violence in general. In this sense these instruments might be more accurately called "dangerousness assessments." Trying to assess the lethality indicators in domestic violence cases by working back from domestic homicides is problematic because it assumes that certain permutations, combinations, and intensities of antecedents, culminate in or indeed cause death. It might be better to assert that the various factors are associative or correlative; with the clear understanding that correlation is not proof of causation. However, it is clear from other research that lethal outcomes may also depend upon the availability of emergency medical services, especially in the first hour after a shooting or stabbing (Doerner, 1983; Mann, 1988; Websdale, 1999). Clearly, battered women in rural areas may be at a distinct disadvantage insofar as it may be more difficult to summon emergency medical assistance for them in cases of serious assaults that produce life-threatening injuries. Put differently, cases that appear to exhibit more classic and intense signs of lethality may not culminate in death because better emergency medical services are available to avert death. These cases would then only be coded as aggravated assaults and would not end up among the population of lethal cases used to generate predictive matrices.

One of the biggest problems with the lethality assessment instruments is that they purport to use "lethality indicators" that are, in fact, characteristics of many domestic violence relationships, the vast majority of which do not end in death. In other words, many relationships where there is domestic violence will exhibit these characteristics such as escalating abuse and entrapment, a pending divorce, obsessive-possessiveness, and perpetrator suicide attempts, but very few will end in death. It may be the case that the antecedents mentioned are present to a more intense degree in those cases that will escalate to death. However, I would argue that it is impossible to measure that intensity in a way that can then be translated into a standardized assessment tool. Much of the meaning for example, of a variable such as "intensity of entrapment" turns upon victims' subjective experiences of their plight, agency interpretation of the information she may be willing or feel able to share, and especially the way in which agencies solicit such information from victims in the first place.

While rare, it is nevertheless the case that domestic homicides occur when none or very few of the typical antecedents are present. I have researched the occasional Florida domestic homicide where everyone associated with the couple were shocked to hear of their deaths and where the research team could find no prior domestic violence, or agency involvement with the family. This is particularly the case in homicide-suicides where the partners tend to be more isolated and have less involvement with the criminal justice system or other service providers (Websdale, 1999). While the absence of antecedents is likely a product of our lack of knowledge, we cannot rule out that domestic homicides may occur without a long history of abuse, entrapment, and service provider involvement. It is therefore incumbent upon us not give women a false sense of security if lethality or dangerousness assessment tools indicate an apparently low level of risk of homicide. As Hart aptly puts it, "One can never really know which batterer will attempt to kill a battered woman or her children" (1988:242). In this sense there may be value in women understanding that any battering relationship might end in homicide.

Another critique of lethality assessment instruments is that they presuppose a population of women who will complete questionnaires. However, we know some groups of women are more likely to provide intimate information than others. Leaving aside the fact that most of the instruments are only written in English, it seems that assessing risk is likely to ignore large numbers of women of color, including large numbers of migrant women. However, it is also the case that Campbell's Dangerousness Assessment derives from a dataset that includes significant proportions of African-American women and has been used with great success with both African-American and Hispanic women, although not Asian women. Research suggests that women of color may be particularly reluctant to disclose personal information to advocates, police, or other criminal justice personnel. For example, one African-American woman in Florida who ended up killing her abusive partner, told me she and women like her were reluctant to use domestic violence shelters because those shelters were culturally insensitive and that the women were also reluctant to share details of their personal lives (see Websdale, 1999: 148-155). A focus group of battered African-American women in Nashville recently told me they are very reluctant to report their black male abusers to the criminal justice system because that system has historically oppressed the African-American community and that they themselves may be seen as "snitches." Similarly, research with Asian battered women reveals a great reluctance on their part to discuss their victimization, a reflection of their desire to maintain a cultural ethic that values the sanctity of the family (See Wang, 1996). It is also the case these instruments may exclude, marginalize, or be ill-suited to lesbian women at risk of lethal violence; although it must be said that relatively small numbers of lesbian women kill their intimate partners and it is not at all clear what dynamics or associative factors precede these deaths.

Perhaps the biggest objection to prediction studies in general and lethality assessments in particular is that they employ a scientific language that seeks to foretell the future. Steeped in the aura of scientific legitimacy, relying upon "clean data" that are checked into boxes on questionnaires, women's lives are stripped of their idiosyncrasies, their complexities, and subsumed into a final score or final solution that obscures the richness of their personal experience. The process of assessment may be impersonal at the very time when victims need individualized and personal care, attention, and respect. This clinical logic is consistent with a broader patriarchal cultural ethic that silences, devalues, and dismisses women's intuitive and subjective ways of knowing. Put differently, predictive studies work as part of an economy of power which involves the fast and frugal screening and classification of women to "efficiently" weed out those at greatest risk of lethal interpersonal violence with the minimum amount of effort on the part of overworked agency personnel. Care for women may take a back seat to the need to produce an assessment of her life that can be readily quantified, compared to others, related to a norm, and subsequently disposed of. However, this is not necessarily the case. Interviewers who administer these instrument do employ their own personal styles, warmer and supportive body language, and adapt the instrument to the experiences and emotional affect of women.

In spite of all these difficulties it is clear that while these instruments are not efficient lethality screens they are powerful dangerousness indicators. For this reason they can be tremendously useful to the domestic violence movement in combating domestic violence, developing more effective safety plans, listening to battered women more carefully, and reducing the incidence of serious injury, and, in some cases, death.

Usefulness of the Instruments and the Research

While it may not be possible to predict which domestic violence cases will end in death, there are many reasons for using the research into the antecedents of domestic homicide and the assessment instruments broadly consistent with and informed by the research. However, the following caveats need to be stated:

1. No instrument, however thorough, however seemingly in-tune with research findings, should form the exclusive basis for safety planning for victims. Rather, the predictive formula produces a score or risk assessment that ought only be used in concert with other information, including the intuitive feelings of advocates who have worked with women and perhaps lived similar experiences.

2. Risk assessment scores should not substitute for listening to battered women and learning about the complexities of their personal lives and broader social circumstances. Police officers who administer risk assessment tools ought not use these instead of working closely with women. Likewise probation officers and prosecutors ought not base their work with battered women on raw scores alone. Rather, raw assessment scores might be integrated into a overall non-judgmental strategy of advocacy and care.

3. Battered women ought not to be filling out these instruments in close proximity to batterers. Batterers can become enraged at the sharing of what they perceive as private and privileged family information. The practice of sending battered women home with a risk assessment instrument so that they can complete it in a "relaxed atmosphere" and then return it to a police department is dangerous. While police may feel batterers are safely behind bars this may not always be the case. If he is released under unusual circumstances and discovers a completed instrument the victim could be in grave danger.

Ideally, rather than producing a "foolproof" predictive instrument, it would be better to train those involved in providing services to battered women in the intricate dynamics of domestic violence. However, in the real world where funding is short, where many agency players do not know much about domestic violence or are hostile to learning because they think "she should just leave him," the instruments clearly have their uses. Consider the following:

* Any thoughtful instrument has the potential to enlighten those who know little about the plight of battered women. For all the concerns about lethality prediction instruments among the advocacy community, I think there are elements in a number of instruments that perceptively capture the kinds of relationship characteristics, batterer behaviors, and various system responses that researchers have documented across the country; including studies of domestic homicides in Florida, New York, Santa Clara, and Philadelphia. As such, the instruments expose players like police officers to issues that they may not otherwise consider or have been trained to think through.

* Risk assessment instruments may not only be an educational tool for service providers. They may also provide a touchstone for victims themselves as they seek to strategize about their futures and those of their children. This is not to say that battered women always minimize their victimization, or that they do not have the wherewithal to work things out for themselves. Rather, risk assessment scores and dangerousness predictions may provide yet another (and perhaps very different) lens through which to see themselves, their batterers, and their overall predicaments.

* At present, we know little about how these assessments are used and what effect they have on intervention and support services (but see Roehl and Guertin, 1998). It might be the case that the administering of these tools applies pressure to multiple service providers, encouraging them to develop a greater sense of care and caution. For example, however sensitive a criminal justice professional may be to battered women's stories, if that professional is informed that this victim has taken a legitimate danger assessment instrument, and has been assessed to be at the highest risk of lethality, then I suspect that professional may exercise greater caution and care.

* Finally, using numbers provides a shared language of risk for all those working with domestic violence cases (see Trone, 1999). Such sharing, albeit in the form of impersonal enumeration, may enhance communication among service professionals, lead to increased awareness and greater proactive interventions, and, hopefully launch further discussions about how best to curb these atrocities.

Author of this document:
Neil Websdale Ph.D.
Associate Professor of Criminal Justice
Northern Arizona University
Flagstaff, AZ

Consultant:
Bahney Dedolph
Community Programs Supervisor
Catholic Charities Community Services
bahney@cox.net

Browne, A. 1987. When Battered Women Kill. New York: Free Press.

Bureau of Justice Statistics, U.S. Department of Justice. 1998. Violence by Intimates: Analysis of Data on Crimes by Current or Former Spouses, Boyfriends, and Girlfriends. NCJ-167237 March 1998.

Buteau, Jacques, Alain Lesage, and Margaret Kiely. 1993. "Homicide followed by suicide: A Quebec case series, 1988-1990." Canadian Journal of Psychiatry. 38:552-556.

Campbell, Jacquelyn C. 1986. "Assessment of risk of homicide for battered women." Advances in Nursing Science, 8 (4), 36-51.

Campbell, Jacquelyn C. (Ed) 1995. Assessing Dangerousness: Violence by Sexual Offenders, Batterers, and Child Abusers. Sage. Thousand Oaks, CA.

Daly, Martin and Margo Wilson. 1988. Homicide. Hawthorne, NY: Aldine de Gruyter.

Doerner, W.G. 1983. "Why does Johnny Reb die when shot? The impact of medical resources upon lethality." Sociological Inquiry, 53, 1:1-15.

Easteal, Patricia W. 1993. Killing the Beloved: Homicide between adult sexual intimates. Canberra: Australian Institute of Criminology.

Fagan, J, D. Stewart, and K. Jansen. 1983. "Violent Men or Violent Husbands? Background Factors and Situational Correlates." In D. Finkelhor, R. J. Gelles, G. T. Hotaling, and M. A. Straus, eds. The Dark Side of Families: Current Family Violence Research, 49-67. Beverly Hills, CA. Sage.

Hart, Barbara. (1988). "Beyond the Duty to Warn: A Therapist's Duty to Protect Battered Women and Children." In Yllo, K. & Bograd, M. (Eds.) Feminist Perspectives on Wife Abuse. (pp. 234-248). Newbury Park, California. Sage.

Klein, A.R. 1993. Spousal/Partner Assault: A Protocol for the Sentencing and Supervision of Offenders. Swampscott, MA: Production Specialties.

Lester, David. 1992. Why People Kill Themselves. Springfield, Illinois: Charles C. Thomas.

Mann, Coramae R. 1988. "Getting Even? Women Who Kill in Domestic Encounters." Justice Quarterly 5(1): 133-151.

Stark, Evan, and Anne Flitcraft. 1996. Women at Risk: Domestic Violence and Women's Health. Sage. London.

Trone, Jennifer. 1999. Calculating Intimate Danger: Mosaic and the Emerging Practice of Risk Assessment. Vera Institute of Justice, New York.

Wang, Karin. 1996. "Battered Asian American Women: Community Responses from the Battered Women's Movement and the Asian American Community." Asian Law Journal 3:151-185.

Websdale, Neil. 1999. Understanding Domestic Homicide. Northeastern University Press. Boston, MA.

Websdale, N., Sheeran, M., and Johnson, B. 1998. Domestic Violence Fatality Reviews: Summarizing National Developments. National Council of Juvenile and Family Court Judges. Reno, Nevada.

West, D.J. 1967. Murder followed by suicide. Cambridge, Mass: Harvard University Press.

Wilson, Margo I. and Martin Daly. 1993. "Spousal homicide risk and estrangement." Violence and Victims, 8:271-294.

Wilt Susan A., Susan M. Illman, and Maia Brodyfield. 1997. "Female Homicide Victims in New York City, 1990-1994." New York Department of Health Injury Prevention Program. Guiliani, Rudolph W., Mayor, and Margaret A. Hamburg, Commissioner of Health.