022 289/90/301-9001903.0/0 ‘Tie Jounal oF hata opti 182 by Wiliams & Wiking Val 23,No.1 Printed in US MEN, WOMEN, AND MURDER: GENDER-SPECIFIC DIFFERENCES IN RATES OF FATAL VIOLENCE AND VICTIMIZATION Arthur L. Kellerman, MD, MPH, and James A. Mercy, PhD To study the potential differences that distinguish homicides involving women as victims or offenders from those involving men, we analyzed Federal Bureau of Investigation Uniform Crime Reports data on homicides that occurred in the United States between 1976 and 1987. Only cases that involved victims aged 15 years or ‘older were included. Persons killed during law enforcement activity and cases in which the victim's gender was not recorded were excluded. A total of 218,273, homicides were studied, 77% of which involved male victims and 23% female victims. Although the overall risk of homicide for women was substantially lower than that of men (rate ratio [RR] = 0.27), their risk of being killed by a spouse or intimate acquaintance was higher (RR = 1.23). In contrast to men, the killing of a woman by a stranger was rare (RR = 0.18). More than twice as many women were shot and killed by their husband or intimate acquaintance than were murdered by strangers using ‘guns, knives, or any other means. Although women comprise more than half the U.S. Popuiation, they committed only 14.7% of the homicides noted during the study interval. in contrast to men, who killed nonintimate acquaintances, strangers, or victims of undetermined relationship in 80% of cases, women killed their spouse, an intimate acquaintance, or a family member in 60% of cases. When men killed with un, they most commonly shot a stranger or a non-family acquaintance. When women killed with a gun, the victim was five times more likely to be their spouse, an intimate acquaintance, or a member of their undetermined relationship. family than to be a stranger or a person of PRESS ACCOUNTS suggest that women’s fear of vie- timization has increased in recent years. Highly pub- licized reports of homicides and sexual assaults have led many women to conclude that they are vulnerable to attack.‘ Firearm manufacturers and others have capital- ized on this trend by aggressively promoting small hand- guns for women as an effective means of self defense. Since homicide is the most serious of violent crimes, it is frequently given disproportionate attention by the print and electronic media.’* Relatively little attention has been paid, however, to the potential differences that distinguish homicides involving women as victims or offenders from those involving men. To study this issue, we analyzed Federal Bureau of Investigation Uniform Crime Reports (UCR) data for the years 1976 through 1987. We were particularly interested in answering the following questions: (1) In comparison with men, how From the Division of Emergency Medicine, Department of Medicine, Unversity of Tennessee, Memphis, and the Dvsion o nry Epidemiol: ‘gy and Control, Cente for Environmental Health and Injuxy, Contr, Centers for Disease Control, Alana, Georgi Presented at the annua! meeting ofthe Society of Academic Emer: gency Medicine, May 15, 1981, Washington, D.C ‘The views expressed are thse ofthe aUthors and do not necessary reflect those ofthe University of Tenaessee or the Contors for Disease Conta ‘Adotess for reprints: Arthur L. Kelermann, MD, MPH, Divison of Emergency Mecicne, 877 Jetlerson, Room G164, Memphis, TN 38708, likely is a woman to be a victim of homicide? (2) When a woman is murdered, how often does the offender prove to be her spouse, intimate acquaintance, or family mem- ber rather than a stranger or unidentified assailant? (3) When a woman kills, is she more likely to kill a stranger or a member of her family? METHODS. ‘The Fedoral Bureau of Investigation’s Uniform Crime Re porting (FBI-UCR) program is @ nationwide cooperative ven- ture of more than 15,000 city, county, and state law enforce: ‘ment agencies, Bight index crimes, including the violent crimes of murder and nonnegligent manslaughter, forcible rape, rob: bery, and aggravated assault, a2 well as the property crimes of burglary, larceny, motor vehicle theft, and arson are currently monitored and tabulated on an annusl basis. Only erimes that come to the attention of law enforcement agencies and are voluntarily reported to the FBLUCR are counted in these Statistics. Not all law enforcement agencies participate in the FBI-UCR, but most do; in 1982, law enforcement agencies representing 224 million US citizens (97% of the nation’s population) reported to the FBL-UCR. This coverage included 99% of persons living in SMSAs, 95% of those living in “other cities", and 90% of persons living in rural areas as defined by the United States Census Bureau, Murder and nonnegligent manslaughter are defined by the FBI-UCR program as “the willful (nonnegligent) killing of one ‘human being by another.” Given the seriousness of the offense and its high rate of detection, homicide is the most reliably2 The Journal of Trauma documented crime reported to the FBI-UCR; detailed case data are captured on approximately 96% of the homicides that occur im any given year. In addition to basic victim demographic information and data regarding the cause of death, participating law enforcement agencies submit a supplemental homicide reporting form (SHR) that includes information about incident circumstances, the vietim-offender relationship, and the type ‘of weapon involved.” Unfortunately, relevant information is sometimes lacking from these forms. Although the nationsside ‘case clearance rate for murder is higher than for any other crime index offense, a substantial minority of homicides are still under investigation at the time the SHR is filed, For this reason, the identity of the offender is listed as “unknown” in approximately 31% of cases each year. ‘The UCR data tapes for the years 1976 through 1987 were utilized for our analysis. Only cases involving suspects and victims aged 15 years and older were included. Homicides involving persons killed by police officers during law enforce- ment activity were excluded, Cases in which the gender of the victim was not reported were algo excluded. For purposes of analysis, multiple vietim homicides were considered separate ceases based on the identity of each vietim. Computations of ‘homicide rates per 100,000 persons were based on data obtained from population surveys conducted by the U.S. Bureau of ‘Cencus for each year from 1976 to 1987." RESULTS During the 12-year study interval, 215,273 homicides involving victims aged 15 years or older were reported to the FBI. This total excludes 4,323 homicides committed by law enforcement officers in the line of duty. In 165,990 cases (77.1%), the homicide victim was male. In the remaining 49,266 cases (22.9%) the victim was female. Overall, the risk of homicide for women was substan- tially lower than that noted for men (4.43 vs. 16.24 per 100,000 person years, Rate Ratio (RR) = 0.27). However, the risk of being killed by one’s spouse or intimate acquaintance was somewhat higher for women than men (female/male RR = 1.23). The higher overall risk of homicide for men relative to women was almost entirely the result of their much higher rate of victimization at the hands of an acquaintance, stranger, or unidentified assailant. In contrast to men, the killing of a woman by @ stranger was rare (0.41 vs. 2.34 per 100,000 person years, RR = 0.18) (Table 1). ‘Almost 40% of women were killed by their spouse, an intimate acquaintance, or a family member. In cases in which the identity of the offender was known, murders of women by spouses, intimate acquaintances, or family ‘members exceeded murders by strangers by a ratio of 4.3 to 1. Men, on the other hand, were killed by strangers ‘more often than they were killed by spouses, intimate acquaintances, or family members combined (Table 1). When homicides were analyzed by the type of weapon employed, knives or other cutting instruments were found to have been used to kill 20.4% of male victims and 20.3% of female victims. Almost 30% of women were bludgeoned, strangled, or suffocated, compared with only 12% of men, However, a firearm (usually a handgun) was the weapon most commonly used to kill victims of either July 1992 sex. Seventy-four percent of all firearm homicides in- volved a handgun (Table 2). Of note, during the entire 12-year study interval, more than twice as many women were shot and killed by their husband or an intimate acquaintance than were murdered by strangers using firearms, knives, or any other means. Although women comprise more than half of the U.S. population and 23% of all homicide victims, they com- mitted only 14.7% of the homicides reported during the study interval (male/female offender RR = 6.33). In contrast to men, who killed nonintimate acquaintances, strangers, or individuals of undetermined relationship in roughly 80% of cases, women killed their spouse, an intimate acquaintance or a member of their family in {60% of cases (Table 3). Despite this striking difference in the proportion of homicides involving a close personal acquaintance, women were less likely to kill than men at any level of relationship. The ratio of female to male offenders narrowed considerably, however, when the homicide involved a close personal acquaintance. Women killed casual acquaintances, strangers, and unidentified victims at rates only one twentieth to one tenth that of men, but their rate of homicide involving a spouse or intimate acquaintance was 70% of that noted for men (Table 3), Substantially more civilian homicides com- mitted by men were classified “legally justifiable” (ie., they involved the killing ofa felon during the commission of a felony) but the proportion of such homicides was not significantly different for women compared with men (2.1% vs, 2.0%, x’ p< NS). However, the FBI's definition of justifiable homicide does not include many killings subsequently ruled to have been committed in self de- fense.® A firearm was the weapon of choice for 56.4% of women offenders and 63.2% of men. Knives and other sharp instruments were used by 27.4% of women and 19.7% of men. Other weapons or means were used by 16.0% of women and 17.1% of men. When men killed with a gun, they most commonly shot a stranger or a non-family acquaintance. When women killed with a ‘gun, their victim was five times more likely to be their spouse, an intimate acquaintance, or a member of their family than to be a stranger or a person whose relation- ship to the offender was undetermined (Table 4). DISCUSSION Our analysis of FBI homicide data between 1976 and 1987 confirms that the circumstances and characteristics of homicides involving women differ substantially from those involving men. In contrast to a man, a woman is far more likely to be killed by her husband, an intimate acquaintance, or a member of her family than she is to bbe murdered by a stranger or unidentified intruder. This observation should not be surprising in light of a growing body of knowledge about violence against women. Ac- cording to Stark and Fliteraft, domestic violence is theVol. 33, No. 1 ‘Men, Women and Murder 3 Table 1 Rate, number, and percentage of homicides by gender of vetim and victim/otfender relationship: United States, 1976-1987 eee Ferae Visine — Tile Visine Faeeae) uber ——Aatey——_—Pwoont “Numba Fate Pecont ‘ate Rato ‘Spouse 11236 «1.01 228 8.383 082 0 123 Intimate Acquaintance sai8 O48 108 4222 oat 25 47 (Other Famiy 2957 028 60 378 098 60 027 ‘Acquaintance 9930 0.89 202 62941 Bie 379 o14 Stranger 4521 Oat 82 27.995 234489 018 Undetermined 153200138 sit 52507 514 316 o27 Total 49262443 1659751624 027 ~ Totes exclide vies aged 15 years or less, persons kiled by poice offcers engaged in law enforcement activity, and cases in which the ‘gender o the victim was not recorded. Eighteen cases in which the vcimjoffender relavonship was not noted are aio excluded, Rate: tumber of Homicides per 100,000 person year. Table 2 [Numer and proportion of homicides by gender of vieim and ‘weapon involved: United States, 1976-1987 Murcer Weapon eed Ramter”Pecont “Number Perent Handgun 17784 96.1 82984 50.0 Long Gun Baas 11921983 132 Other Gun 3298026718543 Knife/Cuttng instrument "10203-93792 204 Blunt Object a199 85772047 Hands/Fiste/Feet 4503 91 38 Strangulaton 2672 54 o7 othert seat 81 32 Total 49,266 * Totals exclude cts aged 15 years or les, parsons Kiled by pace ‘officers engaged inlaw enforcement activity, homcdes im which felon ‘was justifiably kiled by a private ctizen, and cases in which the gender (ofthe vitim was not records, Examples of other weapons include poisoning, explosives, fr, ‘owning, asphyxiation, and being pushed or thrown out a window. leading cause of injury to women and accounts for more visits to hospital emergency departments than car crashes, muggings, and rapes combined."* Battering of women is not isolated to a particular socioeconomic group. Surveys suggest that in a lifetime, as many as one woman in four will be beaten by her husband or an intimate acquaintance. Given the realization that women are most commonly victimized by another member of their household or an intimate acquaintance (individuals who presumably have ‘equal access to any readily available weapon), efforts to encourage women to purchase handguns for self protec tion may be counterproductive. Ironically, some believe that acquisition of firearms hy women could potentially deter episodes of battering. According to Kleck, “Victims of domestic violence can take advantage of behavioral cues which serve as advance warning signs and ready themselves accordingly. In most threatening situations, advance preparations could include securing a weapon...” Although the proportion of such armed confrontations that end in the death of the batterer or his vietim is unknown, domestic homicides of both women and men are often preceded by an escalating pattern of violence. Our data confirm that women who reach a firearm shoot, and kill their husbands almost as often as men shoot their wives, but research by others suggests that their reasons for pulling the trigger are often quite different.” Men commonly kill their female partners in response to the woman's attempt to leave an abusive relationship. Women, on the other hand, often kill their male partners: Table 3 Rate, number and percentage of female and male homicide offenders by victim/offender relationship: United States, 1976-1967 Fomae Orendare ae Ofenaers vi erate Svoterde Retest Number Ratet——_—Persrt_— abaya Peon ‘Spouse e272 ova 314 yo8st 1.06 7a 070 Intimate Acquaintance ass 032 138 3701 056 37 056 Other Famiy 3807038 148 yeg03 126 84 0.28 Acquaintance 7388088 279 71683701 487 0.09 Stranger 1729016 66 ais2 30804 0.05 Undetermined 151804 58 24050 208137 07 Total 26,384 153,440 “Totals exclude cases involving veims aged 1 years or less, persons kiled by pole officers engaged inlaw enforcement activity, cases in \ihich the ienty of the offender was unknown, and cases in which the gender ofthe vicim was not recorded. ‘fate = numberof homicies per 100,000 person year.4 The Journal of Trauma uly 1992 Table 4 [Number and proportion of homicides by gender of offender, weapon involved and victim/ottender relationship: United States, 1976. 7 Freas Tava arent her Vet tender Tes Feat ae Temas Woes Fenaes Naber Paco Humber Peco _Nanter Paani Naber Pecot_Wunber Porat Nunber Perot Spouse 760 +77 5920 007 180) 44 2098 «281 «4as9=«72~=~«CK]SCT A Inmate Ao- S20 G4 BNL SHB? 39ST BIT HB 2B cuanance OterFamiy, —7968«78~=«10H0«70 aah 7888S ga 25 ae 8.2 Requniance «44982484 «3980-257 15695 SIT 2ATD. 411108 422 OTS Sranger 2008s 207 «Note 70 S748 180«=S S075 SAS 207 «2k Nosuspect «1361641749 $0035 1087 SKS a7 8588 “ota 96923 14916 20261 7248 28.256 4220 “Totals excude victims aged 15 years o less, persons kiled by poloe ofcers engaged in aw enforcement activity, cases in which the identity of the offender was net known, and cases in which the gender ofthe victim was pot recorded. in self defense or in retribution for prior acts of vio- lence." Even ifa consensus could be established that homicide is an acceptable response to domestic disputes or at tempted victimization, ownership of a firearm does not guarantee that it will be reached in time or used effec: tively if reached. Although Americans own approxi- mately 120 million firearms (roughly 30% of which are handguns)," less than 2% of robberies in the United States are resisted with a firearm."* In confrontations with assailants, even experienced owners of firearms can find their weapons turned against them. Sixteen percent of the 696 U.S. law enforcement officers shot and killed in the line of duty between 1981 and 1990 lost their lives when their service weapon was turned against them.” Firearms kept in the home may be particularly problem- atic. One study of firearm-related deaths in King County, ‘Washington found that for every time a gun in the home was used to kill in self defense, there were 1.3 uninten- tional gunshot deaths, 4.6 criminal homicides, and 37 firearm suicides. Four limitations to our analysis warrant comment, 1. Because of their high rates of detection and report- ing, our study was restrieted to UCR documented homi- cides, Many robberies, beatings, and sexual assaults are not reported to police, especially when the assailant is ‘well known to the vietim, Data generated by the National Crime Survey suggest that with the exception of rape, men are more likely than women to be victims of serious but nonfatal criminal violence." However, in cases of domestic violence, the relative frequency of injuries among women and men is reversed.” 2. Local law enforcement agencies generally submit UCR case reports early in the course of their investiga- tion, sometimes before the identity of the offender is known. Although many initially unsolved homicides are eventually determined to have been committed by a stranger, follow-up data from one large metropolitan police jurisdiction (Los Angeles, California) suggest that {substantial number turn out to involve an acquaintance or a relative of the victim.” It is not certain, however, that this urban data set is representative of the overall pattern of vietim-offender relationships in the nation as a whole. 8. ‘The complex natuire of interpersonal relationships frequently exceeds classification schemes as simple as the one used by the FBI, When homicides involve mar- ried couples, the victim is invariably listed as the spouse of the offender, even if the couple has been separated for years or has filed for divorce. A woman trapped in an ‘abusive relationship with a man is generally coded as having been killed by her “boyfriend,” regardless of the actual intimacy of the relationship. Drooz has observed that “acquaintances” can include enemies as well as friends.” 4. Although homicide is clearly the most setious con- sequence of interpersonal violence, itis not the only one. Women may elect to purchase a firearm or adopt other self-protection strategies in hopes of decreasing their ‘chances of becoming a victim of rape, robbery, or assault ‘as well. In 1966, the Orlando Police Department trained 6,000 women to use firearms in a highly publicized rape prevention program. In 1982 the town council of Ken: nesaw, Georgia passed a largely symbolic ordinance that requires every homeowner to keep a gun in his or her home.” In both cases, short-term declines in rates of victimization were touted as evidence that these inter- ventions had effect." However, Green’ has questioned the reliability of the Orlando data, and McDowell etal.” have shown that the Kennesaw ordinance had no dis- cemnible effect on that city’s fluctuating rate of burglary over time. Despite public perceptions to the contrary, the overall risk of homicide among women is substantially less than that of men. Statistically speaking, a woman's risk of being killed by a stranger or an unidentified assailant is Jess than one fifth that of a man, In contrast to men, who are most often killed by an acquaintance or a stranger, women are far more likely to be murdered by their spouse, a boyfriend, or a member of their household. Conversely, when women kill, their victim is five times ‘more likely to be their spouse, an intimate acquaintance,Vol. 33, No. 1 or a member of their family than to be a stranger or an individual of undetermined relationship. In light of these data, the wisdom of promoting firearms to women for self protection should be seriously questioned. Since women are most likely to kill or be killed by a close personal acquaintance, nonlethal efforts to recog- nize and break the cycle of domestic violence will prob- ably have a far more favorable impact on rates of vio- lence, serious injury, and homicide than promotion of firearms for self-defense.”**? Greater recognition of gen- der-specific differences in the nature and rate of violence and victimization is needed in order to identify optimal strategies for prevention and control. ‘Acknowledgments We are indebted to the Federal Bureau of Investigation Uniform Crime Reports program for collection and collation of the primary data used in our analysis. We are also indebted to Ms. Carol Conway and Ms, Julie Minnick for ith the preparation of the manuscript. [REFERENCES 1 Church G: ‘The other arms race, Time Magazine, February 6, 20- 26, 1989 2, Johnson P: I fee] more seeure having @ gun around, USA Today, ‘March 1, 1988, pp AL-AZ 8, Davis AB: Women and guns. The Commercial Appeal, Memphis, ‘TN, January 26, 1986, pp G1-G2 4. Buchalter G: Why T own # gun. Parade Magasine, February 21, 1988, pp 4-7 5. Ayoob M? Take back the night. Guns Magazine, 8:66-61,76,78-82, 1989 6. Quigley P: Armed and Female, New York, EP Dutton, 2989 1. Woltgang ME, Figlo RM, Tracy PE, et a: The National Survey of ‘Grime Severity. Washington, DC, US Government Printing Of fice, 1985, 5, Graber DA: Crime, News and the Public, New York, Praeger, 1980 8, Crime in the United Staves 1982: Uniform Crime Reports for the United States Washington, DC, Federal Bureau of Investigation, United States Department of Justice, 1983 Uniform Crime Reporting Handbook. Washington, DC, Federal Bureau of Investigation, United States Department of Justice, 1084 11. US Bureau of Census: Cursent Population Reports, Series P-25, 'No, 870, Estimates ofthe Population ofthe United States by Age, Face and Sex: 1976 to 1973 Washington, DC, US Government Printing Office, 1980 US ‘Bureau of the Ceneus: Current Population Reports, Series P25, No, 1092, United States Population Estimates by ge, Sex ‘and Race: 1880 to 1987. Washington, DC, US Government Print- Ing Office, 1988. 2. 16. 16. i. 18 19. 2 2. 2%. 2, 28, 29, ‘Men, Women and Murder 5 ‘Stark B, Pliteraft A Spouse Abuse In: Surgeon Generals Workshop on Violence and Public Health: Sourcebook. Leesburg, VA. Octo. ber 27-29, 1885. Atlanta, Centers for Diseare Control, SA1-SA(3 Straus M, Gelles R, Steinmete S: Behind Claved Doors: A Survey ‘of Family Violence in America. New York, Doubleday, 1980 ‘Shulman MA: Survey of Spousal Violence Against Women in Ken- tuoky. Lexington, Kentueky Commission on Women, 1979; sty zo, 10927001 ‘Kleck G: Crime control through the private use of armed force. ‘Socal Problems, 35:1, 1958. Mercy JA, Saltzman LE: Fatal violence among spouses in the United Stats, 1976-1985. Am J Public Health 9-575, 1889 Cazemore N, Zahn M: Women, murder and male dominetion: Police reports of domestic homicide in Chicago and Philadelphia, Paper presented atthe annual meeting af the American Society of Criminology, Atlanta, GA, 1986 Browne A: Assault and homicide in the home: When bettered ‘women kill. In Saks MJ, Saxe L (eds): Advances in Social Paychology, Vol 8. Hillsdale NJ, Lawzence Bilbaum Associates, 1987 Wilbanks W: The female homicide offender in Dade County Flor: ida, Criminal Justice Rev 89, 1983, ‘Wright JD, Rossi PH, Daly K, eta: Weapons, Crime and Vinlenoe in America: A Literature Review and Research Agenda. Washing: ton, DC, US Government Printing Offic, 1981, pp 60-67 Law Enforcement Officers Killed and Assoulted. Uniform Crime Reports, Washington, DC, Federal Bureau of Investigation, United States Department of sustice, 1900 Kellermann AL, Reay DT Protection pei? An analysis of firearm ‘lated deaths in the home. N nal J Med 3141557, 1986 US. Bureau of Justice Statistics: Criminal Victimization in the United States, 987 NCJ-115524). Washington, DC, Department of Justin, 1889 Stets JE, Straus MA: Gender differences in reporting marital violence and its medical and peychological consequences. In Straus MA, Gelles RJ (eds): Physical Violence in American Families: Risk Focvors and Adaptation to Violence in 8,145 Fam lies. New Brunswick NJ, Transaction Press, 1990, pp 151-165, ‘University of California, Loe Angeles, Centers for Disease Control ‘The Epidemiology of Homicide in the City of Los Angeles, 1970- 1979, Department of Fiealth and Human Services, Publie Health Service, Centers for Disease Control. August, 1985. Droog RB: Handguns and hokum: « methodological problem. JAMA 238:48, 1877 Cook PJ: The technology of personal violence. In Tonry M (ed Crime and Justice: A Review of Research. Chicago, University of Chicago Press, 1991, Kleck G, Bordua DJ: The foctual foundation for certain key as ions of gun control. Law and Poly Quarterly 6.271, 1983 ‘Citizen gun ownership and criminal deterrence: Theory, research and policy. Criminology 25:68, 1987 MeDowall D, Wiersema B, Loftin C: Did mandatory firearm own: ‘ership in Kennesaw prevent burglaries? Sociology and. Social Research 7448, 1989 ‘The National Committee for Injury Prevention and Control: Do- Inestic violence. In: Injury Prevention: Meeting the Challenge. New York, Oxford University Pres, 1889, pp 222-252 Randall T: Domestic violence intervention calls for mote than ‘eating injuries. JAMA 261:958, 1990