ARTICLE HOMICIDE 10.1177/1088767903262446 Messing, Heeren STUDIES / WOMEN / May MUL 2004TIPLE MURDERERS Another Side of Multiple Murder Women Killers in the Domestic Context JILL THERESA MESSING University of California, Berkeley JOHN W. HEEREN California State University, San Bernardino Although both female and male multiple murderers have been studied, little attention has been paid to women who commit multiple murder. Using a national database of newspaper accounts from two archives, Lexis-Nexis and ProQuest, this article isolates an exploratory sample of U.S. women who, between 1993 and 2001, killed two or more victims during a single episode of domestic violence. These 32 cases are, in some respects, similar to cases of male mass murder but are distinct in other ways. Most significantly, these murders are well planned and the victims are largely, although not entirely, the woman’s own children. These cases are examined in terms of the social characteristics of the offenders as well as their relationship to the victims and the way in which these murders appear to develop. The article concludes that especially with the child killings, there are certain common predisposing factors and precipitating events that play key roles. Keywords: mass murder; family violence; domestic homicide Although there has been a recent spate of research on women as murderers (Jensen, 2001; Leonard, 2001; Mann, 1996) and even on women serial murderers (Hickey, 1997), there have not been significant studies of women who commit multiple murders during a single event. Most studies of multiple or mass murder are concerned with male perpetrators as they far outnumber women in this offense. We attempt to redress this neglect by looking at a sample of cases where women commit multiple murder in the context of domestic life. Given the rarity of this type of crime, our approach will be generally qualitative, centering on providing a contextual account of how these crimes develop in the lives of women. Necessarily, we will draw some parallels with men who commit multiple homicide and note the role played by assorted HOMICIDE STUDIES, Vol. 8 No. 2, May 2004 123-158 DOI: 10.1177/1088767903262446 © 2004 Sage Publications 123 124 HOMICIDE STUDIES / May 2004 variables that have been used in previous quantitative research on murder. Even though multiple murder by women occurs so infrequently and has not been the subject of sustained scholarly attention, it does offer several advantages as a research topic. It has been noted that when women kill, they disproportionately do so in the context of domestic relations (Jensen, 2001; Peterson, 1999; Silverman & Kennedy, 1988). Thus, the domestic scene is the quintessential setting for gaining insight into women’s homicidal activity. In addition, by limiting our attention to multiple murder, we eliminate more commonly found expressions of female homicide. For example, even though it has decreased dramatically in recent years (Fox & Levin, 2001), the most common kind of homicide by a woman continues to be the killing of her intimate partner. As much research shows (Leonard, 2001; Shackelford, 2001; Websdale, 1999), the typical precursor to these homicides is the sustained abuse by the woman’s intimate partner, often during a number of years. Although these situations (and evidence of prior abuse of the female killer) appear among our cases, limiting our concern to multiple murders should remove from consideration the kind of “self-help” (Peterson, 1999) or victim-precipitated murder where women finally retaliate against their abusive intimate partner. There should also be some clarity of focus in the killing of children in that two common types of such homicide—neonaticide and child abuse gone awry—are not likely to be found in our sample. Neonaticide is usually a crime committed by a very young mother who often is unmarried and wishes to cover up the fact of being pregnant by dispensing with her newborn child as soon as it is born (Lambie, 2001). This kind of offense is very unlikely to involve a second victim. Similarly for child abuse carried too far: These episodes are sometimes characterized as accidents (Alder & Baker, 1997; Lambie, 2001) and are unlikely to entail a second or subsequent victim. Moreover, some child murders have as their victims “defective children” (Daly & Wilson, 1988), those born with some physical or mental disability. The self-help implied in eliminating this burdensome offspring is also unlikely to involve a second murder. Finally, there is the frequent speculation in the research literature about whether some proportion of sudden infant death syndrome deaths may be the product of homicide by Messing, Heeren / WOMEN MULTIPLE MURDERERS 125 abusive mothers. Because these kinds of episodes of child death are not likely to include two or more victims in a single episode, focusing on multiple homicides generally dispatches this whole issue. PRIOR RESEARCH ON WOMEN AND MULTIPLE MURDER Although reference is made to multiple murders committed by women in various studies (Alder & Baker, 1997; Alder & Polk, 2001; Block & Christakos, 1995; Crimmins, Langley, Brownstein, & Spunt, 1997; Richards, 2000; Websdale, 1999), it has not been the central concern of much research. This neglect is largely the result of the small number of cases of multiple murder by women that each of these studies has encountered. At the low end is Johnson’s (1996) finding that none of the 36 cases of women who committed homicide in Tuscaloosa County, Alabama, from 1975 to 1992, involved multiple murders. At the other extreme is the Victoria, Australia, sample of maternal filicides from 1978 to 1991 that shows 7 of 32 cases as multiple murders (Alder & Baker, 1997). Most of the other studies find more or less 5% of the sample, depending on whether all murders by females or just filicides were being examined. We did locate two studies that focus exclusively on cases of women multiple murderers, but they were less concerned with the murders than with the media representations of the women (Cleeton, 2001; Heberle, 1999). As there is no previous research directed at women as multiple murderers, we take as relevant background studies those that have looked at male multiple murderers as well as the recent burst of theory and research on women’s homicide in general. Although neither body of literature is entirely parallel, it should be useful to draw comparisons and contrasts with what is known about single murders by women and multiple murder by men. Male Multiple or Mass Murder The quality of research on multiple murders or mass murder is often uneven. Probably the best sample is the 483 cases used by Fox and Levin (1998), derived from FBI Supplementary Homicide 126 HOMICIDE STUDIES / May 2004 Reports (SHR) from 1976 to 1995. However, the kind of information provided by the FBI about these offenses and their perpetrators is limited in case details. Hempel, Meloy, and Richards (1999) provided more of these details in their nonrandom sample of 30 mass murders from 1949 to 1998. They isolated 31 variables that illuminate the commission of these crimes, but the sampling problem is a serious weakness. In spite of the limitations in these and other studies, certain issues emerge repeatedly in discussions of mass murder. First, there is the social isolation of the perpetrator (Cohen, 1995; Fox & Levin, 1998; Hempel et al., 1999; Kelleher, 1997), who is likely to have experienced a recent loss with respect to employment or intimate relationships as a kind of triggering event. In addition, the murderer is likely to have a history of being rejected and frustrated in life (Fox & Levin, 1998; Kelleher, 1997), although these troubles are likely to be blamed on others (Fox & Levin, 1998; Kelleher, 1997; Palermo & Ross, 1999). The mass murderer is likely to be a White male who is somewhat older than killers who claim just one victim (Fox & Levin, 1998; Hempel et al., 1999; Palermo, 1997). He is seen as familiar with and perhaps obsessed about firearms usage (Fox & Levin, 1998; Hempel et al., 1999; Palermo, 1997). Revenge and domination are seen as the fundamental motives for the multiple slayings (Fox & Levin, 1998; Kelleher, 1997; Palermo, 1997), and the episodes often end with the killer’s suicide (Stote & Standing, 1995). A problem with many of these common elements of multiple murder is that they do not appear to be very distinctive. Many of the features noted above would seem to be characteristic of all murder and even domestic violence in general. For example, the motives of control and revenge are widespread in men’s acts of domestic violence (Hamberger & Lohr, 1997) and murder (Baker, 1999; Polk, 1994). Although it is true that most mass murderers are male, this is the case generally with all kinds of violence. Even where women seem to be equal participants in domestic violence (Straus, 1999), the reality is that women’s violence is much more likely to be reactive or defensive (Hamberger & Lohr, 1997; Leonard, 2001). Polk (1994) showed men to be greatly overrepresented among the sample of all 380 murders in Victoria, Australia, from 1985 to 1989. The FBI data used by Fox and Levin (1998) show 87% of the offenders in single murders to be male; this male proportion only rose to 94% in mass murders. Hence, being male does not Messing, Heeren / WOMEN MULTIPLE MURDERERS 127 seem to be a sharply distinguishing characteristic of mass murderers. Because males seem to be the dominant actors with respect to violence in general, it does not seem unexpected that they are even more overrepresented among the most extreme forms of violence, including mass murder. Nor does age or race seem to be crucial. According to data presented by Fox and Levin (1998), the differences between the age groups are minimal and might be even less significant if the school shootings in Paducah, Pearl, and Littleton were included in their sample (p. 436). With respect to race, Fox and Levin (1998) showed that the majority (63%) of the mass murderers are White (p. 436). Although underrepresented when compared to their proportion (75.1%) in the general population (U.S. Bureau of the Census, 2000), Whites are overrepresented as mass murderers when compared to their proportion (46.4%) as perpetrators in single murders (Fox & Levin, 1998). Another aspect of mass murderers noted in previous research was their having experienced a lifetime of frustration and rejection and blaming others for these events. Yet these are common patterns in other kinds of violence, as Fox and Levin (1998) noted. Domestic violence, according to Moore (1997), is often the result of such social conditions as poverty, meager education, and lower occupational position. Polk (1994) found that 54% of his Victoria homicide offenders were unemployed, and the vast majority of offenders were from the bottom of the economic heap. Failures in family relations are also not uncommon among spousal abusers. In fact, as studies show, many battering episodes (Ptacek, 1999) and murders (Websdale, 1999) concern issues of separation and custody, indicating rejection in intimate relations. Rejection in childhood is also typical (Dutton, 1999) and seems most apparent in the kind of brutalization that Athens (1992) regarded as preparing the way for a violent approach to life as an adult. Finally, blaming others for one’s problems seems to be a consistent element in any kind of violence (Athens, 1997; Polk, 1994; Ptacek, 1999) and makes sense of why revenge is such a central motivation for violence in general. If we consider the role of firearms in mass murder, it will be readily apparent that this is not a factor that is distinctive to this kind of offense but is a common contributor to single murders and many other kinds of crime. Zimring and Hawkins (1997) made it 128 HOMICIDE STUDIES / May 2004 clear that the wide availability of guns in the United States is a key factor in explaining the high rate of lethal violence associated with assaults, robberies, and even burglaries in this society. They do note the instrumentality effects of guns in making it easier to kill multiple victims, but this would suggest only that guns increase the efficiency of the slaughter rather than causing new victims to be added to the list of those to be killed. Hence, this ease or convenience would not appear to be important in family annihilation in the same way it might be important in a robbery where the perpetrator wants to eliminate all witnesses. Suicide, as noted, brings to a close a certain proportion of mass murders. Other episodes end through the willingness of the offender to engage in a gun battle with police, what has been called suicide by police. But, this self-destructive end occurs in single murders (Websdale, 1999) as well as in mass murders. It may be more common in mass murders, as Hempel et al. (1999) reported that more than half of their sample of perpetrators took their own lives. This is far higher than the proportion of homicidesuicides found by Websdale (1999) in his study of 230 Florida domestic homicides. Suicide, however, is likely to be a reflection of the degree of loss that has occurred in the preceding homicide (Hempel et al., 1999; Stack, 1997). Thus, one would expect that where the loss of life through homicide has been greater, as it is in mass murder, the attendant suicide would be more likely. What seem, then, to be the most distinctive features of mass murder are (a) the social isolation of the offender and (b) a significant trigger event involving a loss suffered by the offender. The isolation is likely to be profound. Hempel et al. (1999) classed 94% of their mass murderers as “loners.” Being isolated is said to remove any of the normal sources of social support and affectional bonds. In addition to this, isolation is likely to diminish the ability of the murderer to test reality. The triggering event is usually “a sudden loss or threat of a loss, which from his point of view, is catastrophic. The loss typically involves an unwanted separation from loved ones or termination from employment” (Fox & Levin, 1998, p. 439). Hempel et al. also noted the importance of crises at work and in intimate relations. Together, these kinds of precipitants accounted for about three fourths of the cases they studied. These separation crises are likely to add to the isolation already evident in the lives of these men and may be Messing, Heeren / WOMEN MULTIPLE MURDERERS 129 particularly difficult for those who define themselves as the “family breadwinner.” Women as Murderers As noted above, the most common arena for women’s homicide is the domestic sphere where they take the lives of intimate partners and children. Researchers have found the proportion of women’s within-family murders to vary from 57% (Jensen, 2001) to 80% (Ogle, Maier-Katkin, & Bernard, 1995). In general, intimate partners are killed at about double the rate of children. For example, Fox and Zawitz (2001), using SHR data from the FBI, found 424 intimate partners and 201 children (ages 0 to 5, the highest risk group) as victims of female killers in the United States in 1999. Particularly in the case of intimate partner homicides, women offenders are found to be responding to being violently victimized by their male consort both in prior episodes and in the immediate murder scene. This is confirmed by the number of prior arrests of these male victims for domestic violence, evidence of battering of the women, and the use of alcohol by the male victim just prior to the homicide (Jurik & Winn, 1990; Mann, 1996; Websdale, 1999). A significant proportion of women who murder is also likely to be under the influence of alcohol (Block & Christakos, 1995; Jurik & Winn, 1990; Mann, 1996), although some studies have found less intoximcation among women as compared to men (Smith, Moracco, & Butts, 1998). The method of killing used by women depends on the victim targeted, with most domestic partners being killed with guns and knives (Mann, 1996) and children more often victimized by beatings, smothering, and strangulation (Fox & Levin, 2001). Women are less likely to commit suicide after a murder than are men, especially in the instance of killing intimate partners (Block & Christakos, 1995; Daly & Wilson, 1988; Silverman & Kennedy, 1988). After the murder of children, suicide by the female killer is more likely, except in the case of newborns (Daly & Wilson, 1988; Haapasalo & Petaja, 1999). Confirming this general pattern, Daly and Wilson (1988) claimed that “women who resolve to die and take their loved ones with them seem never to include their husbands in their ‘rescue fantasies’” (p. 216). 130 HOMICIDE STUDIES / May 2004 During the past few years, there have been several broad theories offered that attempt to explain why women kill. Although some of these perspectives are primarily meant to illuminate only intimate partner murders (Peterson, 1999) or child homicides (Alder & Polk, 2001; Crimmins et al., 1997), they seem to be in agreement with regard to the basic social and psychological determinants of these murders. Among the social-structural factors, there is repeated reference to poverty, gender inequality, racial discrimination, urban residence, and other expressions of social disorganization (Crimmins et al., 1997; Jensen, 2001; Mann, 1996; Ogle et al., 1995; Peterson, 1999; Vollum & Titterington, 2001; Websdale, 1999). These social factors lead collectively to an environment in which women are stressed and under both physical and psychological threat. At the same time, women who murder are seen as being socially isolated, lacking in social support, and having exhausted other options short of violence (Alder & Polk, 2001; Jensen, 2001; Ogle et al., 1995; Totman, 1978; Websdale, 1999). Women find themselves within a traditional home environment where they are expected to care for their children with limited resources and little spousal assistance, perhaps even while maintaining paid employment “on the side.” This situation, set within the ideology and household reality of patriarchy, confines women to a situation in which they are expected to excel easily in the potentially challenging roles of wife and mother (Jensen, 2001; Ogle et al., 1995). Problems in the home, such as an unhappy husband or a less-than-perfect child, are likely to lead to self-blame and subsequent low self-esteem (Crimmins et al., 1997; Ogle et al., 1995; Vollum & Titterington, 2001). This broadly oppressive situation is exacerbated by acute episodes of violence, drug use, legal violations, mental illness, and economic hardship (Crimmins et al., 1997; Mann, 1996; Peterson, 1999). Although these perspectives seem promising as first steps toward understanding women’s homicide, they also appear to be too general in some respects and thinly supported by evidence in other areas. Social inequality, patriarchy, stress, and low selfesteem, for example, are pervasive features of women’s lives. Yet it is difficult to see how the omnipresence of these factors can be used to logically account for the small number of killings that occur at the hands of women. In other words, if 100 million American women experience stress and inequality, why do only about Messing, Heeren / WOMEN MULTIPLE MURDERERS 131 1,000 of them resort to homicide? One theory that recognizes this problem is Peterson’s (1999), which asks why, if women’s killing of their abusive intimate partners is simply an expression of defensive self-help, there are not more such events? Peterson’s response to this hypothetical is that even though women are reluctant to call on formal agencies of social control, they are even more reluctant to be prosecuted and imprisoned under the criminal justice system. Hence, they murder only infrequently. Ogle et al. (1995) reasoned in the reverse direction, arguing that women are typically traditional in their outlook, passively conforming to the social requirement that they be good mothers. How then to account for this “overcontrolled personality” ever committing murder? The authors suggested that women seem to erratically erupt into violence when the stress becomes overwhelming. Additional difficulties in these general theories are evident when we consider the choice of victim. Self-defense in the face of a lifethreatening partner may seem reasonable, but it has not been spelled out fully in any of these theories how this victimprecipitation perspective can account for instances of child killing. Perhaps it is premature to expect to develop a general theory of women’s homicide. In fact, some researchers (Silverman & Kennedy, 1988) have suggested that it may be more productive to use a middle-range theory strategy of isolating a variety of scenarios of women’s homicide rather then looking for common patterns and forces at work in each. Wilczynski (1997), for instance, after spelling out 10 categories of motives for filicide, suggested that these 10 could be reduced to 3 broader patterns where common risk factors are found. Interestingly, these 3 distinct patterns are very similar to what other studies (Alder & Baker, 1997; Alder & Polk, 2001; Haapasalo & Petaja, 1999) discern and can be labeled neonaticide, murder/suicide, and fatal assault. Although some (Resnick, 1969) have proposed a larger number of patterns of filicide, they seem to run the risk of overlapping with each other, making clear classification difficult. In addition to these three patterns of child killing, one other pattern of homicide by women, that committed against intimate partners, seems to be clearly delineated. Websdale (1999), for example, noted that in his Florida sample, “nearly all the male victims [of female intimate partner murders] precipitated their own demise through the use 132 HOMICIDE STUDIES / May 2004 of proximal and distal acts of violence” (p. 212). For the purposes of our analysis, the value of these 4 patterns of murder by women is that they go beyond the issue of causes and motives to indicate common elements in the victim-offender relationship, in the context of the offense, and in other experiential features of these crimes. These insights will prove useful in the discussion of multiple murder in this study. METHOD Homicide has been studied by a variety of methods, each of which has its distinct strengths and weaknesses. The FBI’s SHR seem to be most widely used for quantitative analysis of the general descriptive details of homicide. Although this data set is taken to be comprehensive in case finding, two recent studies (Duwe, 2000; Langford, Isaac, & Kabat, 1998) point to some problems in the thoroughness of the SHR. Moreover, the reports include largely rudimentary details of police investigations and give little sense of the dynamics of the homicide event (Petee, Padgett, & York, 1997). Another common method of data gathering is the interviewing of offenders after they have become part of a prison or mental hospital population. Although this provides in-depth information on the contours of the murder event, it tends to privilege the offender’s perspective. In addition, this method of case selection can lead to a sampling bias by eliminating those who have committed suicide or have not been convicted, while including such diverse offenses as neglect, abuse, and even conspiracy to compose a general category of homicide (Mann, 1996). To this issue of unrepresentative sampling can be added the problem of exculpatory accounts being solicited, especially when the information or data are gathered prior to the trial. Where whole city and state populations have been used, questions can be raised about how representative they are of the national population (Mann, 1996). In light of these difficulties, the newspaper databases LexisNexis and ProQuest were searched to gather as much information as possible about domestic multiple murder committed by women. These databases can be found in most libraries and together allow for a comprehensive review of the nation’s Messing, Heeren / WOMEN MULTIPLE MURDERERS 133 newspapers. The Lexis-Nexis database covers 141 local, regional, and national newspapers and ProQuest provides full-text access to 237 major U.S. newspapers. Eliminating redundancies, there are 298 separate newspapers contained within these databases. The District of Columbia and 47 of the states have 1 or more newspapers in these online catalogs; there are between 1 and 24 newspapers per state. Therefore, most of the major and midsize cities within the United States are represented, as are some small towns. In addition to the newspaper representation, news wires are included in these online sources. We were interested in women who had, by themselves, killed two or more victims during a single incident within the context of a domestic relationship. Langford et al. (1998) made the case for including, within the sphere of intimate partner homicide, all murders that stem from adult intimate relationships even when the intimate partner was not among the murder victims. Following the same reasoning, we extended the definition of domestic homicide to incorporate intimate partner and child murders, as well as any other deaths that developed out of the domestic context. That is, included were those women who killed in response to difficulties with a romantic partner, difficulties with related children in their care, or those who targeted their domestic partner or children in response to some stressor. Although most victims are family members, the women may have ultimately claimed the lives of strangers or unrelated persons as well. Although two victims is a lower limit than that typically used in research on men’s multiple or mass murder, it represents a pragmatic balancing of the heightened seriousness of the offense and of our ability to find an adequate sample for analysis. As previously mentioned, the inclusion of cases in which there are multiple victims effectively rules out the possibility of an unintended death. However, with a higher victim threshold, the number of cases would have been dramatically decreased and this would have limited our ability to discern plausible patterns. Although murder by women is rare compared to that by men, this rarity becomes dramatically amplified as the number of victims per episode increases. For example, women commit approximately 13% of murders involving a single victim, but where there are two or more victims, the women’s share is cut by more than half, composing only about 6% of such crimes (Fox & Levin, 1998). 134 HOMICIDE STUDIES / May 2004 Within the Lexis-Nexis and ProQuest databases, the search terms murder and domestic were entered to cast the widest possible net for locating potential cases. These search terms were chosen after trying several alternative terms (such as homicide and family) that appeared to generate fewer stories. The resulting abstracts of newspaper articles were then carefully perused for women perpetrators fitting the criteria established. Once a case was identified, further searching was performed using the perpetrator’s name. As many cases went to trial and spanned several years, this method was useful in gaining as much detail as possible. It seemed important to begin the search in the present to include the most up-to-date cases; the databases were then searched in reverse chronological order until a sufficient sample size had been reached. The resulting database includes cases from across the United States occurring between 1993 and 2001. Based on our perusal of all relevant articles returned, we created homiciderelated categories and entered the information into an SPSS data file. Throughout this procedure, articles were printed off of the newspaper databases, collected in files, and summarized in cover sheets. Multiple murder, especially when committed by a female, is not a common event and it is, therefore, reasonable to assume that these newspaper databases will have some mention of this type of incident, even for the geographic areas not specifically covered by the database. In fact, newspaper coverage was often quite detailed for these extraordinary events. Although mass murders that occur in public places do attract much greater media coverage (Petee et al., 1997), these domestic homicides are also the focus of considerable media attention. In addition to interviews with the police and other experts on crime and mental illness, reporters frequently spoke to neighbors, acquaintances, and survivors. This allowed for a more elaborate depiction of the events leading up to and surrounding the killings themselves. Through their information gathering and story composition, reporters seemed to maintain an objective stance, often appearing to be unconcerned for whether the details reflected poorly on any of the participants, including the victims. From a journalist’s viewpoint, the coverage was aimed at providing a balanced and thorough account meant to satisfy the inevitable question among readers of why this event occurred. Messing, Heeren / WOMEN MULTIPLE MURDERERS 135 Nevertheless, much like other sources of data, questions can be raised about this method of data collection. For example, at times we found details were missing that would have been useful for teasing out patterns across cases (see Richards, 2000). As with any secondary data source, there is the possibility that the data or aspects of the data are inaccurately reported. A reporter may have misstated facts or arrived at misleading conclusions, which we subsequently took as factual and analyzed for the purposes of this research. Often, however, multiple stories or multiple news sources included the same details, increasing the probability of their accuracy. Moreover, it is possible that although this type of murder was committed, newspapers elected not to provide coverage of this event, perhaps due to its being “crowded out” by the simultaneous occurrence of “larger” news stories. Perhaps in some cities, these multiple murder stories were not seen as sufficiently newsworthy to warrant coverage. Nor can it be claimed that the geographic coverage of the Lexis-Nexis and ProQuest databases is exhaustive. In addition, the relatively short (9-year) time span covered by this research is a limiting factor. It is difficult to know whether the causes and contours of women’s domestic murder have changed in recent years; during a longer time span, the patterns may have differed. Although the aforementioned problems lead us to be cautious in the interpretation of findings, it should be noted that we did not select a sample from the newspaper database but included every instance found that met our criteria. Thus, we do feel this is an adequate portrait of women’s domestic multiple homicide as it is played out currently within the United States. Each method of data gathering has its inherent difficulties, but searching newspaper databases allowed for a balancing of detailed case information, representative sampling, and a diversity of perspectives on the events. RESULTS Our newspaper search produced 32 cases of women who, between 1993 and 2001 in the United States, had committed multiple murder in the context of domestic relations. Because we searched a 9-year period, these multiple murders are certainly 136 HOMICIDE STUDIES / May 2004 TABLE 1 Number of Victims by Victim-Offender Relationship Number of Deaths by Relationship Case Type By VictimOffender Relationship Filicide (20) Familicide (6) Other combinations Ex-intimate partner and other (2) Own child and other (1) Miscellaneous (3) Own Children (n) Intimate Partner Other Relative Nonrelative Number Injured 57 12 — 6 — — — — 5 1 — 1 — 1 — — — 1 2 3 — 7 3 3 — quite rare, with just more than 3 events per year. To provide some sense of the incidence of these cases, we can note that during our newspaper search, we were also gathering information about men who committed multiple murder within the domestic sphere. However, we used three deaths as the criteria for the men’s killings, rather than the two established for women. For the 1993-2001 period searched in U.S. newspapers, we found 141 cases of “domestic” multiple murders by men and only 16 cases of women who committed three or more murders in this same context. Thus, of the total number of cases where three victims were murdered, women perpetrators represent about 10% (16/157) of these episodes. Although this is higher than the 6% found by Fox and Levin (1998), it is likely the result of the exclusive focus on domestic situations, where women’s murders are more frequently found. Description and Contributing Factors Table 1 shows the distribution of these cases by type and the number and kinds of victims associated with each category. The largest category was the killing of children, filicides, that included 20 cases, with 57 child deaths and 5 injuries. Most of the women (14) killed all of their (biological) children, but some children were spared the maternal attack and some survived it. As with murder in general, the injuries could just as well have been deaths, except for such contingencies as poor shooting, rapid emergency care, Messing, Heeren / WOMEN MULTIPLE MURDERERS 137 and so on. Familicides accounted for another 6 cases, encompassing 6 intimate partner deaths, 12 child killings, and 1 injury. Among the 6 couples, 3 were married and living together, 1 was separated, and 2 were in cohabiting relationships. Beyond these familicides, it seems appropriate to add that 2 former intimate partners were also attacked in these multiple murder cases. In both cases, the ex-spouse had established a new romantic relationship, and the woman offender attacked the new couple. The deaths included 1 of the ex-spouses, both of the new intimate partners, and a friend; the injured included the other ex-spouse and 2 additional people. The last 4 cases are more complicated, with 1 involving a mother who killed her own child and her (the mother’s) younger brother who was 11 years old. Fire was used as the means of killing in this case and led to the injury of 3 others, including some of the mother’s other children. Also among the miscellaneous category, there was 1 case of a grandmother involved in a custody dispute with her son-in-law, who killed her 2 grandchildren. In addition, there was an older mother who was trying to show support for her son in his dissolving romantic relationship. She located the apartment house where her son’s former lover lived with a new man and set it on fire. The only victims in this case were the 5 strangers in the apartment building who were unintended deaths in the mother’s plan of revenge. The final case was a wife who killed her husband’s girlfriend and the child of that extramarital affair. The case was distinctive in that she did not target her husband. Half of our sample killed only the 2 people that represented our lower limit for inclusion, and the highest death toll for any one event was 6. Taking all cases together, the average number killed in these offenses is almost 3 (2.8), and 10 lives are claimed each year in these multicides by women. The number wounded ranged from 0 to 3, with the vast majority of cases (78%) having no wounded. With so few collateral injuries, the women were highly efficient or controlled killers. This is contrary to male murder sprees that often include nonfatal injuries (Dietz, 1986; Hempel et al., 1999). Of the 7 women that left people wounded, 4 used fire as their weapon and 3 used guns. One of these women, Patricia Reager, was intent on murdering her family and herself when the gun that she was using jammed. She called 911 when the gun stopped firing, having already murdered her husband and one 138 HOMICIDE STUDIES / May 2004 daughter and leaving another daughter alive with a bullet wound to the head. In statements to police, she said that the shooting was a “botched job.” She had established plans, but they were upset by an unexpected contingency. More typically though, the murders were carried out as planned. Both the paucity of injuries and the infrequency with which strangers were killed seem to reflect that planning. Hempel et al. (1999) suggested that these patterns are typical of nonpsychotic (male) mass murderers. Another way to understand the efficiency of our women murderers, however, is as an extension of their roles as wives and mothers. Although their notion of “caring” is highly misaligned, the forethought and preparation that goes into the actual homicide shows that even as they are committing life-ending acts of murder, the women are still, in their own minds, caring for their families (Alder & Polk, 2001; Daly & Wilson, 1988). These murders, therefore, seem to be calculated and are very rarely committed on the spur of the moment. This is in sharp contrast to the traditional view of male mass murderers as violent, out-of-control madmen (Palermo, 1997). Although thought to be common among male perpetrators, this lack of control is uncharacteristic of the women in our sample. An apt instance is Andrea Yates, who calmly and systematically drowned her five children in the bathtub. She then laid them side by side in the bedroom, under a sheet, before calling her husband and the police. Similarly, there is Sandi Nieves, who convinced her four daughters to have a sleepover in the kitchen, turned on the gas oven, and started the fire that killed them all. One other interesting pattern that emerges from an overview of our sample has to do with the choice of victims. As noted earlier, when women in general commit murder, about half of their victims are intimate partners. Child victims are killed at perhaps half of this intimate partner rate, making them less frequent targets of women’s homicide. These proportions are sharply reversed with our sample of women multiple murderers. Of our cases, 63% are filicides, with the murderer’s own children, then, representing 78% of the total number of deaths. Only 8% of victims were intimate partner deaths. However, these killings are notable in that they occurred in the course of 6 familicides. Other researchers (Daly & Wilson, 1988; Steffensmeier & Allen, 1996; Websdale, 1999) have remarked on the extreme rarity of these events, but in Messing, Heeren / WOMEN MULTIPLE MURDERERS 139 our study this annihilation of one’s whole family did represent 19% of our cases. Although the reasons for these killings will be discussed below, it can be added here that 5 of these 6 cases, following the common male pattern, ended with the suicide of the murderer. Table 2 provides an overview of the details of the murders and some characteristics of their perpetrators. With respect to age, most of the women were in the middle child-rearing years, with only 1 woman younger than 20 and 1 older than 50. The mean age for the sample was 34.4 years. Although race and ethnicity were known for only 30 cases, half of these women were non-Hispanic White and 23.3% were Hispanic. Asians and Blacks each made up 13.3% of the sample. These patterns are interesting in the way that they contrast to the rates for these racial and ethnic groups when women kill just their domestic partner or single children. Several studies (Block & Christakos, 1995; Johnson, 1996; Mann, 1996) report, for example, that African American women murder their intimate partners at considerably higher rates than that of other women. In our study, they cannot even be seen as significantly overrepresented among these multiple murder offenders. Similarly, non-Hispanic Whites would appear to be somewhat underrepresented when considering their proportions in the population. On the other hand, when we consider the proportions of Asians (3.5%) and Hispanics (12.5%) in the general population (U.S. Bureau of the Census, 2000), these two groups are overrepresented in our sample. This is contrary to earlier research (Riedel & Best, 1998) that finds both Asian and Hispanic (Latina) women commit few intimate partner murders when compared to those committed by men in these same ethnic groups. As explanation with respect to Latinas, Riedel and Best (1998) interpreted this as showing that due to extended family networks and continuing immigration, tradition and patriarchy have ongoing importance in Latina/Latino culture. It seems that similar arguments could apply to traditional intimate partner relations in Asian cultures. Although this traditionalist interpretation of the sex ratio of killing is interesting, we will indicate below how traditionalism plays a different role in women’s multiple murder. The weapon used to commit the crime varied widely across cases. However, guns were used most often, with almost 50% of the murders carried out by this means. Some earlier research 140 HOMICIDE STUDIES / May 2004 TABLE 2 Offender and Offense Characteristics (All Cases) Variable Offender age at time of the murders Under 20 20 to 29 30 to 39 40 to 49 50 and older Race and ethnicity Black White Hispanic Asian Primary homicide method Gun Knife Strangulation/smothering Drowning Fire Combination Offender abuse during childhood Physical Sexual Physical and sexual Spousal abuse By the offender’s partner By the offender Combination Reported as none Substance use Yes No Seasonal differences Fall Winter Spring Summer Regional differences West Midwest South Northeast Precipitating event Role loss Breakup Affair/another woman Custody dispute Lost autonomy Frequency Percentage 1 9 13 8 1 3.1 28.1 40.6 25.1 3.1 4 15 7 4 13.3 50.0 23.3 13.3 15 4 4 2 6 1 46.9 12.5 12.5 6.3 18.8 3.1 1 3 1 3.1 9.4 3.1 3 1 5 14 13.1 4.3 21.7 60.9 6 9 40.0 60.0 11 7 8 6 34.3 21.8 25.0 18.8 13 5 9 5 40.6 15.6 28.1 15.6 22 6 7 5 4 68.8 27.3 31.8 22.7 18.2 Messing, Heeren / WOMEN MULTIPLE MURDERERS 141 TABLE 2 (continued) Variable Psychological disturbance Depression/suicidal intent Deluded thinking Outcome Suicide Attempted suicide Trial judgment Death Prison Insanity Frequency Percentage 10 4 6 31.2 40.0 60.0 11 8 35.5 8 9 3 25.8 29.0 9.7 (Block & Christakos, 1995; Riedel & Best, 1998) on women’s methods of murder finds guns to be a less common choice than other means, especially when the victim was a child (Chew, McCleary, Lew, & Wang, 1999; Silverman & Kennedy, 1988). A central reason that guns are rarely used in these cases is the defenselessness of young children; firearms are simply not needed to kill them, as Fox and Levin (2001) noted. In addition, the wide use of guns by our sample participants is somewhat surprising in light of the numbers of suicides and suicide attempts (59%) among our offenders. When women choose to take their own lives, they are much less likely than their male counterparts to use guns (Killias, vanKesteren, & Rindlisbacher, 2001). So, in spite of the high proportion of child killing and suicide, the efficiency of guns in multiple murder (Fox & Levin, 1998) seemed to be the overriding consideration. As one instance, we cite 19-year-old Miguelina Estevez who shot each of her 2-year-old triplets in the head, killing them, before shooting herself. Yamileth Greene also shot and killed her two young children while they were sleeping and then shot and killed herself. Precipitants Although it would be too much to expect news reporters to delve deeply into the upbringing of the offenders in their crime stories, it was interesting that little childhood abuse was reported as having been experienced by our female killers. There were only 5 cases where such physical and/or sexual abuse was noted in the 142 HOMICIDE STUDIES / May 2004 news reports. One of the reasons for the paucity of such information may be the large number of cases that ended in suicide or a determination of mental illness. Issues such as past abuse experienced by offenders are most likely to emerge as part of a defense strategy when a case goes to trial, so suicides and judgments of mental illness short circuit that possibility. Another obvious reason is, as noted, the inability or lack of concern of the reporter to gain access to such information, especially when the main focus of news stories is likely to be on the immediate crime scene, the victims, and the aftermath of these killings (Websdale & Alvarez, 1998). Apart from being abused as children, we also did not find much evidence of spousal abuse in the offenders’ intimate relationships prior to the murders. Only 9 cases included such documented family violence, with another 14 cases reported as having none. Although this is certainly another element a news reporter might omit for reasons similar to those noted above regarding abuse of the offender during her childhood, attempts were clearly made to document this family violence. For instance, with 1 mother who killed her three children, the newspaper reported that “police had no record of previous calls from that address or reports involving the family. No records of criminal charges could be found in San Mateo or San Francisco counties” (Zamorra & Hatfield, 1998, p. A1) In no way can this inquiry be seen as exhaustive or definitive, but the lack of such evidence in these cases may represent a difference between our cases and single murders. Similar cautions need to be exercised with respect to substance use just prior to the killings. Information on more than half the cases is missing, but that which is available suggests that such drug or alcohol consumption was a possible contributing factor in only 6 cases. For example, Socorro Caro, described as a devoted stay-at-home mother, volunteer, and churchgoer, knew that her marriage was “on the rocks” and had begun drinking heavily and taking prescription drugs to combat her growing feelings of depression. Alcohol and drugs have been shown to be important factors in a substantial proportion of single murders committed by women (Block & Christakos, 1995; Mann, 1996), so the relative unimportance in the present cases once again underlines the observation that these murders, in large part, were committed with considerable forethought. Messing, Heeren / WOMEN MULTIPLE MURDERERS 143 When we look at seasonal and regional variations, the patterns do not appear to suggest any extraordinary skewing of cases. Most of the murders occurred in the fall, with October being the month with the highest number of episodes. Regionally, the West and, to a lesser extent, the South were the locations of most of these crimes. California, in particular, stood out as being the state with the largest contingent of women multiple murderers. Although a few studies of single murders have looked into the precipitating event that contributes to the occurrence of murder, this concern is a staple in understanding multiple or mass murders (Fox & Levin, 1998; Hempel et al., 1999). Our categories of precipitating events (see Table 2) may seem to be very similar to one another, even overlapping. However, what was sought in each of these episodes was the immediate event that seemed to ignite the killing rampage. In general, we found two broad types of triggering events, with several discrete subcategories in each. The first and most common category of events (almost 70% of cases) involved changes in the domestic situation of the offender. These changes represent a loss of fundamental roles for the women in our sample. Primary here were cases in which the status of wife was stripped from the women, and they were no longer able to define the self in relation to an intimate partner. Six of these cases involved the breakup of a relationship, most often through divorce. Maria Montalvo’s situation is illustrative. Her husband had decided to leave her and had, as a result, gone to stay at his parents’ home. She followed him there, parked in their driveway, and ignited her car with gasoline. Although she apparently intended to kill herself, only her two toddlers perished in the flames. Another 7 cases were similar in that the woman was responding to the loss of her intimate partner, but this loss was symbolized by the discovery of the partner’s affair or new relationship. For example, Marilyn Lemak saw her husband with his new girlfriend and realized that she and their three children were no longer the most important things in his life. Wishing to relieve him of the responsibility of caring for them, she killed her three children before attempting suicide. Five of the cases entailed the loss of the maternal role, with custody problems being the trigger for the fatal violence. Terri Esterak had already lost custody of her three daughters and was at the end of a month-long visitation 144 HOMICIDE STUDIES / May 2004 period. The day before she was to return the girls to their father, she fatally shot all three and then herself. Finally, there were the 4 cases that we label lost autonomy. Here the problem was more a matter of child care responsibilities interfering with other life prospects, such as a new romantic relationship or a desire for independence. It seems that legitimately giving up child-rearing responsibilities was potentially too stigmatizing for these women, so they tried to resolve their personal crises by killing the children. These women are different from the others but similar to neonaticide cases, in that suicide was not considered. Indeed, 2 of these women attempted to blame the deaths on others so that they could be free from the responsibility of children and of crime at the same time. Another feature of these murders that seems to set them apart from many of the others in our sample is the lack of resignation about their situation. For here what is evident is that the offender seems to have a plan for reclaiming her future. Typical of this subcategory is Susan Smith, who drowned her two boys and initially attempted to claim that a stranger committed the crime. Her motivation appears to have been in beginning a new romantic relationship by eliminating the responsibilities of motherhood. What seems most remarkable about these cases is that the crucial events in triggering the murders revolved around the woman’s position in the family. Whether losing one’s intimate partner or children, or the prospect of having another romantic relationship, these women’s lives heavily revolve around family and domestic relationships. There were no precipitating events such as loss of job or financial setback that were crucial factors. Where any such considerations came into play in these cases, it was only secondary to the loss of a relationship to a male breadwinner. In studies that have investigated male multiple murderers (Fox & Levin, 1998; Hempel et al., 1999), financial and occupational losses are commonly found to be key precipitating events, even when the killer strikes out exclusively at his family. It seems clear, then, that a fundamental source of identity for men is lost or damaged through setbacks that occur outside of the home. However, for women who commit multiple murder, the sphere of loss is not unexpectedly located among kin or intimate relations. Nevertheless, the losses suffered by women seem to entail just as devastating a blow to social position and identity as those Messing, Heeren / WOMEN MULTIPLE MURDERERS 145 experienced by men. Lisa Peng’s case shows the degree to which one’s life may become unraveled. At 44 years old, she was the wife of a wealthy businessman and the mother of his grown children. On discovering that a child had been born to her husband’s 25-year-old mistress, Lisa Peng murdered the younger woman and her 5-month-old child. Replaced as an intimate partner, her children were grown and she had even lost the ability to claim that she was the only mother of her husband’s children. The role loss left no part of her life unscathed. The other main category of precipitants has less to do with relational difficulties than with disturbances at the level of the individual. Once again, however, we note the overlap between triggering events, as these more psychological problems are not without a context in the interpersonal relations of the offenders. The one difference in these last 10 cases is that the psychological aspects come to outweigh the interpersonal in terms of initiating the violence. Our two subcategories of psychological disturbances correspond to the common psychiatric distinction between thought and mood disorders. In the first subcategory, there were 4 cases of depression or suicidal intent that led to multiple homicide. In Christina Riggs’s case, she had contemplated taking her own life on several occasions but was constrained by thoughts of her two children’s fates. Ultimately, she realized that she could take her children’s lives as well. Her suicide note said of the children, “I did not want them to be a burden on mom or anyone else or to be separated” (Frye, 2000, p. A1). The other subcategory suggests distortions of thinking on the part of the killer. Most often this involved the delusion that the murder was in the best interest of the child. A case in point is that of Patricia Reager. During the O. J. Simpson spousal murder trial, Reager thought that Simpson’s attorney, Johnny Cochran, was going to send people to her house to torture and kill the family. She believed that the only way she could save them from this fate was to kill them herself. These psychological disturbances in thinking seem quite consistent with what recent research (Link, Monahan, Stueve, & Cullen, 1999) describes as a configuration of symptoms of mental illness that typically lead the offender to strike out against persons perceived as harmful. In other words, although the delusions are internal events or precipitants, they are real in the violent consequences they produce in the murderer. 146 HOMICIDE STUDIES / May 2004 Outcomes With respect to outcomes of these cases, more than one third (35.5%) of the women took their own lives after their murder spree. Another 25% attempted suicide but failed and were brought to trial. Of these attempted suicides and the other cases brought to trial, only 3 (9.7%) were determined to be mentally ill and were sent to mental health facilities in lieu of prison. About one fourth of the women (25.8%) were given the death penalty for their killings, and 29% were found guilty and given prison sentences, most commonly life terms. Some of those judged sufficiently sane to be put on trial seemed to have obvious symptoms of psychiatric disturbance. For example, Andrea Yates, with the birth of her fifth child, was suffering again from recurring postpartum depression that apparently had psychotic elements. Her defense asserted that she believed her children were not developing normally, that she thought she was possessed by the devil, and that her children would be better off dead than having her as a mother. These delusional thoughts do seem to have some explanatory power in her murdering all her children. In another instance, Cheryl Burns set fire to the mobile home that she shared with her children and several other relatives; she had been previously diagnosed as bipolar and was suffering from auditory hallucinations. Both of these women were put on trial and given lengthy sentences. These decisions are a marked contrast with the decisions made in European criminal justice systems in similar circumstances (Haapasalo & Petaja, 1999). Child Killings If we consider the number of victims alone, the greatest harm inflicted through these multiple murders is in the children whose lives are lost. To take a closer look at these losses, we would like to narrow our focus to the filicide and familicide cases in which this occurred. These 26 cases entailed the deaths of 69 children who were the offspring of the offenders. In Figure 1, we show the ages of the children who died in these episodes. The most remarkable aspect of this representation is how it completely reverses the age trends shown in other studies of the victimization of children through homicide (Chew et al., 1999). Generally, according to the Bureau of Justice Statistics (2002), the homicide victimization rate Messing, Heeren / WOMEN MULTIPLE MURDERERS 147 12 10 Frequency 8 6 4 2 16 Years 15 Years 14 Years 13 Years 12 Years 11 Years 10 Years 9 Years 8 Years 7 Years 6 Years 5 Years 4 Years 3 Years 2 Years 1 Year Infancy 0 Age of Child at Time of Murder Figure 1: Age of Murdered Children in Filicide and Familicide Cases for neonates is relatively high at about 5 per 100,000, but that rate begins to drop dramatically once infants get past the first few days and weeks of life. The rate of victimization approaches its trough by age 5 or 6 and continues near this low rate until the child moves into adolescence, when it becomes more common to be killed outside the domestic context. This victimization rate from age 5 to 15 is so low that it is not matched again until a person reaches old age. These trends do not correspond to those found within our sample. Because of the selection criteria that at least two deaths occur in our cases, neonaticides were unlikely. As Figure 1 shows, the highest rate of victimization is at age 2, and the rate remains relatively high until age 9. Surprisingly, there continue to be deaths up to the age of 16 at the hands of these mothers. The generally lowered prospects for suffering homicide after one emerges from infancy are not seen among these victims. Studies of child homicide (Alder & Polk, 2001; Richards, 2000; Wilczynski, 1997) may lead one to think that child abuse is a significant factor in these deaths. However, the evidence offered by the news accounts we found does not provide much support for this notion. As Table 3 shows, only 4 cases were reported as having documentation of previous child abuse. Many more than half of the 26 cases reported that no prior child abuse had occurred. 148 HOMICIDE STUDIES / May 2004 TABLE 3 Offender and Offense Characteristics in Cases of Filicide and Familicide Variable Child abuse Physical, by the father Physical, by the mother Physical, by both parents Neglect Reported as none Family and work roles Married Stay at home Working Separated Stay at home Working Divorced Stay at home Working a Number of children Two Three Four or more Frequency Percentage 1 0 0 3 17 3.8 0.0 0.0 14.3 81.9 7 5 2 10 5 5 5 3 2 7 11 7 71.4 28.6 50.0 50.0 60.0 40.0 28.0 (64.1)b b 44.0 (25.2) b 28.0 (10.7) a. Excluding those mothers who gave birth to only one child, both in the sample (n = 25) and in the general population. b. Percentages in parentheses from U.S. Bureau of the Census (2000). Although it is necessary to stress, once again, that reporters may not provide sufficiently deep investigation to uncover instances of such abuse, it does seem that because our sample is composed of multiple murderers, it is not likely to include cases of murder where child abuse simply got out of control. A more promising avenue for understanding these cases is in the relationship between the mothers and their children in these domestic situations. One difference we see within this relationship in our sample is how early the women in our sample generally had children. Although the mean age of these mothers’ first birth is 23.5, this does not adequately show the degree of difference between our mothers and mothers within the general population. As is apparent from Figure 2, the general population gives birth at a much later age than did these women who murdered their children and families (U.S. Bureau of the Census, 2000). Although only 9.5% of the general female population has given birth by the time that they have reached the age of 20, more than Messing, Heeren / WOMEN MULTIPLE MURDERERS 149 40 Sample General Population 38% 35 30 26.9% 20.7% 24.1% 20 13.8% 15 16.1% 19.4% Percent 25 0% 3.4% 5 1.1% 8% 9.5% 10 0 Under 20 20 to 24 25 to 29 30 to 34 35 to 39 40 and Over Figure 2: Mother’s Age at the Time of Her First Child’s Birth SOURCE: General population figures from U.S. Bureau of the Census (2000). one quarter of our sample of 26 women were mothers before this age, thus representing almost 3 times the percentage of teenage mothers in the general population. Between the ages of 20 and 24, 26.9% of the general population gave birth for the first time whereas 34.6% of our sample did so. Differences are negligible for the next higher age groups, but it is notable that 9.1% of the general population of mothers has their first birth after age 35 whereas only 3.4% of our sample did so. 150 HOMICIDE STUDIES / May 2004 In addition to the age at first birth, we can note the number of children that each of these women had. Although our sample is skewed because it required at least two deaths for the inclusion of a case, it still appears that these mothers had more children than is typical in the general population. Obviously, our filicidal mothers included none who had just one child, and only 1 of our 6 familicides involved a single child and the significant other as victims. If we limit our comparison to U.S. families with two or more children, this makes our child-killing cases parallel and allows us to see any contrasts. Among the general population with two or more children, 64% had just two children, a far higher proportion than the 28% of our sample in this same category. About one fourth of the U.S. population has three children, compared to 44% in our sample. Four or more children are found among only 11% of the U.S. population, but this family size was found among 28% of our cases. This comparison suggests that this larger size is a distinctive element of our families. Illustrative of these trends of early births and larger numbers of children is Khoua Her, a mother of six by the age of 24, who had her first child at age 13. Miguelina Estevez had triplets at the age of 17, whereas Megan Hogg had her first child at 18 and two more before the age of 23. It is undeniable that these kinds of child care demands are certainly fulfilled regularly by many women without any resulting murder or even child abuse. In our sample, however, we are interested in how these elements may interact with others in bringing about a potentially violent outcome for the children of the offenders. One other element in the lives of these women is their status as stay-at-home mothers. In conjunction with the early births and larger number of offspring, this has to be considered as contributing significantly to their life pattern. As Table 3 shows, most of the women in our sample on whom we have such information were stay-at-home mothers. Overall, this was so for 14 of the 22 women (64%); this included 50% (7 of 14) of those in the sample who were separated or divorced. The latter proportion is fairly surprising as, in these cases, the spouse’s income might not be as helpful or dependable in meeting household expenses. We see this domestic status as playing several roles with respect to the lives of the women in our sample. First, although the proportions are not greatly different from the general population, having the status of stay-at-home mother is a further indicator of traditionalism Messing, Heeren / WOMEN MULTIPLE MURDERERS 151 among the women in question. Looking at the larger picture, we see that these women had children earlier, had a larger number of children than average, and stayed at home with their offspring. When this domestic status was disrupted, these women felt keenly this loss of status and rather than trying to recreate their lives and homes, chose to end their anguish through violence. A parallel, then, can be drawn to mass murders by males where the forced severing of ties to work and/or family are typically the triggering force behind their spree. Being a stay-at-home mother is also likely to result in social isolation from adult relationships. Although this social isolation is the product of different forces in the lives of each of these women, it does appear to be a pervasive presence. A typical pattern is that of an immigrant woman who marries young, has children, and although living in the United States, fulfills the traditional domestic roles expected in the country of her birth. When the family equilibrium is disturbed, the physical distance from her native culture and her family may assure that the traditional social supports are not present to help the woman through the crisis. An instance of this type of isolation is Nirmala Devi Katta, who left her family home in India to come to the United States to get married. When she discovered that her husband had, and continued to have, numerous affairs with other women, she had only his family in the United States to which she might turn. Although she begged her parents to let her return to India, they would not allow it, as it would be a disgrace to the family. With no meaningful social support and a disastrous family life that she could not escape, she eventually resolved her crisis by killing her husband and children. Overall, it was possible to identify isolating factors in more than three fourths of the cases in our sample. Child care responsibilities, restriction to domestic duties, and status as an immigrant all seemed to be significant details that pointed to the isolation of these women multiple murderers from adult relationships. Even mild states of mental illness that set one apart from others in terms of feelings and perceptions also operated in ways that tended to separate the women from those around them. Andrea Yates’s unstable mental state was discounted by those apparently closest to her, separating her emotionally from even her husband. In other instances, it was a configuration of elements distinctive to 152 HOMICIDE STUDIES / May 2004 the woman’s situation that served to cut her off from sources of support. For some of the single working mothers, feelings of isolation seemed to be more the product of their attempts to balance work and family obligations. One such woman, Dellfrances Bennett, was described as “always hurried” to meet child care and work responsibilities and never able to have time for friends or acquaintances of her own. Another woman, Tina Marie Cornelius, had so few intimate relations that her disappearance was recognized only when her children’s bodies were discovered in a creek bed and the police feared that she had been a victim of violence as well. In fact, she had fled her home in the wake of the filicide. The lack of social support in the lives of these women, regardless of the form taken, is a double-edged sword; it both creates feelings of disjuncture from reality and, at the same time, separates them from those who may be able to recognize their plight and offer assistance. DISCUSSION To an extent, these women multiple murderers seem to resemble the portraits that have been produced of male mass murderers. Most prominent as contributing factors are the social isolation we found among our sample participants and the crucial role played by certain precipitating events. Insofar as these women have come to center their lives in the domestic sphere, their acts of violence are appropriately seen as expressions of their desperation at losing this domestic status. Rather than wait out the crisis and their likely adjustment to a newfound social reality, they resort to violent means to resolve their personal troubles. Beyond the isolation and the triggering events, one other element that seems similar is in the realm of beliefs or ideology. In male mass murder, one often sees clear indications of the patriarchal ideology that men “own” their families and have absolute control over them. When faced with losing that control, the patriarchal response is to use violence, even to the extreme degree of mass murder, to reestablish the dominant male position. In this sense, men often feel that they are entitled to kill their families. In our cases of women who commit multiple murder, a parallel kind of belief seems to be operating. In many instances, these women Messing, Heeren / WOMEN MULTIPLE MURDERERS 153 believed that they had some unassailable proprietary right to their children’s lives. Unlike male patriarchy, this typically did not extend to the entire family, but the women often believed unquestioningly that, as mothers, they owned their children (see Daly & Wilson, 1988). Among our cases, this mentality was evidenced by Denise Pierce who, in a suicide note written to her husband, proclaimed, “I just can’t live without you and I couldn’t leave my kids behind” (Cannizaro & Turni, 1996, p. A1). Another of our murderers, Khoua Her, expressed a similar idea when she said that although committing suicide, she needed to protect her children and thus could not “leave them behind.” This notion finds some legitimation in various cultural stories and traditions of which Medea is the most familiar. This Greek tragedy tells of Medea murdering the sons she had with Jason to avenge his marital abandonment. One psychiatrist (Wertham, 1949) has taken the interpretation further and seen Medea’s murders as committed in the name of women’s rights. A parallel story is found in the Mexican myth of La Llorona. This folktale is about a spurned wife who, angry and vengeful toward her unfaithful husband, kills her two children and then wanders the earth crying while searching for them. In an Asian context, we also found reference to a Japanese pattern of parent-child suicide, oyako shinju, that seems to provide some degree of cultural rationale for filicide and/or familicide (Iga, 1996). In this Japanese cultural model, a mother is seen as having a duty to her children that entails “taking care” of them even through murdering them, if her life’s burdens become too unbearable. The emotional identification between mother and child is thought to be so strong that a suicidal mother who left her children behind would be seen as worthy of condemnation. This leads us to the issue of what burden may be so unbearable that it can be dealt with only through murder and possibly suicide. Several studies (for example, Alder & Polk, 2001) make reference to the fact that murderers do not see themselves as having any option other than fatal violence. Yet from the point of view of an observer, it is not clear that these mothers have considered or exercised all options short of taking another’s life. Totman (1978), for one, described her sample of “murderesses” as having “exhausted all other alternative courses of action either actually or in fantasy and find[ing] them not viable” (p. 94). Much, then, 154 HOMICIDE STUDIES / May 2004 has to do with the perceptions held by those who ultimately resort to murder. Many women who have been divorced or found their husbands involved in a new romantic relationship have turned their backs on the crisis and begun anew in rebuilding their lives. This may not be a simple or painless process, but lives go on. Similarly, many women who no longer could provide for children or found them to be a burden have given them up for adoption. It appears that the main reason why the women in our sample did not pursue these other options is to be found in what might be called a matriarchal corollary to patriarchal ideology, the idea of mothers owning their children. This corollary seems to be a key ingredient in directing action away from life-preserving options and toward the disastrous outcomes our research uncovered. If we understand the offender’s perceptions as a critical explanatory element, this allows us to tie together the cases that seem to be more the product of mental illness or psychiatric problems and those that stem from more common domestic role losses. The triggering events that seem central for some women had their foundation in psychological states or idiosyncratic perspectives held by the offender alone. These psychological elements were not readily shared or were quite likely to be disbelieved by others when they were divulged. As aforementioned, Andrea Yates was able to talk to others about her deep depression, but she was apparently unable to convince her husband or professional counselors how serious her desperation was. Or when Patricia Reager felt that she was in imminent danger of being attacked by Johnny Cochran’s thugs, these fears were not likely to be taken seriously by her confidants. As with the relationship or role loss cases, the effect that these distorted perceptions had on the women’s actions was to make it seem that murder was the only viable option. In short, the perceptions of the offenders are crucial in triggering their homicidal attacks. It seems important to begin with this inner perspective in the understanding of these particular murder cases. What does the research presented here offer as insight with regard to some of the major perspectives on murder by women? A common perspective on women’s homicide is that their murders, particularly those committed against intimate partners, are an expression of self-help. In brief, the argument (Peterson, 1999) is that offenders try to resolve their life problems through their own Messing, Heeren / WOMEN MULTIPLE MURDERERS 155 efforts as a result of having found no assistance through formal or legal channels. In a large proportion of the cases looked at here, especially the child murders, it appears that the offenders did not necessarily find that formal avenues of justice were inaccessible. Instead, they often wanted to achieve goals that had little to do with justice and thus could not be brought about by any formal system. For example, the revenge or resignation evident in these killings does not appear to be an attempt to rectify some injustice as much as it is following through on a particular emotional path (Katz, 1988). In other words, like most instances of domestic violence, it is much more an expressive than an instrumental murder (Alder & Baker, 1997). The concept of self-help as a source of criminal activity appears to carry a much more rational or instrumental connotation. Nor can it be said that from the point of view of the mother who murders her child, this child is seen as a deserving victim; there is little or no sense of any kind of victim precipitation. Another perspective is that which contrasts the distinct attributional styles of men and women and sees women’s lower murder rate as the product of a tendency to blame themselves for frustrations and stresses of everyday life. Hence, women are more likely to commit suicide rather than homicide, whereas men are more likely to blame others and therefore, commit more homicide (Vollum & Titterington, 2001). A major difficulty the present research presents for this theory is in the frequent occurrence in our sample of homicide followed by suicide or attempted suicide. These cases would seem to confound the expectation that one chooses one or the other direction for expressing lethal violence. Moreover, we do not get much sense in most of the child killings in our sample that the victim is actually being blamed for stresses in the mother’s life. Our examination of these cases of multiple murder has turned up features among these women that are both similar to and different from those that have been found among male mass murderers. The similarities are seen in the common contributing factors, the occurrence of several cases of familicide, and in the association of the multiple homicides with the suicide of the offender. Yet these similarities have to be qualified by noting the extreme rarity of these murders by women. Although rare, these murders are, nevertheless, typical of women as compared to, for 156 HOMICIDE STUDIES / May 2004 example, workplace murders by women or school shootings by adolescent females. What seems distinctive, as well, among these women murderers is that they were responding to traditional concerns about the loss of their position in the family. Killing multiple victims, especially their own children, demonstrates the continuing prevalence of the kind of patriarchal ideology that does not give equal weight to the right to life of all those who have been born. In the present instance, the most common presumptive hierarchy that allows for these killings to occur is that between mother and child. REFERENCES Alder, C. M., & Baker, J. (1997). Maternal filicide: More than one story to be told. Women and Criminal Justice, 9(2), 15-39. Alder, C., & Polk, K. (2001). Child victims of homicide. Cambridge, MA: Cambridge University Press. Athens, L. H. (1992). The creation of dangerous violent criminals. Urbana: University of Illinois Press. Athens, L. H. (1997). Violent criminal acts and actors revisited. Urbana: University of Illinois Press. Baker, N. V. (1999). Family killing fields: Honor rationales in the murder of women. Violence Against Women, 5(2), 164-185. Block, C. R., & Christakos, A. (1995). Intimate partner homicide in Chicago over 29 years. Crime & Delinquency, 41(4), 496-526. Bureau of Justice Statistics. (2002). U.S. Department of Justice, Office of Justice Programs. Available from http://www.ojp.usdoj.gov/bjs/ Cannizaro, S., & Turni, K. (1996, April 4). Violent woman slaughters family: Mother distraught over end of marriage. Times-Picayune, p. A1. Chew, K. Y., McCleary, R., Lew, M. A., & Wang, J. C. (1999). The epidemiology of child homicide in California, 1981 through 1990. Homicide Studies, 3(2), 151-169. Cleeton, E. R. (2001). How could a mother . . . ? Matricide and text mediated relations of family discourse. Cultural Studies Critical Methodologies, 1(4), 430-449. Cohen, D. A. (1995). Homicidal compulsion and the conditions of freedom. Journal of Social History, 28(4), 725-764. Crimmins, S., Langley, S., Brownstein, H. H., & Spunt, B. J. (1997). Convicted women who have killed children. Journal of Interpersonal Violence, 12(1), 49-63. Daly, M., & Wilson, M. (1988). Homicide. New York: Aldine. Dietz, P. E. (1986). Mass, serial, and sensational homicides. Bulletin of the New York Academy of Medicine, 62, 477-490. Dutton, D. G. (1999). Limitation of social learning models in explaining intimate aggression. In X. B. Arriaga & S. Oskamp (Eds.), Violence in intimate relationships (pp. 16-44). Thousand Oaks, CA: Sage. Duwe, G. (2000). Body count journalism: The presentation of mass murder in the news media. Homicide Studies, 4(4), 364-399. Fox, J., & Levin, J. (1998). Multiple homicide: Patterns of serial and mass murder. Crime and Justice, 23, 407-455. Messing, Heeren / WOMEN MULTIPLE MURDERERS 157 Fox, J., & Levin, J. (2001). The will to kill: Making sense of senseless murder. Boston: Allyn & Bacon. Fox, J. A., & Zawitz, M. (2001, January 4). Homicide trends in the United States (Crime and Justice Electronic Data Abstracts, Department of Justice, Bureau of Justice Statistics). Retrieved August 2002 from http://www.ojp.usdoj.gov/bjs/homicide/homtrnd.htm Frye, C. (2000, May 7). Special report: “I’m sorry, Momma.” Arkansas Democrat-Gazette, p. A1. Haapasalo, J., & Petaja, S. (1999). Mothers who killed or attempted to kill their child: Life circumstances, childhood abuse and types of killing. Violence and Victims, 14(3), 219-239. Hamberger, L. K., & Lohr, J. M. (1997). An empirical classification of motivations for domestic violence. Violence Against Women, 3(4), 401-424. Heberle, R. (1999). Disciplining gender: Or, are women getting away with murder? Journal of Women in Culture and Society, 24(4), 1103-1112. Hempel, A., Meloy, J., & Richards, T. (1999). Offender and offense characteristics of a nonrandom sample of mass murderers. Journal of the American Academy of Psychiatry and the Law, 27(2), 213-225. Hickey, E. (1997). Serial murderers and their victims. San Francisco: Wadsworth. Iga, M. (1996). Cultural aspects of suicide: The case of Japanese oyako shinju (parent-child suicide). Archives of Suicide Research, 2, 87-102. Jensen, V. (2001). Why women kill: Homicide and gender equality. Boulder, CO: Lynne Rienner Publishers. Johnson, I. M. (1996). Female murderers in a southern city, 1975-1992. American Journal of Criminal Justice, 20(2), 207-224. Jurik, N. C., & Winn, R. (1990). Gender and homicide: A comparison of men and women who kill. Violence and Victims, 5(4), 227-242. Katz, J. (1988). Seductions of crime: Moral and sensual attractions in doing evil. New York: Basic Books. Kelleher, M. D. (1997). Flash point: The American mass murderer. Westport, CT: Praeger. Killias, M., vanKesteren, J., & Rindlisbacher, M. (2001). Guns, violent crime, and suicide in 21 countries. Canadian Journal of Criminology, 43(4), 429-449. Lambie, I. (2001). Mothers who kill: The crime of infanticide. International Journal of Law and Psychiatry, 24, 71-80. Langford, L., Isaac, N., & Kabat, S. (1998). Homicides related to intimate partner violence in Massachusetts. Homicide Studies, 2(4), 353-377. Leonard, E. D. (2001). Convicted survivors: Comparing and describing California’s battered women inmates. The Prison Journal, 8(1), 73-86. Link, B. G., Monahan, J., Stueve, A., & Cullen, F. T. (1999). Real in their consequences: A sociological approach to understanding the association between psychotic symptoms and violence. American Sociological Review, 64, 316-332. Mann, C. R. (1996). When women kill. New York: State University of New York Press. Moore, A. M. (1997). Intimate violence: Does socioeconomic status matter? In A. P. Cardarelli (Ed.), Violence between intimate partners: Patterns, causes, and effects (pp. 90100). Needham Heights, MA: Allyn & Bacon. Ogle, R. S., Maier-Katkin, D., & Bernard, T. J. (1995). A theory of homicidal behavior among women. Criminology, 33(2), 173-193. Palermo, G. (1997). The berserk syndrome: A review of mass murder. Aggression and Violent Behavior, 2(1), 1-8. Palermo, G., & Ross, L. E. (1999). Mass murder, suicide, and moral development: Can we separate the adults from the juveniles? International Journal of Offender Therapy and Comparative Criminology, 43(1), 8-20. Petee, T. A., Padgett, K. G., & York, T. S. (1997). Debunking the stereotype: An examination of mass murder in public places. Homicide Studies, 1(4), 317-337. Peterson, E. S. L. (1999). Murder as self-help: Women and intimate partner homicide. Homicide Studies, 3(1), 30-46. 158 HOMICIDE STUDIES / May 2004 Polk, K. (1994). When men kill: Scenarios of masculine violence. New York: Cambridge University Press. Ptacek, J. (1999). Battered women in the courtroom: The power of judicial responses. Boston: Northeastern University Press. Resnick, J. (1969). Child murder by parents: A psychiatric review of filicide. American Journal of Psychiatry, 126, 325-333. Richards, C. E. (2000). The loss of innocents: Child killers and their victims. Wilmington, DE: Scholarly Resources. Riedel, M., & Best, J. (1998). Patterns in intimate partner homicide: California 1987-1996. Homicide Studies, 2(3), 305-320. Shackelford, T. K. (2001). Partner killing by women in cohabiting relationships. Homicide Studies, 5(3), 253-266. Silverman, R. A., & Kennedy, L. W. (1988). Women who kill their children. Violence and Victims, 3(2), 113-127. Smith, P. H., Moracco, K. E., & Butts, J. D. (1998). Partner homicide in context: Apopulationbased perspective. Homicide Studies, 2(4), 400-421. Stack, S. (1997). Homicide followed by suicide: An analysis of Chicago data. Criminology, 35(3), 435-453. Steffensmeier, D., & Allan, E. (1996). Gender and crime: Toward a gendered theory of female offending. Annual Review of Sociology, 22, 458-487. Stote, R., & Standing, L. (1995). Serial and multiple homicide: Is there an epidemic? Social Behavior and Personality, 23(4), 313-318. Straus, M. (1999). The controversy over domestic violence by women: A methodological, theoretical and sociology of science analysis. In X. B. Arriaga & S. Oskamp (Eds.), Violence in intimate relationships (pp. 16-44). Thousand Oaks, CA: Sage. Totman, J. (1978). The murderess: A psychological study of criminal homicide. San Francisco: R & R Research. U.S. Bureau of the Census. (2000). Census 2000. Available from http://www.census.gov/ main/www/cen2000.html Vollum, S., & Titterington, V. B. (2001). Gender, attributional styles, and direction of lethal violence. Homicide Studies, 5(2), 227-252. Websdale, N. (1999). Understanding domestic homicide. Boston: Northeastern University Press. Websdale, N., & Alvarez, A. (1998). Forensic journalism and patriarchal ideology: The newspaper construction of homicide-suicide. In F. Y. Bailey & D. C. Hale (Eds.), Popular culture, crime, and justice (pp. 123-143). San Francisco: Wadsworth. Wertham, F. (1949). The show of violence. Garden City, NJ: Doubleday. Wilczynski, A. (1997). Child homicide. Oxford, UK: Oxford University Press. Zamorra, J. H., & Hatfield, L. D. (1998, March 25). Awful details of 3 girls’ slayings: Mother was taking prescription drugs for depression, seizures. San Francisco Examiner, p. A1. Zimring, F. E., & Hawkins, G. (1997). Crime is not the problem: Lethal violence in America. New York: Oxford University Press. Jill Theresa Messing is a doctoral student in the School of Social Welfare at the University of California, Berkeley. Her research interests include family violence, child welfare, and foster care. John W. Heeren received his Ph.D. at Duke University and has taught for many years at California State University, San Bernardino. His current research interests are in multiple murder and religion in the mass media.