Differences in female and male involvement in lethal violence in Russia

Abstract

Relatively little is known of the distributions of homicide event characteristics in non-Western nations in which women relative to men are involved. This article utilizes unique homicide narratives drawn from Russian court and police records to compare homicide victim, offender, and event characteristics by sex of victim and separately by sex of offender. Results from logistic regression show that homicides in which a female was the victim or offender were more likely to occur between intimates and to occur in the home, whereas homicides involving males were more likely to occur in a public place, to be alcohol-related, to involve a firearm, and to involve a victim and offender who did not know each other well. These results not only present an important first glimpse at women as homicide victims and offenders in Russia specifically, but also provide a point of comparison with findings from similar analyses undertaken in the West, and present further initial observations upon which to construct a cohesive theory about female involvement in serious violent events.

Introduction

The past two decades have seen an increase in studies addressing the differences between homicides involving women as victims or offenders and those involving men. At the same time, there has been a growing understanding among scholars of the importance of studying homicide in different countries and different cultural contexts—particularly at the individual- and incident-levels (see, for instance, the comments of Cheatwood, Landau, & Petee in Smith, 2000). Likely due to difficulties accessing the necessary data, however, research on situational characteristics of male- versus female-involved homicides has rarely extended beyond the borders of North America, western Europe, and Australia.

The current study utilized unique homicide narratives drawn from Russian court and police records to compare homicide incidents by sex of offender, and separately sex of victim, with respect to victim, offender and event-level characteristics. Specifically, this article addresses the following questions with regard to homicide events in Russia: (1) how do the situational characteristics of female-perpetrated homicides differ from those committed by men, (2) how do the situational characteristics of homicide with female victims differ from those with male victims, and (3) are there characteristics of male-on-male homicide that are unique compared with homicides in which a female is involved (as victim, offender, or both)?

By seeking answers to these questions in a new and culturally different context, this study provided an opportunity for assessing the generalizability of Western-based theories of homicide offending and victimization. Further, though the current project was primarily descriptive in nature, research of this type is an important step in developing broadly informed grounded theory regarding gendered involvement in lethal violence.

Previous findings

The empirical literature on the distribution of homicides by sex reveals a fairly consistent pattern. Across cultures and over time homicide offenders tend to be overwhelmingly male, with most studies showing that men perpetrate 82 to 95 percent of all homicides (Carcach & Conroy, 2001; Chimbos, 1993; Connor, 1973; Kellermann & Mercy, 1992; Mouzos, 1999; Pridemore & Eckhardt, 2008; Salfati & Dupont, 2006; Shaw et al., 2006; Silverman & Kennedy, 1987; Wilbanks, 1983a; Wolfgang, 1958). Men are also the most frequent targets of lethal violence, though the sex ratio for homicide victims tends to vary more than for offenders. Past studies have reported that males comprise between 59 and 78 percent of all homicide victims (Carcach & Conroy, 2001; Chimbos, 1993; Kellermann & Mercy, 1992; Pridemore & Eckhardt, 2008; Salfati & Dupont, 2006; Shaw et al., 2006; Wolfgang, 1958).

With regard to who is killing whom, research has repeatedly shown that offenders of both sexes are more likely to kill a male victim (Carcach & Conroy, 2001; Jurik & Winn, 1990; Mouzos, 1999; Wilbanks, 1982, 1983a). Incidents with male offenders and male victims tend to account for just over half of all homicides, with male-on-female homicides comprising an additional 20 to 35 percent. Females rarely kill other females. Previous studies have generally found that cases in which a woman kills another woman made up less than 3.5 percent of all homicides (Carcach & Conroy, 2001; Mouzos, 1999; Wilbanks, 1983a).

Homicide characteristics by sex of offender

Prior studies comparing male- and female-perpetrated homicides have consistently found sex of offender to be correlated with a number of contextual factors. With regard to victim-offender relationship, for example, previous research has repeatedly found that female homicide offenders overwhelmingly kill those close to them, usually an intimate partner, or less often, a family member. Women much less frequently target friends or acquaintances and only very rarely kill strangers. By comparison, studies have shown that homicides involving friends or acquaintances comprise the largest percentage of male-perpetrated killings (approximately half), followed by those involving strangers (Chimbos, 1993; Jurik & Winn, 1990; Kellermann & Mercy, 1992; Kruttschnitt, 1995; Mann, 1996; Miethe & Regoeczi, 2004; Wilbanks, 1983a, 1983b; Wilson & Daly, 1992).

Findings from prior research have also been fairly consistent with regard to the correlation between the sex of offender and the location of the violent event. Studies have generally found that the vast majority of female-perpetrated homicides were committed in the home, while those committed by men were much more likely to occur in a public place (see Jurik & Winn, 1990; Wilbanks, 1983b; Wolfgang, 1958). Greek homicides, however, deviate from this pattern in that male offenders in Greece have been found equally likely, if not slightly more likely, to commit homicide in the home as they were to kill in a public place (Chimbos, 1993). This suggests that the spatial patterns of male-perpetrated homicide may be more affected by cultural differences than homicides committed by women.

With regard to motives, most studies have found that both male- and female-perpetrated homicides most commonly occur in the context of an argument or dispute as opposed to being motivated by profit or related to the commission of another crime (such as rape). This pattern, however, appears to be more pronounced for female offenders relative to their male counterparts. For instance, Jurik and Winn (1990) found that an argument was the precipitating incident in a higher percentage of female-perpetrated homicides as compared to those with male offenders (80 percent versus 63 percent), while profit-motivated killings accounted for a smaller percentage of killings by women (14 percent versus 25 percent for men)—results that supported the earlier findings of both Wilbanks (1983a) and Wolfgang (1958). Studies have also repeatedly found that female-perpetrated homicides are about twice as likely as those committed by males to be victim-precipitated (Goetting, 1988; Jurik & Winn, 1990; Wilbanks, 1983b).

In addition, prior research showed that homicides committed by male offenders are more likely than those committed by females to involve alcohol (Carcach & Conroy, 2001; Dauvergne, 2003; Wells & Graham, 2003; Wolfgang, 1958). Recent research on homicides in Russia, however, found no significant difference between alcohol- and nonalcohol-related homicides with respect to sex of offender (Pridemore & Eckhardt, 2008). This may suggest a stronger association between alcohol and violence in Russia relative to other nations, or it may simply reflect the much higher rates of consumption in that country (Pridemore & Kim, 2006).

Despite important cross-cultural differences in the proportion of homicides committed with a gun, studies have consistently shown that men are more likely than women to use a firearm in the commission of a homicide (Carcach & Grabosky, 1997; Chimbos, 1993; Goetting, 1988; Jurik & Winn, 1990; Kellermann & Mercy, 1992; Wilbanks, 1983a, 1983b). Results from a number of prior studies also showed that female offenders are more likely than males to kill using a knife (Goetting, 1988; Jurik & Winn, 1990; Kellermann & Mercy, 1992; Wilbanks, 1983a, 1983b). Much of this research, however, had been conducted in the U.S. where gun use predominates, particularly among male homicide offenders. Such research may be of limited utility in predicting results for nations where firearm availability is limited. In Russia, for example, guns are utilized in only about 10 percent of all homicides (Pridemore, 2006). Formulating a hypothesis with regard to sex differences in the use of bodily or blunt force is also difficult, as some past studies have found that male offenders are more likely than females to use bodily or blunt force in the commission of a homicide (Jurik & Winn, 1990; Wolfgang, 1958), while others have found the opposite to be true (Chimbos, 1993; Wilbanks, 1983b).

Homicide characteristics by sex of victim

With respect to a number of situational characteristics, findings from studies comparing male- and female-victim homicides differed little from those pertaining to sex of offender. Prior research on victim-offender relationships, for instance, had consistently found that male victims are most commonly killed by a friend or acquaintance (38 to 50 percent of cases), followed by strangers (17 to 32 percent), and intimates or family members (6 to 23 percent) (Kellermann & Mercy, 1992; Mouzos, 1999; Wilson & Daly, 1994; Wolfgang, 1958). By comparison, female victims are more than twice as likely as their male counterparts to be killed by an intimate partner, with studies indicating that between 38 and 70 percent of all femicides are committed by someone with whom the victim had an intimate relationship (Chimbos, 1993; Craven, 1996; Kellermann & Mercy, 1992; Moracco, Runyan, & Butts, 1998; Mouzos, 1999; Rennison & Welchans, 2000; Wilson & Daly, 1994; Wolfgang, 1958). Research further indicated that less than one-third of female victims are killed by a friend or acquaintance and even fewer (8 to 15 percent) are killed by a stranger (Chimbos, 1993; Kellermann & Mercy, 1992; Moracco et al., 1998; Mouzos, 1999; Wolfgang, 1958).

Previous studies have also found that women are more likely than men to be killed in the home, though the gender gap for homicide location appears to be less pronounced than it is in comparisons by offender sex. Female homicide victims are killed in a private residence in 52 to 75 percent of cases, while research suggests that homicides involving male victims are almost equally divided between public and private locations (Chimbos, 1993; Frye, Hosein, Waltermaurer, Blaney, & Wilt, 2005; Moracco et al., 1998; Mouzos, 1999; Wolfgang, 1958).

Prior findings additionally suggest a correlation between the sex of the victim and the underlying motive for the homicide. More specifically, male victims generally are more likely than females to be killed in the course of a profit-motivated crime or following a nondomestic altercation, while females are more likely than males to be killed in the context of a domestic dispute or to be the victim of a homicide motivated by jealousy (Mouzos, 1999; Tardiff, Gross, & Messner, 1986; Wolfgang, 1958).

Given that a large percentage of homicides, regardless of sex of victim, occur in the context of a dispute (domestic or otherwise), research on victim-precipitation may help one to better understand how these events unfold. Studies have found that in those homicides that were victim-precipitated, the vast majority of victims were male (89 to 94 percent) (Voss & Hepburn, 1968; Wolfgang, 1958). Though these studies compared victim-precipitated homicides with those that were not (rather than homicides with male versus female victims), their findings still suggested that homicides with male victims are much more likely to be victim-precipitated.

Though some studies have found no significant gender gap with regard to alcohol involvement (see, e.g., Pridemore & Eckhardt, 2008; Tardiff et al., 1986), the results of prior research have generally indicated that male victims of homicides are more likely than female victims to have been drinking at the time of the violent encounter. For instance, Goodman et al. (1986) found that male homicide victims were almost twice as likely as female victims to have been drinking—a finding supported by more recent studies in Canada (Fedorowycz, 1996) and Finland (Lunetta, Penttilä, & Sarna, 2001). Still other studies have provided support for the finding that male victims are more likely to have been drinking, but suggest that the gender gap is much smaller than indicated by the aforementioned studies (Mouzos, 1999; Shaw et al., 2006; Wilbanks, 1983b; Wolfgang, 1958).

Lastly, studies analyzing the correlation between sex of the victim and weapon choice have resulted in some inconsistent findings. Moracco et al.’s (1998) finding that female victims were less likely than males to be killed using a gun was supported by results from Kellermann et al. (1992) and Tardiff et al. (1986). Findings from these latter studies also suggested that females are more likely than males to be killed using blunt or bodily force. Tardiff et al. (1986), however, found that knives were more likely to be used in homicides involving female victims as compared with males, while Kellermann et al. (1992) found that the likelihood of a knife being used in a homicide was not correlated with the sex of the victim. Still other studies suggested that there are no significant differences between male and female victims with regard to type of weapon used (Mouzos, 1999; Wolfgang, 1958).

Summary of hypotheses

Based on this review of the empirical literature, several working hypotheses were generated. For example, it was expected that homicides involving male offenders will be more likely than those with female offenders to occur in the context of a profit-motivated crime, occur in a public place, involve a firearm, and involve a victim who is a friend or acquaintance of the offender. It was further anticipated that homicides involving female offenders will be found to be more likely than those involving male offenders to take place in a private residence, involve a victim and offender who are intimates, and be victim-precipitated. With respect to victims, it was expected that relative to males, the killing of female victims is more likely to be perpetrated by an intimate, occur in the home, and be carried out through the use of bodily force. By comparison, it was predicted that homicides involving male victims will be more likely than those with female victims to be victim-precipitated, involve a firearm, and involve an offender who is a friend or acquaintance of the victim. Finally, while the literature did not provide enough evidence for strong hypotheses, the rich data available for use in this study were employed to explore several other homicide characteristics by sex of victim and offender, including alcohol involvement, the time of the event, if the homicide was premeditated, and if the homicide was carried out in an attempt to hide another crime.

Data and method

The data for this study came from a unique set of narratives extracted from homicide court verdicts in the Udmurt Republic in 1989–1991 (101 cases) and 1998 (124 cases). Udmurtia is a typical Russian industrial region (i.e., state) located in the western Ural mountains. It has a population of about 1.6 million, a capital city (Izhevsk) of about 650,000, and 70 percent of the population lives in urban areas. There is a smaller proportion of ethnic Russians (60 percent) in the region relative to the nation as a whole. About 30 percent of the population, however, is ethnic Udmurt, which is an eastern Finnish group with hundreds of years of history in Russian culture. The age-standardized death rate due to homicide in the region (27.4 per 100,000 residents in 2000) is nearly exactly the same as that of Russia (27.5) as a whole (Russian Ministry of Health, 2001). Due to its representative nature on a number of characteristics, the Udmurt Republic has been the site of past (Shkolnikov & Chervyakov, 2000) and ongoing studies of Russian mortality, and it has also been the basis of previous studies of homicide (Chervyakov, Shkolnikov, Pridemore, & McKee, 2002; Pridemore, 2006).

Each one- to two-page narrative contained a description of the homicide event according to police and court records for cases resulting in a conviction (almost always including firsthand reports taken from offenders and/or witnesses). Among other items, they usually contained information on alcohol use, victim-offender relationship, motive, location, situational context, and type of weapon. In order to ensure consistency, all narratives were extracted from the records by one court secretary. The narratives were originally created as part of a United Nations Development Programme-sponsored project undertaken by Shkolnikov and Chervyakov (2000) to examine the Russian mortality crisis of the 1990s. They were provided by Dr. Shkolnikov to the second author of the present study, who translated them from Russian to English and coded them for analysis. In order to reduce errors and ensure reliability, each narrative was coded twice and the two records checked for consistency.

Three sets of analyses were carried out. The first compared the characteristics of homicide events when women were victims relative to those when men were victims (regardless of the sex of the perpetrator). The second was the same except for offenders. The third compared the characteristics of male offender-male victim events to all others.¹ In these analyses, each victim was employed as an incident. Thus, in those few cases where there were multiple victims, each victim was treated as an individual case and thus victim, offender, and event characteristics were coded for each.

Incident characteristics included measures of the proportion of events (1) with male and female offenders (in the victim model); (2) with male and female victims (in the offender model); (3) in which alcohol was involved (including victim, offender, both, neither); (4) in which a gun, knife, or bodily force was the primary means of assault; (5) that occurred between 6:00 p.m. and 6:00 a.m., during the weekend (i.e., Friday, Saturday, or Sunday), and during the winter months in the Udmurt Republic (i.e., November–April); (6) in which the victim and offender were intimates (i.e., current/former spouse or current/former boyfriend/girlfriend), family members (not including intimates), acquaintances (i.e., they had a loose connection and knew of each other in some way; this category did not include friends), or strangers (i.e., no prior contact at all between them); (7) in which male/female jealousy, rape, an acute argument (i.e., one that immediately preceded the homicide, not an ongoing argument or revenge for an earlier action), or some sort of profit motivated the offense (e.g., robbery, burglary, car theft) led to the homicide; (8) that occurred in someone’s (usually the victim’s or offender’s) home or in a public place (e.g., outside in a park or on the street, or inside in a bar or restaurant); (9) that were committed in order to hide another crime (e.g., rape or burglary); (10) that were victim-precipitated; and (11) that were premeditated. Victim-precipitation and premeditation were gauged by the second author based upon the information provided in the narratives.

In order to examine similarities and differences on these characteristics between women and men, logistic regression was employed to estimate crude odds ratios. Each characteristic acted as a single independent variable, while the dependent variable was the presence of a female victim (in Table 2), female offender (in Table 3), or male offender-male victim (Table 3). Full models were then estimated that provided adjusted odds ratios for each characteristic, controlling for the other incident characteristics (though this was impossible to do for the “female offenders” model, since there were only sixteen cases). Since some of the independent variables were dependent upon other independent variables (e.g., gun versus bodily force, acute argument versus profit motivated, etc.), only the following were employed when estimating adjusted odds ratios: male victim, bodily force, 6:00 p.m.–6:00 a.m., weekend, winter, acquaintance, acute argument, public place, if committed to hide another crime, victim-precipitated, and premeditated.

Table 2.

Comparison of homicide characteristics when female was victim relative to when male was victim (regardless of sex of perpetrator)

Characteristic	Sex of victim		Crude odds ratio (95% CI)	p-value	Adjusted odds ratio (95% CI)	p-value
	Female	Male
Male offender	93.8	94.7	0.84 (0.31–2.34)	.748	--
Female offender	6.2	5.3	--	--	--
Alcohola
Offender drinking	65.5	63.2	1.10 (0.66–1.81)	.724
Victim drinking	31.0	47.8	0.51 (0.31–0.84)	.009	0.35 (0.18–0.71)	.003
Both drinking	29.2	40.2	0.62 (0.37–1.03)	.064
Neither drinking	32.7	30.2	1.14 (0.68–1.90)	.625
Weaponb
Gun	3.5	12.5	0.26 (0.09–0.78)	.016	0.39 (0.10–1.56)	.181
Knife	42.5	48.8	0.77 (0.47–1.24)	.274
Bodily force	32.7	19.7	1.56 (0.89–2.73)	.122
Time
6:00 p.m.–6:00 a.m.	69.9	73.9	0.75 (0.43–1.32)	.320
Weekend	34.8	36.7	0.93 (0.56–1.53)	.763
Winter	43.3	45.0	0.95 (0.59–1.53)	.834
Relationshipc
Intimate	18.6	3.0	7.48 (2.73–20.50)	< .001	13.80 (3.86–49.44)	< .001
Family member	12.5	6.6	2.04 (0.89–4.68)	.091
Acquaintance	36.2	46.7	0.64 (0.39–1.05)	.078
Stranger	28.3	34.9	0.71 (0.43–1.20)	.201
Contextd
Jealousy	6.2	2.4	2.74 (0.78–9.59)	.115
Rape	15.9	0.6	32.02 (4.21–243.64)	.001
Acute argument	22.1	36.1	0.50 (0.29–0.87)	.013	0.98 (0.47–2.04)	.954
Profit	31.0	21.3	1.60 (0.93–2.75)	.088
Location
Home	75.2	45.2	3.72 (2.20–6.27)	< .001	2.82 (1.53–5.18)	.001
Public place	23.9	51.2	0.30 (0.18–0.50)	< .001
Hide crime	38.9	23.7	2.06 (1.23–3.45)	.006	1.98 (1.02–3.83)	.043
Victim precipitated	5.3	34.3	0.10 (0.04–0.25)	< .001	0.13 (0.05–0.35)	< .001
Premeditated	28.3	36.7	0.68 (0.40–1.13)	.136	0.53 (0.27–1.05)	.069

^aWhen “victim drinking” was replaced with “both drinking” in the full model, inferences for all other variables remained the same. For both drinking: adjusted odds ratio = 0.41 (CI: 0.21–0.82), p-value = .011.

^bWhen gun was replaced with bodily force, inferences for all other variables remained the same. For bodily force: adjusted odds ratio = 1.15 (CI: 0.55–2.39), p-value = .720.

^cWhen intimate was replaced (individually) with family, acquaintance, and stranger, inferences for all other variables remained the same, except that hide crime became nonsignificant (with p-values ranging from .075 to .170). Results for family and acquaintance were nonsignificant. For stranger: adjusted odds ratio = 0.48 (CI: 0.24–0.94), p-value = .033.

^dWhen acute argument was replaced (individually) with jealousy and profit, inferences for all other variables remained the same, and the results for both were nonsignificant. When it was replaced with rape, hide crime became nonsignificant and the p-value for premeditated (which is nearly significant in all other models) increased substantially (to .539). For rape: adjusted odds ratio = 40.79 (CI: 4.70–354.23), p-value = .001.

Table 3.

Comparison of homicide characteristics when female was offender relative to when male was offender (regardless of sex of victim)

Characteristic	Sex of offender		Crude odds ratio (95% CI)	p-value
	Female	Male
Male victim	56.3	59.9	0.85 (0.31–2.36)	.757
Female victim	43.8	39.7	--	--
Alcohol
Offender drinking	37.5	65.9	0.31 (0.11–0.88)	.028
Victim drinking	56.3	40.1	1.92 (0.69–5.32)	.209
Both drinking	25.0	36.3	0.58 (0.18–1.86)	.363
Neither drinking	31.3	31.1	1.01 (0.34–2.99)	.989
Weapon
Gun	0.0	9.4	--	--
Knife	50.0	46.2	1.16 (0.42–3.19)	.770
Bodily force	18.8	22.9	0.56 (0.21–2.81)	.699
Time
6:00 p.m.–6:00 a.m.	75.0	72.2	0.96 (0.30–3.10)	.951
Weekend	37.6	35.6	1.06 (0.37–3.01)	.991
Winter	31.3	44.9	0.56 (0.19–1.65)	.290
Relationship
Intimate	31.3	7.9	5.24 (1.66–16.50)	.005
Family member	31.4	7.4	5.61 (1.78–17.75)	.003
Acquaintance	18.8	43.9	0.29 (0.08–1.03)	.056
Stranger	18.8	33.3	0.45 (0.13–1.62)	.223
Context
Jealousy	0.0	4.1	--	--
Rape	0.0	7.1	--	--
Acute argument	12.5	31.5	0.31 (0.07–1.39)	.126
Profit	6.3	26.6	0.18 (0.02–1.41)	.103
Location
Home	93.8	54.9	12.33 (1.61–94.68)	.016
Public place	6.2	42.4	0.09 (0.01–0.69)	.021
Hide crime	18.8	30.5	0.53 (0.15–1.90)	.328
Victim precipitated	56.3	20.9	4.86 (1.73–13.64)	.003
Premeditated	31.3	33.6	0.90 (0.30–2.67)	.848

Results

The first column in Table 1 reveals overall frequencies for all homicide incidents analyzed here. For example, females were about 6 percent of all homicide offenders and 40 percent of all victims in this sample. The offender was drinking in about two-thirds of the homicides, the victim in about 40 percent, and both in 36 percent. A knife (or some other sharp object) was the most often used weapon, three-quarters of all incidents occurred between 6:00 p.m. and 6:00 a.m., and a little over one-third of the cases occurred during the weekend. In 42 percent of cases the victim-offender relationship was acquaintance, and in 33 percent of the cases the victim and offender were strangers. Acute arguments (30 percent) and profit (25 percent) were most often the situations in which the homicide incident occurred, and more than half of the incidents occurred in someone’s home. Finally, 30 percent of the homicides were carried out to hide another crime, 23 percent were victim-precipitated, and one-third were premeditated. Columns two through four of this table are addressed in the discussion section below.

Table 1.

Overall frequencies

Characteristic	Overall	Female victim	Female offender	Male victim/male offender
Female offender	5.7	6.2	--	--
Female victim	39.9	--	43.8	--
Male offender	94.3	93.8	--	100.0
Male victim	59.7	--	56.3	100.0
Alcohol
Offender drinking	64.3	65.5	37.5	64.3
Victim drinking	41.0	31.0	56.3	45.4
Both drinking	35.7	29.2	25.0	40.0
Neither drinking	31.1	32.7	31.3	31.3
Weapon
Gun	8.9	3.5	0.0	13.2
Knife	46.5	42.5	50.0	50.3
Bodily force	22.7	27.4	18.8	19.5
Time
6:00 p.m.–6:00 a.m.	75.7	69.9	75.0	77.6
Weekend	36.1	34.8	37.6	36.1
Winter	44.2	43.3	31.3	44.5
Relationship
Intimate	9.3	18.6	31.3	0.0
Family member	8.8	12.5	31.4	5.6
Acquaintance	42.4	36.2	18.8	48.1
Stranger	32.5	28.3	18.8	36.9
Context
Jealousy	3.9	6.2	0.0	2.5
Rape	6.7	15.9	0.0	0.6
Acute argument	30.4	22.1	12.5	37.5
Profit	25.4	31.0	6.3	22.5
Location
Home	57.1	75.2	93.8	42.8
Public place	40.4	23.9	6.2	53.5
Hide crime	29.8	38.9	18.8	23.9
Victim precipitated	22.9	5.3	56.3	32.1
Premeditated	33.5	28.3	31.3	38.6

Table 2 compares homicide incidents in which a female was the victim to those in which a male was the victim, regardless of the sex of the perpetrator. The first two columns show the proportion of incidents in the female victim and male victim samples with each of the incident characteristics. The rest of the columns show the corresponding crude and adjusted odds ratios, together with the p-values for the odds ratios. The odds ratios were calculated here to represent the homicides with female relative to male victims. For example, female victims were significantly less likely to be drinking at the time of the event than male victims (adjusted odds ratios = 0.35, p = .003), which largely accounted for the finding that when a female was the victim, it was significantly less likely that both the offender and victim were drinking (adjusted odds ratio = 0.41, p = .011). There were no significant differences between these two samples for the use of a knife or bodily force. On the other hand, a firearm was less likely to be used when a female was victimized. Although the adjusted odds ratio was not significant (adjusted odds ratio = 0.39, p = .181), it appeared that this was the result of a low number of cases and thus a wide confidence interval.

Examining the relationships between victims and offenders, females were significantly more likely than males to be victimized by an intimate (adjusted odds ratio = 13.80, p < .001) and significantly less likely than males to be victimized by a stranger (adjusted odds ratio = 0.48, p = .033). The only significant difference between female and male victims in terms of context was for rape (adjusted odds ratio = 40.79, p = .001). Females were also more likely than males to be murdered as a result of jealousy, but this was another example of where a small number of cases created wide confidence intervals and thus a nonsignificant finding. Although females are less likely than males to be victimized in the context of an acute argument, the results for the adjusted odds ratio suggested that this difference was due to other characteristics of the homicide event. As expected, females were more likely than males to be murdered inside someone’s home (adjusted odds ratio = 2.82, p = .001). They were also more likely to be murdered in order to hide another crime (adjusted odds ratio = 1.98, p = .043), usually rape. Finally, homicides with female victims were less likely to have been victim precipitated (adjusted odds ratio = 0.13, p < .001) and premeditated (adjusted odds ratio = 0.53, p = .069).

Table 3 compares the characteristics of female-offender relative to male-offender homicides, regardless of the sex of the victim. Since the number of female offenders was so small (n = 16), only crude odds ratios were estimated and the drawing of strong conclusions was discouraged. Nevertheless, it is worthwhile to point out a few differences. Male offenders, for example, were more likely than female offenders to have been drinking at the time of the homicide event. They were also more likely than female offenders to murder an acquaintance and to carry out their crime in a public place. Further, although odds ratios could not be generated for the rape and jealousy characteristics since no female offenders carried out their crimes under these circumstances, it could be seen that several male offenders acted on these motives. Similarly, even though the small number of female offenders who committed their crime during an acute argument or as the result of a profit motive made it difficult to reach a significant p-value, the results suggested that males were significantly more likely than females to commit their homicides in these two contexts. Female offenders, on the other hand, were significantly more likely than male offenders to murder an intimate and to murder a (non-intimate) family member. Females were also more likely than males to carry out their crimes in someone’s home. Finally, female-perpetrated homicides were significantly more likely to have been victim precipitated, and these were almost always in the context of a victim of domestic violence finally responding herself with lethal violence. The latter was based upon the interpretation of the event by the police and court as provided within the qualitative narratives.

Table 4 shows the results when homicides where both the victim and the offender were males compared to all other homicides. Although the study was mainly concerned with examining homicides with female involvement, and in comparing such homicides with those without females, this final analysis was carried out in view of the fact that male-on-male events comprised the majority of homicides. Thus an examination of these events compared to all others could reveal what might be unique features of male-on-male homicide events that were distinct from other homicides. Relative to all other homicides, the male-male events were more likely to have both the offender and the victim drinking at the time of the event (adjusted odds ratio = 1.76, p = .070). Next, it appeared as if these male-on-male homicides were more likely than all others to involve a gun, though the small number of cases (of all other homicides) made it difficult to determine if this difference was significant. Examining the results for victim-offender relationships, male-male homicides appeared to be more likely than other homicides to occur between acquaintances (adjusted odds ratio = 1.62, p = .086) and more likely to occur between strangers (adjusted odds ratio = 2.01, p = .039). As would be expected, male-male homicides were significantly less likely than all other homicides to occur in the context of a rape (adjusted odds ratio = 0.03, p = .033), and significantly more likely to occur in the context of an acute argument (adjusted odds ratio = 2.04, p = .034). Finally, the male-male homicides were also more likely than all the other homicides to occur in a public place, to be victim-precipitated (adjusted odds ratio = 2.28, p = .026), and to be premeditated (adjusted odds ratio = 2.13, p = .020).

Table 4.

Comparison of homicide characteristics when both victim and offender were male relative to all other incidents

Characteristic	Male victim and male offender?		Crude odds ratio (95% CI)	p-value	Adjusted odds ratio (95% CI)	p-value
	Yes	No
Male victim	100.0	7.3	--	--
Female victim	--	91.9	--	--
Alcohola
Offender drinking	64.3	64.2	1.00 (0.61–1.64)	.992
Victim drinking	45.4	35.5	1.51 (0.92–2.46)	.101	1.45 (0.78–2.68)	.236
Both drinking	40.0	30.1	1.55 (0.94–2.55)	.085
Neither drinking	31.3	30.9	1.02 (0.61–1.69)	.949
Weaponb
Gun	13.2	3.2	4.53 (1.51–13.56)	.007	2.45 (0.71–8.47)	.158
Knife	50.3	41.5	1.43 (0.89–2.30)	.140
Bodily force	19.5	26.8	0.66 (0.38–1.16)	.146
Time
6:00 p.m.–6:00 a.m.	77.6	73.1	1.28 (0.73–2.23)	.390
Weekend	36.1	36.0	0.99 (0.60–1.61)	.960
Winter	44.5	43.9	1.02 (0.64–1.64)	.037
Relationshipc
Intimate	0.0	21.1	--	--
Family member	5.6	13.0	0.40 (0.17–0.94)	.035	0.64 (0.24–1.68)	.364
Acquaintance	48.1	35.0	1.77 (1.09–2.87)	.021
Stranger	36.9	26.8	1.62 (0.97–2.71)	.064
Contextd
Jealousy	2.5	5.7	0.43 (0.12–1.49)	.180
Rape	0.6	14.6	0.04 (0.01–0.28)	.001
Acute argument	37.5	21.1	2.26 (1.32–3.88)	.003	2.04 (1.06–3.94)	.034
Profit	22.5	29.3	0.71 (0.41–1.21)	.207
Location
Home	42.8	75.6	0.24 (0.14–0.41)	< .001	.028 (0.16–0.50)	< .001
Public place	53.5	23.6	3.72 (2.21–6.26)	< .001
Hide crime	23.9	37.4	0.53 (0.31–0.88)	.015	0.66 (0.35–1.24)	.198
Victim precipitated	32.1	11.4	3.67 (1.92–7.04)	< .001	2.28 (1.10–4.71)	.026
Premeditated	38.6	26.8	1.72 (1.03–2.86)	.039	2.13 (1.13–4.03)	.020

^aWhen “victim drinking” was replaced with “both drinking” in the full model, inferences for all other variables remained the same. For both drinking: adjusted odds ratio = 1.76 (CI: 0.96–3.24), p-value = .070.

^bWhen gun was replaced (individually) with knife and bodily force, inferences for all other variables remained the same. Results for knife and bodily force were nonsignificant.

^cWhen family was replaced (individually) with acquaintance and stranger, inferences for all other variables remained the same (though the p-value for victim drinking dropped substantially (to .11) in the latter model). For acquaintance: adjusted odds ratio = 1.63 (CI: 0.93–2.84), p-value = .086. For stranger: adjusted odds ratio = 2.01 (CI: 1.03–3.89), p-value = .039.

^dWhen acute argument was replaced (individually) with jealousy and profit, all other inferences remained the same, and the results for both were nonsignificant. When it was replaced with rape, all inferences remained the same, except for premeditated (p = .268). For rape: adjusted odds ratio = 0.03 (CI: 0.01–.027), p-value = .033.

Discussion

The empirical literature has long suggested that certain event-level homicide characteristics are associated with sex of victim, offender, or both. This study was an effort to extend this body of literature by examining female involvement in homicide incidents in a non-Western culture.

Turning first to Table 2, it can be seen that homicides involving female victims differed significantly from those in which the victim was male on several characteristics. For instance, female-victim homicides were nearly three times more likely to occur in the home and almost fourteen times more likely to be committed by an intimate than homicides involving male victims. Femicides were also much less likely than male-victim homicides to be victim-precipitated. The overall pattern of these findings was consistent with those from prior studies in North America, Europe, and Australia (Chimbos, 1993; Craven, 1996; Moracco et al., 1998; Mouzos, 1999; Rennison & Welchans, 2000; Wilson & Daly, 1994; Wolfgang, 1958). At the same time, the results in the current study supported Gondolf and Shestakov’s (1997) finding that the gender gap in intimate homicide in Russia is more pronounced than in many other industrialized countries. The male-to-female victim ratio for intimate homicide is about 1:2 in the U.S., 1:2.5 in Scotland, and 1:3.2 in Canada and Australia (Gondolf & Shestakov, 1997; Wilson & Daly, 1992, 1994). By comparison, these data indicate that for every male victim of intimate homicide in the Udmurt Republic, more than four women are killed by an intimate partner.

Several factors might be contributing to this sizeable gender gap in Russian intimate homicides. First, domestic violence may have increased (in frequency and/or severity) in the years following the fall of the Soviet Union as families were forced to deal with the strains of social, political, and economic transition (Johnson, 2005; Pridemore, 2006). At the same time, widespread housing shortages and the residence permit requirements of some cities severely limited mobility for many Russians (Gondolf & Shestakov, 1997; Johnson, 2001). Compounding the problem, services and shelters for victims of domestic violence, while currently growing in number, were virtually nonexistent in Russia during the years from which these data were collected, leaving victims of domestic violence with few options for escape even if they perceived that their lives were at risk (Gondolf & Shestakov, 1997; Johnson, 2005; see Dugan, Nagin, & Rosenfeld, 1999, for an analysis of the role of domestic violence resources in reducing intimate partner homicide in the United States). Finally, despite the lack of options for escape, Russian women may be less likely to “kill their way out” of abusive relationships than their counterparts in other countries—in other words, the larger gender gap in Russia may be a product of a small proportion of male victims rather than (or in addition to) an unusually high proportion of intimate partner femicides (Messner & Savolainen, 2001).

The finding that female- and male-victim homicides did not differ significantly with regard to being motivated by jealousy was not necessarily at odds with the general pattern of female-victim homicides described thus far. Though male offenders committed all of the jealousy-motivated homicides in this sample, the target of their lethal violence may have been the “other man” as often as it was a wife or girlfriend suspected of cheating.

Relative to homicides with female victims, those with male victims were significantly more likely to involve a drinking victim and more likely to occur outside the home. In addition, when controlling for the other variables in the model, male-victim homicides were twice as likely to be committed by strangers and almost eight times more likely to be victim-precipitated. These findings were consistent with those from prior studies comparing homicide characteristics by sex of victim (Foote, 1998; Jurik & Winn, 1990; Kellermann & Mercy, 1992; Tardiff et al., 1986; Wolfgang, 1958; Zahn & Sagi, 1987). They were also consistent with the general pattern of results from previous research on alcohol and homicide in Russia (using similar data: Pridemore & Eckhardt, 2008) and elsewhere (Goodman et al., 1986; Lunetta et al., 2001; Mouzos, 1999; Shaw et al., 2006). Taken together, these findings also made sense contextually. Heavy drinking is a predominantly male activity in Russia (Malyutina, Bobak, Kurilovitch, Nikitin, & Marmot, 2001). Moreover, an intoxicated individual may be both more likely to instigate a dispute with someone he does not know (or does not know well) and the alcohol may render him less capable of defending himself from a violent counterattack (Pridemore & Eckhardt, 2008). It is also possible that women (whether drinking or not) are less likely to instigate an argument or physical fight, or when they do, their targets are less likely to respond with lethal violence.

Though a greater percentage of male-victim homicides occurred in the context of an acute argument compared to those with a female victim, the difference was not statistically significant. As suggested by the research of Mouzos (1999), Tardiff et al. (1986), and Wolfgang (1958), it may be that the broad nature of the “acute argument” category served to mask some important differences in motivation by sex of victim. In other words, it is possible that female-victim homicides were more likely to occur in the context of a domestic dispute and male-victim homicides to arise out of some other type of altercation (as prior studies suggest), but the result obtained by including both types of acute argument under one classification suggested any difference was nonsignificant. Future research might consider separating domestic disputes from other types of acute arguments in order to provide a more detailed understanding of the possible correlations between victim sex and the context that gives rise to the criminal event.

As expected, homicides involving female victims were significantly more likely than those with male victims to occur in the context of a rape. This was not surprising as, cross-culturally, women comprise the vast majority of rape victims, and Russia is no exception (Gavrilova, Gavrilov, Semyonova, Evdokushkina, & Ivanova, 2005). The findings also showed that homicides involving female victims were almost twice as likely as those involving male victims to be carried out in order to hide another crime. This difference was in all likelihood correlated with the finding that femicides are more likely to occur in the context of a sexual assault. The descriptions from the homicide narratives made it clear that the majority of the rape-related homicides in the sample were carried out in order to prevent the victim (who could possibly identify her attacker) from reporting the crime to authorities, and as noted above, these victims were overwhelmingly female.

Rape-related homicides are not the sole (and may not even be the primary) reason that femicides were more likely than male-victim homicides to be undertaken for the purpose of concealing some other criminal activity. The large percentage of profit-motivate femicides is also a factor. Though the difference was not significant, both the higher percentage of profit-motivate homicides involving female relative to male victims and the fact that profit-motivated killings comprised the largest percentage of female-victim homicides were notable because they were inconsistent with findings from previous studies, which had overwhelmingly found that homicides committed for pecuniary gain were more likely to involve male victims (Mouzos, 1999; Tardiff et al., 1986; Wolfgang, 1958). The unusually high percentage of profit-motivated femicides in Russia may be related to the limited availability of firearms in that country. Without access to the “equalizing” power of a firearm, profit-motivated offenders may be less likely to take on a target whose size and/or strength appears equal to or greater than their own and may be more likely to choose as a target someone they perceive as being relatively weak—namely, a lone female and/or elderly person. Examples of this were not uncommon in the narratives:

NARRATIVE [98038]²: The defendant was serving time in a penal colony and learned from a fellow prisoner that in the village there were two women who lived alone and who had a lot of money. When he was released [from the colony], he went to the village and was able to find the house.

NARRATIVE [98135]: The defendant decided to murder the victim out of “profit” motives in order to take her property. He went to the victim’s home, knowing she was an elderly woman who could not protect herself.

The possibility of a correlation between profit-motivated homicide, firearm availability/use, and sex of victim was consistent with the finding that female-victim homicides were less likely than those with male victims to involve a gun (see also Felson & Messner, 1996). In fact, none of the profit-motivated female-victim homicides in this Russian sample were committed with a firearm.

Female-victim homicides were also less likely than those with male victims to be premeditated, which was consistent with the empirical literature and surprising here only because it was expected that premeditation would be correlated with profit-motivated homicides. A closer look at the narratives revealed that the profit-motivated homicides in the current sample were often not premeditated. Such situations were often the case when an offender attempted to burglarize an “empty” residence only to be surprised by the owner, or when a robbery victim unexpectedly put up resistance. In other words, the theft may have been premeditated, but the homicide itself had not been part of the offender’s original plan (Block, 1977).

Table 3 compares homicides committed by female offenders with those by male offenders. The results indicate that the two groups are similar with respect to some characteristics. For instance, female perpetrated killings did not differ significantly from those committed by males in terms of sex of victim or time of incident. Consistent with the empirical literature, both male and female offenders more often targeted male victims (Carcach & Conroy, 2001; Jurik & Winn, 1990; Mouzos, 1999; Wilbanks, 1983a) and committed their crimes during the late night or early morning hours (Tardiff et al., 1986). In addition, male- and female-perpetrated homicides were about equally likely to occur during a weekend, to occur during the winter months, and to be premeditated.

Also in accordance with findings from previous research, homicides involving female offenders were found to differ significantly from those perpetrated by male offenders on a number of situational characteristics. Relative to male-perpetrated homicides, for example, those committed by female offenders were twelve times more likely to occur in the home and were much more likely to involve a victim who was an intimate or family member³—findings that were consistent with those from previous studies in North America and Europe (Chimbos, 1993; Jurik & Winn, 1990; Kellermann & Mercy, 1992; Mann, 1996; Miethe & Regoeczi, 2004; Wilbanks, 1983a, 1983b). Female-perpetrated homicides were also significantly more likely to be victim-precipitated than those with male offenders, which supported the earlier findings of Goetting (1988), Jurik and Winn (1990), and Wilbanks (1983a, 1983b). Taken together with evidence of a high domestic violence rate in Russia, these findings also made sense contextually. With few shelters or other resources for victims of domestic violence, some women may determine that the best chance for survival is the death of their abuser, as exemplified in the following narrative:

NARRATIVE [98129]: Having heard the threats [made by her husband against her] and remembering his abusive behavior toward her and the children, she decided to kill her husband. At about 2:00am that night, having waited for [her husband] to fall asleep, she took an axe from the entryway…By the light of the TV, which was still on, she found her sleeping husband and struck him in the neck with the axe, from which [he] died immediately.

Though incidents like this one made up a small portion of total homicides in the sample, they comprised a fairly substantial proportion (25 percent) of those perpetrated by female offenders.

Compared with homicides committed by female offenders, male-perpetrated homicides were significantly more likely to involve a victim and offender who were acquaintances and an offender who had been drinking. These results were in accordance to previous findings on the correlation between victim-offender relationship and sex of offender (Chimbos, 1993; Jurik & Winn, 1990; Kellermann & Mercy, 1992; Wilbanks, 1983a, 1983b), as well as with the overall pattern of findings from prior research on alcohol and homicide in Russia (Pridemore & Eckhardt, 2008). Also as predicted, homicides involving male offenders were more likely to occur in a public place, which in the context of Russia was most often an outdoor setting—in the street or in a park or wooded area—rather than a bar, restaurant, or other indoor public location. In this respect, Russia is similar to both Greece (Chimbos, 1993) and the United States (Wilbanks, 1983b).

Male-offender homicides were more likely to be committed with a firearm. Guns were utilized in nearly 10 percent of homicides committed by male offenders, whereas no female-offender homicides were committed using a firearm. Though the sex difference in firearm use was consistent with the empirical literature, most previous studies had found that a greater percentage of homicide offenders (of both sexes) used a firearm to commit their crime—generally in the range of 25 to 60 percent for female-offender homicides and 35 to 75 percent for those with male offenders (Chimbos, 1993; Goetting, 1988; Kellermann & Mercy, 1992; Wilbanks, 1983a, 1983b; Wolfgang, 1958). Much of this previous research, however, had utilized data from countries where gun ownership/use was much more common than it is in Russia, where handgun ownership is restricted. The low level of gun ownership in Russia likely contributed to the finding that male- and female-perpetrated homicides did not differ significantly with respect to the use of knives or bodily force, as would be expected based on a review of the literature. Given that most homicides in the current sample were not premeditated and few Russians carry guns, offenders of both sexes likely utilize whatever weapon is convenient at the time of the encounter— and the latter two methods are equally available to offenders of both sexes in the form of a kitchen knife, household axe, fists, and/or feet.

Though the small number of female homicide offenders in the sample precluded drawing strong conclusions from the findings, it is worth noting that all of the findings in this study with respect to offenders were generally consistent with those of previous research. This suggests that the correlations between offender sex and specific homicide characteristics (particularly victim-offender relationship and location) may be less moderated by cultural differences than those with respect to victim sex.

As in most other countries, the majority of Russian homicides (56 percent) are male-on-male events (see also Carcach & Conroy, 2001; Chimbos, 1993; Wilbanks, 1983a; Wolfgang, 1958), and these data provided a unique opportunity to see if the situational characteristics of male-on-male homicides in Russia are distinct from homicides in which a female is involved (as victim, offender, or both). Table 4 provides the results of this comparison. The findings revealed that homicides in which both the victim and offender were male differed from all other (female-involved) homicides with respect to several characteristics. Compared with all other victim-offender pairings, male-on-male homicides were more likely to occur in a public place, more likely to arise in the context of an acute argument, and more likely to involve a victim and offender who had both been drinking. This pattern of findings was consistent across all three models, suggesting that they were strongly correlated with male involvement in fatal violence. Taken together, these results also offered support for Lofland’s (1969) theory that certain locations serve as “facilitating places” for lethal violence—or more specifically, that character contests in public locations are a facilitating context for male-on-male homicides, particularly when combined with alcohol consumption (Luckenbill, 1977; Miethe & Regoeczi, 2004; Polk, 1994).

In addition, homicides in which both participants were male were also more likely to involve a victim and offender who were acquaintances or strangers. This was consistent with the findings of Decker (1993), Silverman and Kennedy (1987), and Zahn and Sagi (1987), and offers some support for the possibility that closer personal relationships (for instance, close friends or family) may serve to provide a measure of insulation from lethal violence, particularly when the confrontation occurs in public space (Decker, 1993).

While an acute argument was the motivating factor in the largest percentage of male-on-male homicides, this was not necessarily inconsistent with the finding that homicides involving a male victim and offender were more likely to be premeditated. More than 60 percent of male-on-male homicides occurred in a context other than an acute argument, and many of these motivations—profit, revenge, jealousy, and ongoing argument—frequently involved premeditation. In addition, incidents that arise out of an acute argument do not always lack premeditation. The narratives contained a number of cases in which two men became embroiled in an acute argument and/or physical fight after which (1) one participant, still angry or embarrassed, decided to return and kill the other or (2) one participant (usually the instigator) became fearful the other would report the assault to the authorities and decided later to kill the victim in order to prevent this from occurring (see Felson & Messner, 1996; Weaver et al., 2004).

Not surprisingly, male-on-male homicides were significantly less likely than female-involved homicides to occur in the context of a rape. It should be noted, however, that (attempted) rape was involved in several of the homicides in the sample in which both the victim and offender were male. Further, given the acute stigma still associated with homosexuality in Russia, male offenders may be less likely to disclose that rape was a motivating factor when questioned by authorities.

Finally, although the difference was not found to be statistically significant due to the small number of gun-related killings, it appeared that male-on-male homicides were also more likely to involve a firearm than female-involved homicides. The apparent correlation between male-involved lethal violence and gun use was consistent across all three models in the study, but did not suggest a specific causal mechanism. It may be that a male offender is more likely to take up a gun when confronting a male victim because another male is perceived as a greater physical threat than a female (Felson & Messner, 1996). Alternately, possessing the power of a firearm may provide an offender with more freedom to choose a male target for confrontation in the first place.

Theoretical implications

The main objective of this study was primarily descriptive. To be of serious scientific value, however, repeated descriptive findings from different studies must eventually be employed as initial observations upon which to generate theory. In this case, by extending the type of research to a culturally different and previously understudied nation, this study provided an opportunity for incorporating findings from a non-Western culture with those from prior analyses in the larger project of building a broadly informed grounded theory of gendered involvement in homicide. With this in mind, two theoretical implications of the findings are worth noting here.

In their study of homicide in the United States and Finland, Messner and Savolainen (2001) determined that the larger gender gap in acquaintance homicide in Finland was largely related to gendered patterns of alcohol consumption in that country. The relationship between alcohol and violence in Russia has been noted in previous studies (Pridemore, 2004, 2006; Pridemore & Eckhardt, 2008), and like Finland, Russia is characterized by heavy episodic (‘binge’) drinking (Pridemore, 2004; World Health Organization, 2004). Russia differs from Finland, however, with respect to the sex distribution of such drinking. In the Russian population as a whole, more males than females engage in heavy drinking. Yet for the group most likely to be at risk of homicide offending and victimization—younger adults—the gender gap with respect to binge drinking is virtually nonexistent (World Health Organization, 2004). Based on Messner and Savolainen’s (2001) hypothesis, the findings reported here were precisely what one would expect: very narrow gender gaps both with respect to acquaintance killings and alcohol involvement in homicide. These findings from Russia, taken in conjunction with those from the U.S. and Finland (Messner & Savolainen, 2001), offer support for the theory that cross-national variations in the gender gap for acquaintance killings are associated with cultural differences in drinking patterns. More generally, it also suggests that the relationship between alcohol and violence is moderated by situational factors, such as victim-offender relationship.

Numerous studies had also sought to explain why the size of the gender gap in intimate partner homicides varies cross-nationally. Among such studies, there was general agreement on two key points. First, cross-national and temporal variations in the gender gap owe more to variations in the extent to which women kill male intimates than the reverse. In other words, a small gender gap is most often the result of larger numbers of women resorting to lethal violence against a male intimate (as opposed to a reduction in intimate femicide). Second, women are more likely to take lethal action against an intimate when they perceive it to be necessary to their own safety and/or that of their children (Dugan et al., 1999; Wilson & Daly, 1992). Based in part on these observations, it has been theorized that there is a causal relationship between reduced exposure to violent relationships and lower rates of female-perpetrated intimate homicide, and thus wider gender gaps in this category of lethal violence (Dugan et al., 1999; Messner & Savolainen, 2001).

It is unlikely, however, that the smaller number of female offender-male victim intimate partner homicides in Russia can be explained by exposure reduction given that two of the three factors most likely to contribute to a reduced exposure to domestic violence—greater access to domestic violence services and increased economic status of women (Dugan et al., 1999)—were largely absent in Russia during the years examined here (see LaFont, 2001; Rhein, 1998, for analyses of women’s social and economic status since Perestroika). According to the theory of exposure reduction, the social, political, and economic climate of Russia should place women at a heightened, rather than reduced, risk of exposure to domestic violence. Yet, the findings in the present study showed that the gender gap in intimate homicide in Russia is closer to that of Finland (low exposure, large gender gap) than it is to that of African Americans in the U.S. (higher exposure, very small gender gap) (Messner & Savolainen, 2001).

Accounting for Russia’s deviation from theoretical expectations raises some important questions for future research. Given their increased risk of exposure to intimate violence, what factors contribute to Russian women’s reluctance to ‘kill their way out’ of abusive relationships? Does Russia’s declining domesticity reduce exposure to violent relationships to an extent that it would offset deficiencies with regard to economic independence and social services? As it stands, the failure of exposure reduction theory to account for the Russian findings suggests that there may be limits to the theory’s generalizability outside of Western contexts. Alternatively, it may be that the usefulness of exposure reduction theory is limited to explaining variations in intimate homicide within one country over time and it is simply unable to sufficiently account for cross-national differences.

Methodological limitations

It is important to note a few issues that might limit the scope of these findings and the interpretations of them. First, a relatively small number of cases (225) were examined in this study. While this number was not restrictively small for the type of analysis carried out here, the number of cases in a given cell with a particular characteristic and a particular sex could be small and thus reduced the efficiency of the estimates.

Second, in homicide events with multiple victims, each victim was coded as an individual incident and counted as a separate observation. This could potentially lead to biased estimates since such observations would have very similar, if not identical, situational characteristics. Though the number of multiple-victim homicides in the present sample was small and therefore unlikely to have a significant impact on the findings, the sensitivity of the results to this issue were tested by (1) estimating a model in which only the first victim in each multiple-victim homicide was included as an observation and (2) estimating a model in which all observations were retained but multiple-victim homicides were weighted based on the number victims. Inferences resulting from both alternative models did not vary from those presented above.

Finally, the homicide narratives were extracted from court verdicts in which the perpetrator was convicted and may not be representative of all homicides in the region. For example, certain types of homicides—such as those that are mafia-related or in which the victims were homeless or sex workers—may be less likely to be reported, more difficult to solve, and less likely to be resolved in court, and the empirical literature provides some evidence that characteristics of unsolved homicides may differ systematically from those in which the cases are cleared and/or a verdict reached (Decker, 1993, 1996; Regoeczi & Miethe, 2003).

Conclusion

This study utilized detailed narratives of Russian homicides to examine the differences between the characteristics of male- and female-involved lethal violence. It was found that males and females in Russia tend to kill and be killed in different situational contexts. For example, homicides in which a female was the victim or offender were more likely (than homicides in which a male was the victim or offender) to occur between intimates and to occur in the home, whereas homicides involving males (as victim, offender, or both) were more likely to occur in a public place, to be alcohol-related, to involve a firearm, and to involve a victim and offender who did not know each other well, if at all. The comparison of homicides by victim sex also indicated that femicides were more likely to occur in the context of rape and to be carried out in order to hide another crime, while male-victim homicides were more likely to involve strangers and be victim-precipitated. By utilizing narrative data to examine the characteristics unique to male-on-male lethal violence, it was also found that these homicides differ from those that involve a female participant in several respects: they are more likely to involve acquaintances or strangers, to arise in the context of an acute argument, to be victim-precipitated, to occur in a public place, to involve a firearm, to be alcohol-related, and to be premeditated.

The careful examination of situational characteristics of homicide is important to understanding the nature of violent interactions, as is the extension of homicide research to countries and cultures beyond the U.S. and western Europe. Before now, such research was rarely possible in eastern Europe due to the lack of systematic data. Through the use of unique and newly available narrative data, this research represented one of the first studies of its kind in eastern Europe. These findings provided important knowledge of Russian homicide events while also allowing for the first time a comparison of results from Russia with those from Western nations.