Punishing femicide: Criminal justice responses to the killing of women over four decades

Abstract

Spanish

French

The way in which nation states respond to femicide has become the focus of much attention in the past decade. The establishment of specialized police and prosecution units has been recommended and some countries have implemented specific legislation or criminal offences specific to femicide. Part of the challenge in moving beyond these legislative and policy initiatives is the dearth of reliable data that show how states are actually punishing crimes of femicide on the ground. Using data that document punishment outcomes in cases of femicide over four decades in Canada’s most populous province, this article examines how punishments compare for female and male homicide victims, across femicide subtypes and over time. Results show that cases involving female victims attract more punitive court responses overall than cases with male victims. Second, intimate and familial femicides are treated more leniently at several stages than other femicides. Finally, there have been positive changes in the punishment of femicide over time, paralleling legislative and policy responses to violence against women in Canada. Priorities for future research that address the role played by dominant stereotypes in punishment related to particular types of femicide as well as some women’s increased risk are highlighted.

Keywords

Femicide gender intimacy punishment violence

Introduction

The way nation states respond to femicide has become the focus of international attention, particularly in Latin America where more than half the countries with high femicide rates are located (Laurent et al., 2013; Nowak, 2012). However, no country is free from this type of violence. For example, in Canada, inadequate state responses as well as historical and current impacts of colonization have been identified as contributors to the high femicide risk faced by Aboriginal women (NWAC, 2010). This underscores the need to understand how states are responding to femicide, regardless of world region. The establishment of specialized investigation and prosecution units has been recommended by the Special Rapporteur on violence against women to address the perceived impunity for those who perpetrate femicide (United Nations, 2011). Some countries have passed legislation pertaining to femicide or codified femicide as a crime (Laurent et al., 2013). These are positive steps because it is recognized that those who impose the law must recognize the seriousness of violence before society can effectively respond. The challenge, however, in moving beyond legislative and policy initiatives is the dearth of reliable data documenting how states are actually responding on the ground. This article begins to address this gap by examining three research questions using four decades of femicide data from Canada’s most populous province:

How do punishments compare for cases with female and male homicide victims in general?

Do punishments for femicide vary by the relationship between victims and killers?

Have punishments for femicide changed over time?

For this study, femicide is defined as the killing of women by men, consistent with the definition used by most international quantitative data because such acts are the most obvious and easy to document for prevalence and comparative purposes (Alvazzi del Frate and Nowak, 2013; UNODC, 2011). The debate as to ‘whether the word femicide should encompass all murders of women or, alternatively, be restricted to some’ is ongoing (Mujica and Tuesta, 2014: 5). Given recognized difficulties in establishing motive and understanding of what would be ‘femicidal’ offences (Campbell and Runyan, 1998; Widyono, 2008), coupled with the lack of consensus on gender-based indicators (Mujica and Tuesta, 2014), this broader, albeit less nuanced, definition is arguably the most appropriate for the first systematic examination of punishments for femicide. Following a brief overview of femicide in Canada, the current state of knowledge with respect to femicide and punishment is summarized. Given the paucity of research that has examined femicide specifically, the primary focus is the ‘female victim effect’ in punishment outcomes.

Femicide in the Canadian context

There are no official national data on femicide in Canada; however, in 2013, Canada had the fifth highest homicide rate (1.44 per 100,000) among 17 similarly situated countries (Cotter, 2014: 3).¹ Men outnumbered women as victims (71% and 29% respectively), but were also the majority of perpetrators (88%). While women have most to fear from current/former male partners (Cotter, 2014; WHO, 2005), homicide rates have declined during the past 30 years in Canada. Much of this decrease is attributed to decreasing intimate partner homicide (IPH). In 2009, female rates were one third of those experienced by Canadian women in 1979 (Hotton Mahoney, 2010: 17), although women’s risk remains three to five times higher than that of men. Women’s risk of death at the hands of their male partners was also underscored in a study examining femicide in Ontario from 1974 to 1994 (Dawson and Gartner, 1998; Gartner et al., 1999). This research showed that, during this period, 1206 women aged 15 and older were killed and, in 1120 (93%) of the solved cases, the killers were current/former male partners. The authors further showed that, during a period when women’s risk from strangers and men’s risk from spouses decreased, women’s risk from male partners increased (Gartner et al., 1999: 156).

For some women, male partners may not necessarily pose the greatest risk, however. The Sisters in Spirit initiative launched in 2005 by the Native Women’s Association of Canada documented that, by 2010, over 580 Aboriginal women and girls across Canada were murdered or went missing (NWAC, 2010: i). Among the findings was that Aboriginal women were more often killed by male acquaintances or strangers than by partners, but official statistics indicate that Aboriginal women remain eight times more likely to experience IPH than non-Aboriginal women (Statistics Canada, 2006). Nearly half of these Aboriginal femicides remain unsolved and no charges were laid in about 40% of the cases (NWAC, 2010; Pearce, 2013). In contrast, official figures show that, in 2013, about 76% of homicides were cleared and charges laid (Cotter, 2014), declining from about 95% in the past four decades (Hotton Mahony and Turner, 2012: 5). The Royal Canadian Mounted Police (RCMP) reported clearance rates that were higher than NWAC figures, ranging from 80 to 100%, but it was not reported what proportion resulted in charges or convictions (RCMP, 2014: 15). Despite repeated calls for a national inquiry into the treatment of Aboriginal women, the Canadian government has yet to move in this direction despite additional investigations documenting the gravity of the situation. For example, a 2013 investigation by the Inter-American Commission on Human Rights concluded a national action plan was required to address the roots of the problem (IACHR, 2014).

The continued absence of systematic Canadian court data allowing researchers to link case characteristics to punishments has largely prevented a systematic examination into the criminal justice response to violence against Aboriginal women and violent crime more generally (Doob, 2011). This dearth of data is common internationally (UNODC, 2011). Using the Integrated Criminal Court Survey, Statistics Canada regularly publishes court statistics providing aggregate information on completed cases, including type and length of sentence and information on age and sex of accused. These figures show that males are sentenced to custody more frequently than females in Canada except in cases of homicide – where more females are sentenced to custody than males (Boyce, 2013: 15). However, no other characteristics are available to examine why this might be the case, such as the gender of the victim and the multitude of other factors that may be associated with dispositions.

In Canada, Section 222(4) of the Criminal Code of Canada includes three types of culpable homicide: murder (first- and second-degree), manslaughter and infanticide. First-degree murder involves one or more of the following: (1) planning/deliberation; (2) death of police officer(s), custodian(s), or prison personnel while on duty; or (3) occurring during commission of certain criminal acts (e.g. hijacking, kidnapping, forcible confinement, criminal harassment, or sexual assault) (Manson, 2001). If one of these criteria is not met, the homicide may be classified as second-degree murder. The line separating first- and second-degree murder is often obscure, but usually at issue is how much planning/deliberation was undertaken by the accused prior to the killing (Grant et al., 1998). Section 235 stipulates that those convicted of first- or second-degree will be sentenced to life in prison. First-degree murder carries a mandatory minimum 25-year parole ineligibility period; this ranges from 10 to 25 years for second-degree murder.

Manslaughter is culpable homicide that is not murder because ‘the person who committed it did so in the heat of passion caused by sudden provocation’ (s. 232(1)). Manslaughter is subject to a maximum of life in prison, but has no mandatory minimum. The mental element has been the crucial factor in distinguishing between murder and manslaughter, so decisions about degree of accused culpability or blameworthiness have not only been affected by the presence of provocation, but also by other factors (e.g. intoxication) (Grant et al., 1998). Finally, punishable by a maximum of five years, infanticide is the killing of a child less than one year, by wilful act or omission, by the mother who has not fully recovered from the effects of childbirth (i.e. mental disturbances, postpartum disorders).

Why might victim gender matter?

Two dominant theoretical paradigms draw attention to how gendered stereotypes may lead one to expect that gender of the victim will play a role in determining punishments (Richards et al., 2014). First, focal concerns theory argues that court actors may view offenders who kill women as more culpable or blameworthy and as more of a threat to society than those who kill men (Steffensmeier et al., 1998). However, assessments of culpability may be gendered given still common and dominant perceptions of gender roles as well as the varying types and levels of female and male crime. Specifically, women are more often stereotyped as vulnerable and in greater need of protection than men (Baumer et al., 2000). They are also perceived to be less violent in nature and less likely to engage in criminal behaviour more generally. As such, compared to male victims, women may be seen as less responsible for their own victimization when it does occur (Baumer et al., 2000).

Second, the chivalry/paternalism perspective provides two explanations for the role of gender stereotypes in determining differential treatment of crimes with female victims. On the one hand, some argue that the criminal justice system seeks to protect more vulnerable and weaker women by punishing their victimizers more harshly than victimizers of males (Baumer et al., 2000; Curry, 2010; Curry et al., 2004). On the other hand, some feminist writers argue that the criminal justice system remains patriarchal in nature; thus, it is men’s ownership of women – and their right to control them – that creates this protectiveness or desire to be chivalrous toward women victims (or offenders). ‘Chivalry is associated with placing an individual on a pedestal and behaving gallantly toward that person, whereas paternalism involves taking care of the powerless and dependent’ (Belknap, 2001: 133). In contrast, some gender conflict theorists argue male offenders who victimize women will be treated more leniently than those who victimize men because of the subordinate value placed on women’s lives (Belknap, 2001; Daly and Tonry, 1997). Women have less power than men and their gendered roles continue to be valued less; thus, they argue that crimes against women will be punished less severely than crimes against men (Franklin and Fearn, 2008).

At first glance, it seems difficult to reconcile these arguments; however, radical feminists and proponents of the intersectional framework underscore how not all women – victims or offenders – will meet the dominant stereotypes of ‘worthiness’ or ‘weakness’ (Crenshaw, 1991; Franklin, 2008; Miller, 2005). For example, despite the law’s claimed impartiality, Rapaport (1994) demonstrated how patriarchal values continue to operate by excusing male violence against female partners by emphasizing the role played by some victims in their demise. ‘Good’ victims are those who tried to reduce their risk (and that of their children) by leaving an abusive partner, but female victims may also be seen as ‘bad’ if they left for no obvious reason (at least in the court’s eyes), provoking their male partner’s loss of control (Mahoney, 1991–1992). Race/ethnicity of the victim (and their offender) may also impact the court’s response to violence (Crenshaw, 1991). Finally, Black’s (1976) theory of the behaviour of law draws attention to the importance of victim–offender dyads in determining punishments beyond the gender combination of those involved to consider, for example, the relationships they shared as a key factor in determining punishment. For example, crimes of femicide are often perpetrated by male partners and these acts are viewed primarily as private, expressive forms of violence (‘crimes of passion’) and may be treated more leniently as a result (Dawson, 2006; Rapaport, 1994). In contrast, male homicide victims are more often killed by male acquaintances and strangers, violence typically seen as more public, instrumental and serious.

Does gender of the victim matter?

Historically, and today, when examining gender and punishment, the emphasis has been on the gender of the accused rather than the victim. However, an evolving body of research has identified what is referred to as the ‘female victim effect’ demonstrating that, even when controlling for other case factors, crimes involving female victims appear to be punished more harshly than crimes with male victims. While much of this work has examined capital cases in the United States (Phillips et al., 2012; Royer et al., 2014), the majority of this research and other studies have demonstrated that the ‘female victim effect’ occurs at many stages of the criminal justice process and, in particular, for homicide (Baumer et al., 2000; Beaulieu and Messner, 1999; Curry, 2010; Curry et al., 2004; Farrell and Swigert, 1986; Glaeser and Sacerdote, 2000). For example, when victims are female, defendants are more likely to be prosecuted (Baumer et al., 2000), convicted on the most serious charges (Baumer et al., 2000), prosecuted as a capital rather than non-capital case (Baldus et al., 1990; Songer and Unah, 2006), to result in death penalty outcomes (Holcomb et al., 2004; Pierce and Radelet, 2002; Richards et al., 2014; Stauffer et al., 2006; Williams and Holcomb, 2004; Williams et al., 2007) or, in non-capital homicide cases, to receive longer sentences (Curry, 2010; Curry et al., 2004; Glaeser and Sacerdote, 2000), but less likely to receive charge reductions (Beaulieu and Messner, 1999; Glaeser and Sacerdote, 2000).

Consistent with theoretical expectations (Black, 1976; Crenshaw, 1991), some research has shown that the association between victim gender and court outcomes may be conditioned by other case characteristics, however. For example, several studies have shown that cases involving white female homicide victims are the most likely to result in death penalty outcomes (Hindson et al., 2006; Holcomb et al., 2004; Williams et al., 2007) or longer sentences in non-capital homicide cases (Curry, 2010) compared to other gender–race combinations. In addition, the behaviour of the victim, victim–defendant relationship, or the presence of sexual violence have been shown to condition the relationship between victim gender and punishment outcomes (Gillespie et al., 2013). However, the main effect of gender of the victim on its own remains a significant predictor of punishment in the majority of studies.

Gaps in prior research

Despite research supporting the ‘female victim effect’ in punishment outcomes, the primary emphasis on capital cases in the United States means that findings to date are mostly specific to that country and to a subset of killings. This leaves open the question of how homicides more generally may be treated elsewhere, underscoring the recent call for more international research (Ulmer, 2012). The dearth of existing court data internationally has precluded such an examination in most countries, particularly in non-western contexts. Therefore, despite its geographical proximity to the United States, understanding responses to femicide in Canada by drawing from a unique dataset can begin to address this international gap in knowledge. Further, while it is often assumed that Canada parallels the United States in its response to crime, there is little empirical basis for such a contention (Doob and Webster, 2006).² Given the focus on violence against women by male partners in recent decades, an understanding of how various subtypes of femicide are treated by the courts is also important as the ‘female victim effect’ may not be present for all women. This is particularly important given the emphasis in US-based research on victim gender and death penalty cases because intimate femicide cases are rarely prosecuted as such (Rapaport, 1994). Furthermore, enhanced efforts to address intimate femicide may have precluded attention to or an examination of punishment for other forms of femicide. Finally, violence against women more generally, and related issues about gender equality, have been the focus of international legislative and policy attention. This is true for Canada as well, resulting in significant changes in society’s response to violence against women. As such, a systematic examination of whether there have been corresponding changes in how the courts respond to femicide over time is long overdue. To begin to address these knowledge gaps, three hypotheses are examined in this study:

H1: Defendants in female victim cases will be subject to more severe sanctions than defendants in male victim cases.

H2: Punishment will vary across femicide subtypes with femicides involving victims and defendants in closer relationships subject to less serious sanctions than victims and defendants in more distant relationships.

H3: Defendants in cases of femicide in recent years will be subject to more severe sanctions than defendants in femicide cases disposed in earlier years.

The current study

Data

Using data on the total population of femicides in Ontario,³ this study describes the court response to femicide over four decades. The data were drawn from a larger project documenting all Ontario homicides, including both female and male victims (Dawson, 2012), for which data collection occurred in three stages and is ongoing. While each stage expanded information collected, details about victim, accused and case characteristics remain constant. Data sources include coroner’s records, police and newspaper reports, prosecutor and court files. Two different subsamples were examined. First, cases involving both female and male victims processed by the courts from 1985 to 2013 (N = 3425)⁴ were used to determine if there is evidence of a ‘female victim effect’ in a Canadian context (H1). The second sample captures femicides from 1974 to 2013 in which a male accused was identified, charges were laid and the relationship was known (N = 1381) to examine whether there were variations in punishment across femicide subtypes (H2) and over time (H3).

Dependent variables

Four dependent variables capture key court outcomes: initial charge, likelihood of conviction, conviction severity and sentence. First, initial charge is a key determinant of later court outcomes. Given that almost 90% of charges were for first- or second-degree murder, this variable compares first-degree murder charges (coded 1) to other charges (e.g. second-degree murder/manslaughter; coded 0). The second dependent variable distinguishes between those convicted (coded 1) and acquitted (coded 0; including ‘not criminally responsible on account of mental disorder’). Third, murder is distinguished from manslaughter by the existence of a specific intent (or mens rea; Criminal Code of Canada s. 229; Grant et al., 1998). Accused convicted of murder (coded 1; first- or second-degree) are distinguished from those convicted of less serious charges (coded 0; primarily manslaughter). Finally, sentence length is captured by the number of years to be served before parole eligibility. While there are statutory minimums for murder, there are none for manslaughter. Sentences can range from as low as 0 (e.g. suspended sentences) to 25 years.⁵

Independent variables

The effect of three independent variables is examined, paralleling hypotheses: victim gender (H1), victim–defendant relationship (H2) and time period (H3). Victim gender, as noted, is used to examine the ‘female victim effect’. Next, narrowing the focus to subtypes of femicide, victim–defendant relationship is measured using a four-category variable comparing intimate femicide, familial femicide, other-known-perpetrator femicide and stranger femicide (Russell and Harmes, 2001). Conceptualized as the closest victim–defendant relationship (Silverman and Kennedy, 1993), intimate femicide includes current/former legal spouses, common-law partners and girlfriends, typically the largest proportion of femicides in many countries (WHO, 2005). Familial femicide victims are killed by other family and include mothers, daughters and other female relatives (by blood or marriage). Other-known-perpetrator femicide includes primarily friends and acquaintances such as housemates, roommates, co-workers, neighbours and business relationships (legal and illegal). Finally, representing the most distant victim–defendant relationship, stranger femicide includes defendants who had no prior or minimal contact with the victim.

The final variable captures when cases were disposed, distinguishing three time periods: 1974–1984, 1985–1996 and 1997–2013. Pro-charging and pro-prosecution policies were introduced in Canada in 1983 and 1984. These new practices affected how charges were laid and processed, arguably serving as an impetus for change in attitudes toward intimate partner violence and violence against women more generally. In the early 1990s, following several highly publicized cases, criminal harassment provisions in the Criminal Code of Canada came into force criminalizing acts that cumulatively amounted to stalking. As such, the period 1985–1996 is expected to differ from the earlier time period, 1974–1984. Outcomes in cases disposed from 1997 to 2013 are expected to differ from the first two periods because Bill C-41 was passed in 1996 which included a statutory statement that offenders who abused spouses or children could now be subject to harsher penalties (Grant, 2010).

Available for analysis in this study and included as control variables to isolate the unique effects of focal variables were: number of defendants, number of victims, defendant and victim age, gun use, public homicide as well as charge and conviction severity, important predicators of later court outcomes. Defendant criminal record is a well-documented legal factor, but information is currently missing in over half the cases and, as such, could not be included in this analysis. This was also the case with race/ethnicity of the victim, which was included as a control in the latter analyses because there was a lower proportion of missing data in the femicide sample.

Analytic procedures

Logistic regression was used to examine the impact of each variable on the first three dependent variables. This is a common technique for dichotomous and skewed outcome variables, predicting odds an event will occur (i.e. murder versus manslaughter; Demaris, 1992). For sentence length, ordinary least squares regression was used, common for continuous dependent variables (Johnson et al., 2010). Similar to other studies using data drawn from official records, some variables were missing information for some cases.⁶ To retain these cases, for each independent variable with missing data, dummy variables were coded 0 if data were present and 1 if data were missing (Cohen and Cohen, 1975). This allows one to retain the maximum sample, increase statistical power and reduce bias in estimates (Orme and Reis, 1991).⁷ Finally, because earlier decisions will affect later outcomes, unmeasured variables that affect one stage may be correlated with unmeasured variables that affect a later outcome, producing a correlation between error terms on the two dependent variables. Thus, without including information on how cases are screened at each stage (i.e. only those convicted will receive a sentence), estimates of the effect of variables included in each equation may be biased (Berk, 1983). To correct for this in the model predicting sentence length, a two-equation estimation procedure can be used to control for incidental selection bias in the coefficients of the variables affecting various decision points (Klepper et al., 1983). Analyses were conducted both with and without this correction and the results did not differ, so findings are reported without the correction for ease of interpretation.⁸

Results

Is there a ‘female victim effect’ in Canada?

Descriptive and bivariate patterns

To examine the first research question, 3425 homicides were analysed, comprising 1200 female and 2225 male victims. Table 1 shows that female victim cases were significantly more likely to result in first-degree murder charges (52%), convictions (95%), including murder convictions (61%), and longer sentences (14 years) than cases involving male victims (38%, 87%, 41% and 11 years respectively). Underscoring the need to control for differences in female and male homicide, Table 1 also shows significant differences in case characteristics by victim gender. For example, compared to males, female victims were significantly more likely to be killed by intimate partners (50%) and other family members (19%) and less likely to be killed by friends/acquaintances (17%) and strangers (9%). However, some factors may increase the severity of punishment in female victim cases (e.g. more often multiple victims) compared to male victims, but other factors may reduce perceived case severity (e.g. fewer guns and public killings). To more closely examine the presence or absence of the ‘female victim effect’, these factors are controlled in the subsequent analysis.⁹

Table 1.

Descriptive and bivariate patterns for court outcomes in homicide by female and male victims, Ontario, 1985–2013.^a

	Female victims		Male victims		Total
	%	N	%	N	%	N
	35	1200	65	2225	100	3425
First-degree charge	52***	629	38	850	43	1479
Conviction	95**	966	87	1449	90	2415
Murder conviction	61***	584	41	598	49	1182
Sentence (years)	14***		11		12
Victim–accused relationship
Intimate partner	50***	603	7	155	22	758
Other family	19***	231	12	261	14	492
Friends/acquaintances	17***	205	49	1082	38	1287
Strangers	9***	103	20	440	16	543
Missing relationship	5***	58	13	287	10	345
Time period
1985–1996	48***	580	40	895	43	1475
1997–2013	52***	620	60	1330	57	1950
Female defendant	29***	93	71	232	10	325
Multiple accused	8***	96	29	636	21	725
Multiple victims	13***	157	9	202	10	355
Accused age (years)	35***		30		32
Victim age (years)	35		34		34
Gun used	13***	161	24	532	20	693
Missing gun use	7***	83	15	339	12	422
Public killing	14***	163	35	768	27	931
Missing public killing	14***	172	15	326	19	637

The sample at sentencing (N = 2127) is less than at initial charging (N = 3425) because cases drop out at particular stages.

Chi-square significance *p < .05; **p < .01; ***p < .001.

Multivariate analysis

Results in Table 2 for Models 1 through 4 demonstrate that cases involving female victims attract more serious sanctions at all stages examined, even when controlling for available factors. Consistent with the bivariate patterns, this supports the first hypothesis: compared to cases with male victims, defendants who killed female victims were more likely to be charged with first-degree murder, convicted overall, convicted of murder and to receive longer sentences. As expected, other factors also played a role in determining outcomes, as shown in Table 2, including those that differentiate between male and female victim cases. However, gender of the victim remained a consistently strong predictor of punishment, providing preliminary support for the ‘female victim effect’ in the Canadian context, at least in this jurisdiction. Whether the treatment of female victims depends on the relationship they shared with their killers is examined next.

Table 2.

Multivariate analyses of gender of victim on court outcomes for female and male victims, Ontario, 1985–2013.

Variables	Model 1 (Logit)		Model 2 (Logit)		Model 3 (Logit)		Model 4 (OLS)
	1st degree (N = 3425)		Convicted (N = 3425)		Murder (N = 2415)		Sentence (N = 2127)
	b	Odds	b	Odds	B	Odds	b
Female victim	.894 (.098)***	2.445	.749 (.199)***	2.114	.689 (.124)***	1.992	3.741 (.383)***
Intimate partner	.215 (.146)	1.240	.441 (.273)	1.554	.200 (.186)	1.221	−.639 (.562)
Other family	−.060 (.153)	.942	.776 (.285)**	2.174	−.746 (.191)***	.474	−1.495 (.611)*
Other known	.210 (.114)	1.233	.310 (.186)	1.363	−.067 (.142)	.935	−.325 (.439)
Missing relationship	−.039 (.155)	.962	−.548 (.261)	.578	.142 (.245)	1.152	−1.085 (.763)
(Reference: stranger)
Time period: 1985–1996	.123 (.076)	1.131	−.153 (.141)	.858	−.136 (.098)	.873	−1.367 (.303)***
(Reference: 1997–2013)
First-degree charge	–	–	.874 (.162)***	2.397	1.718 (.098)***	5.576	–
Murder conviction	–	–	–	–	–	–	−.006 (.009)
Female defendant	−.415 (.140)**	.661	−.501 (.215)*	.606	−1.252 (.196)***	.286	−4.295 (.561)***
Multiple accused	.908 (.097)***	2.479	−.158 (.169)	.854	.112 (.129)	1.119	2.205 (.392)***
Multiple victims	.866 (.126)***	2.377	.928 (.325)**	2.529	.818 (.163)***	2.266	4.711 (.474)***
Accused age	−.003 (.004)	.997	−.014 (.008)	.986	−.010 (.005)*	.990	.018 (.016)
Victim age	.007 (.002)**	1.007	.024 (.005)***	1.024	−.005 (.003)	.995	.007 (.010)
Gun use	1.059 (.098)***	2.884	−.388 (.177)*	.679	.401 (.129)**	1.494	2.342 (.393)***
Missing gun use	.283 (.157)	1.328	−.065 (.295)	.937	.302 (.221)	1.352	2.206 (.693)**
Public killing	−.027 (.095)	.973	−.109 (.169)	.896	−.347 (.123)**	.707	−.634 (.381)
Missing public killing	.006 (.135)	1.006	.036 (.247)	.965	−.122 (.183)	.885	.012 (.576)
Constant	−1.419 (.177)	.242	1.376 (.311)	3.960	−.493 (.223)	.611	9.786 (.689)

Notes: Standard errors in parentheses. Significance *p < .05; **p < .01; ***p < .001.

Are all femicides treated alike?

Descriptive patterns for the 1381 cases shown in Table 3 demonstrate that, consistent with other countries, the majority of femicide cases were intimate femicides (58%) followed by other-known-perpetrator femicides (18%), familial femicides (15%) and stranger femicides (10%). First-degree murder charges were laid more often in cases of other-known-perpetrator femicide (55%) and stranger femicide (65%) compared to intimate (48%) and familial femicides (43%). Patterns are somewhat consistent with expectations, although intimate femicides, conceptualized to represent the closest victim–defendant relationship, were expected to have the lowest proportion of first-degree murder charges. The proportion of cases resulting in convictions was similar across femicide subtypes with the exception of stranger femicide. With respect to conviction severity, familial femicide was found to have the lowest proportion of murder convictions (44%) followed by intimate femicide (57%). Finally, looking at sentence length, similar patterns were documented with defendants who killed women with whom they had no prior relationship – stranger femicide – receiving the longest sentences (17.5 years). The multivariate analysis includes controls for other factors that may differentiate these killings, similar to the model above, to further isolate the role that femicide subtype may play in determining court outcomes.

Table 3.

Descriptive patterns for court outcomes by femicide subtype, Ontario, 1974–2013 (N = 1381).

Variables	Intimate partner		Other family		Other known		Stranger		Total
	(58%; 801)		(15%; 206)		(18%; 242)		(10%; 132)		(N = 1381)
	%	N	%	N	%	N	%	N	%	N
First-degree charge	48	382	43	89	55	134	65	86	50	691
Likelihood of conviction	96	681	94	174	93	199	86	113	94	1167
Murder conviction	57	387	44	76	66	132	66	74	57	669
Sentence	14		13		15		17.5		13.5

Multivariate analysis

Results in Table 4 show that, overall, closer relationships did lead to less serious punishments, but not always the closest relationships and with one exception. First, intimate and familial femicide cases were significantly less likely to result in first-degree murder charges (Model 1) and longer sentences (Model 4) compared to stranger femicide. Second, familial femicides were less likely to result in murder convictions (Model 3) compared to stranger femicide; however, this was not true for intimate femicide, conceptualized to be the closest relationship. Finally, contradicting this hypothesis, Model 2 shows that intimate femicides were three times more likely to result in convictions than stranger femicide. This finding may stem from stereotypes about intimate femicide as violence that is more spontaneous, impacting the likelihood of charge reductions and guilty pleas for such defendants. These issues will be returned to in the discussion.

Table 4.

Significant predictors of court outcomes by key independent variables for femicide victims in Ontario, 1985–2013 (N = 1381).

Variables	Model 1 (Logit)		Model 2 (Logit)		Model 3 (Logit)		Model 4 (OLS)
	1st degree (N = 1381)		Convicted (N = 1381)		Murder (N = 1146)		Sentence (N = 963)
	b	Odds	b	Odds	B	Odds	b
Intimate partner	−463 (.216)*	.629	1.216 (.416)**	3.372	.002 (.267)	1.002	−2.733 (.652)***
Other family	−.837 (.249)**	.433	.929 (.509)	2.532	−.840 (.304)**	.431	−3.159 (.797)***
Other known	−.310 (.234)	.734	.710 (.432)	2.034	.180 (.290)	1.197	−2.421 (.708)**
(Reference: stranger)
Time period: 1974–1984	−.737(.163)***	.478	−1.263 (.392)**	.283	−.921 (.202)***	.398	−2.854 (.538)***
Time period: 1985–1996	.054 (.140)	1.056	−.704 (.383)	.495	−.167 (.174)	.847	−1.830 (.437)***
(Reference: 1997–2013)
First-degree charge	–	–	.594 (.289)*	1.812	1.806 (.140)***	6.087	–
Murder conviction	–	–	–	–	–	–	9.026 (.386)***
Multiple accused	.490 (.252)*	1.632	−.826 (.432)	.438	.232 (.321)	1.261	1.577 (.727)*
Multiple victims	.760 (.176)***	2.138	.256 (.419)	1.292	.549 (.223)*	1.732	3.095 (.558)***
Accused age	−.005 (.005)	.995	−.025 (.012)*	.976	−.001 (.007)	.999	−.001 (.017)
Victim age	−.005 (.004)	.995	.001 (.008)	1.001	−.006 (.004)	.994	−.025 (.012)*
Victim Aborig./minority	−.064 (1.69)	.938	.064 (.477)	1.066	−.079 (.212)	.924	−.485 (.541)
Missing Aborig./minority	−.275 (.147)	.938	−.440 (.350)	.644	.038 (.181)	1.039	−.918 (.467)
Gun use	.366 (.156)*	1.442	.052 (.375)	1.053	.053 (.191)	1.055	.168 (.479)
Missing gun use	.452 (.291)	1.572	−1.108 (.651)	.330	.061 (.426)	1.063	.731 (1.056)
Public killing	.376 (.172)*	1.456	−.113 (.379)	.893	.129 (.214)	1.138	1.042 (.528)*
Missing public killing	−.034 (.228)	.967	.614 (.645)	1.848	.315 (.302)	1.371	1.128 (.747)
Constant	.814 (.296)	2.257	3.543 (.678)	34.56	−.009 (.373)	.991	11.947 (.960)***

Notes: Standard errors in parentheses. Significance *p < .05; **p < .01; ***p < .001.

Punishments for femicide over time

This study spans a 40-year period during which there were significant transformations in social and legal environments responding to violence against women (Johnson and Dawson, 2011; WHO, 2005). As such, one might expect that practices on the ground have changed as well and, given the emphasis on criminal justice as a key component of overall solutions, an examination of whether law has changed in practice is crucial. Also shown in Table 4, supporting the final hypothesis, results demonstrated that there have been changes over time in punishments for femicide at all stages examined. Specifically, first-degree murder charges, convictions and murder convictions were significantly less likely to occur in femicide cases in the early period (1974–1985) compared to the most recent time period (1997–2013). Femicide cases disposed in both the early and middle time periods resulted in significantly shorter sentences than cases disposed in the more recent time period. It could be argued that what appears to be increasingly punitive sanctions for femicide may simply be a product of an increasing punitiveness overall. Focusing on the middle and recent time periods,¹⁰ a similar analysis was conducted determining sentence length for male victims (not shown here), which showed there was also a significant increase in sentence severity between the middle and recent time periods for male victims (1.196; p < .001) although the relationship was not as strong as for females (2.854 and 1.830; p < .000).

Discussion and conclusion

This study represents the first systematic examination of punishment outcomes for the total population of femicides in one Canadian province over several decades. Overall, preliminary support was documented for the three hypotheses. First, a ‘female victim effect’ is evident in the Canadian context: cases involving female victims were treated more punitively than those with male victims at all four stages examined. Second, the way in which femicide victims were treated by the courts depends upon their relationships with the killers. Defendants who killed women with whom they shared closer (but not always the closest) relationships were subject to less punishment than those who shared more distant relationships. Finally, there have been changes over time in how the courts respond to femicide, supporting a tentative conclusion that legislative and policy changes may be contributing to improved social and legal responses to these crimes in Canada. Each of these findings is discussed in more detail below, including limitations of the current study and implications for future research priorities.

Unpacking the ‘female victim effect’

In North America, at least, it appears that femicide is generally treated as a serious crime, at least in comparison to male victims, if one accepts at face value the ‘female victim effect’ demonstrated in this study and US-based research (e.g. Royer et al., 2014). However, while this and other studies controlled for factors that may differentiate killings of women and men, before concluding the ‘female victim effect’, additional research is needed to systematically examine whether femicide cases are typically more serious in nature and circumstances than male homicides. For example, femicide often involves sexual violence, which in some countries can aggravate the homicide (Baldus et al., 1990), and research has shown that sexual violence conditions the association between victim gender and punishment (Gillespie et al., 2013; Williams et al., 2007). That is, when the sexualized nature of homicide is considered, the gender of the victim may no longer be a significant determinant of punishment as shown for capital cases. As such, cases involving female victims may not be treated more punitively than those with male victims because of a ‘female victim effect’; rather it could be the nature of the killing that increases the severity of punishment.

It was not possible to examine the effect of sexual violence on punishment outcomes in this study; however, in Canada, murder is first-degree if the killing occurred during the commission of various sexual assault offences (see s. 231(5)(b)(c)(d) of Criminal Code), underscoring the presence of statutory aggravation. Given that sexual violence could also be one obvious gender-based indicator of femicide, collecting more comprehensive data on its presence in homicide as well as excessive violence – often referred to as ‘overkill’ and common in cases of femicide – is a priority for future research. As part of this move, more nuanced indicators of femicide could also be identified, moving beyond the basic definition of ‘women killed by men’ to highlight the very different contexts in which this might occur, although similar in its social structural origins and the role of gender inequality. Understanding how sexual violence interacts with gender of the victim across different femicide subtypes would also be crucial given that feminists have argued that the legacy of patriarchal legal doctrines continues to see women treated as property of men in some relationships (Rapaport, 1994), an issue discussed next.

A discount for intimacy?

The results of this study demonstrate that not all femicide victims are treated alike; in particular, cases of familial and intimate femicide were subject to less punishment than other femicides. In other words, men who killed their wives, lovers, mothers, sisters, daughters, aunts and so on were subject to what some have referred to as a domestic discount (Rapaport, 1994), and conceptualized here as an intimacy discount because it also occurs beyond the domestic setting. The one exception to the intimacy discount was that intimate femicide had a higher likelihood of conviction; however, it may be that gender stereotypes which are often intricately linked with stereotypes about intimacy and violence can shed some light on this finding.

For instance, a common stereotype related to intimate femicide is that these crimes are more often spontaneous, ‘crimes of passion’ rather than premeditated events. Although this has been challenged (Dawson, 2006), many historical and traditional stereotypes are so entrenched in the public mind, including court actors, they are hard to eradicate even if there is no empirical basis for them. Given the perceived spontaneity in intimate femicide, research has shown that these cases may more often be resolved through guilty pleas thereby increasing the likelihood of conviction. Relatedly, research has also shown that defendants in cases of intimate femicide have a higher solvability or clearance rate because defendants often remain at the scene or confess to the killing, making such cases inherently easier to prosecute. While intimate femicide has a greater likelihood of conviction, these cases subsequently received shorter sentences, so understanding whether the plea process or other factors (e.g. perceived victim provocation) explain this relationship is an important next step. Hot-blooded crimes or ‘crimes of passion’ continue to be seen to be as synonymous with intimate femicide yet there have been few systematic examinations of whether this is actually the case (but see Dawson, 2006). In addition, an examination of whether and how stereotypes about gender, intimacy and violence may make such pleas more amenable to criminal justice actors is also crucial.

The presence of an intimacy discount also shows that the intense focus of legislative and policy initiatives on intimate femicide in recent decades, at least in North America, has not precluded attention to the severity of other types of femicide and, more specifically, those that occur in the public sphere. Overall, in Canada at least, this research has shown that stranger femicide is treated as a more serious crime, particularly at the sentencing stage, compared to intimate and familial femicide. This may mean that women killed by male partners are still seen as property and, as such, these femicides are not treated as seriously as other femicides – the continuing reign of patriarchal legal doctrines. It may also be that sexual violence, if it aggravates the punishment as hypothesized above, is more often present in stranger femicides, or, alternatively, when sexual violence occurs in intimate femicide, it is not deemed to be as serious in nature; but this remains speculation until future research examines these issues.

Changing responses to femicide over time

Recent discussions surrounding femicide have identified the need for more attention to its prevention and punishment worldwide (Laurent et al., 2013). Despite what is argued to be a continuing legacy of patriarchal legal doctrines that lead to the more lenient treatment of intimate and familial femicide, this study documented what might be considered an improvement in the court’s response to femicide in recent decades. This may be explained, in part, to increasing punitiveness in Canadian sentencing overall (Doob and Webster, 2006) but, arguably, some of this improvement may also be due to increased legislative and policy attention to femicide and violence against women more generally. This study poses the latter as a tentative conclusion, arguing that the emphasis on legislative and policy mechanisms to respond to violence against women in various world regions, and to femicide in particular, may have a positive effect over time.

Recent contributions of the intersectionality approach (Crenshaw, 1991), however, underscore the need to understand who has been affected by these changes and in what way. In Canada, the disproportionately higher risks faced by Aboriginal women highlight the need to prioritize responses to killings of this group of women, in particular. Collecting information on race/ethnicity from court records is a challenge and missing information is a common problem, not unique to this study. The situation has been exacerbated in Canada by a long-standing reluctance, and sometimes informal ban, on the collection and dissemination of such data (primarily for researchers). One exception has been official government statistics that document the overrepresentation of Aboriginals in the criminal justice system as victims and as accused. However, even these examinations are limited, data are not readily accessible to researchers and reliability of race/ethnicity data is low. Regardless, understanding how the law’s symbolic – and real – response might be contributing to risk for these and other groups of vulnerable women is paramount.

Aggregate data such as those examined in this study, while sometimes missing more nuanced information, can begin to identify initial patterns that warrant further examination. Some initial avenues for future research have been discussed above which also have the potential to contribute to identifying gender-based indicators of femicide that can be more closely examined in future studies. Examining what sanctions are imposed for these crimes may help us understand whether what are often viewed as symbolic state gestures, although nonetheless important, actually impact practices on the ground.

Footnotes

Acknowledgements

Thank you to Danielle Sutton, Julie Poon and Guila Benchimol for collecting the data used in the third stage of this study. Thank you to the anonymous reviewers for their constructive comments on a previous version of this article.

Funding

I wish also to thank the Social Sciences and Humanities Research Council for the funding which supported this research as well as the Canada Research Chair programme.

Notes

Author biography

Myrna Dawson is a Canada Research Chair in Public Policy in Criminal Justice, Associate Professor, Department of Sociology and Anthropology, and Director, Centre for the Study of Social and Legal Responses to Violence (), University of Guelph, Ontario, Canada. Her most recent project examines intimacy, violence and the law in the international context.

References

Alvazzi del Frate

Nowak

(2013) Femicide in global perspective. In: ACUNS, Femicide: A Global Issue that Demands Action. Vienna: Academic Council on the United Nations System (ACUNS).

Baldus

Woodworth

Pulaski

(1990) Equal Justice and the Death Penalty: A Legal and Empirical Analysis. Boston, MA: Northeastern University Press.

Baumer

Messner

Felson

(2000) The role of victim characteristics in the disposition of murder cases. Justice Quarterly 17(2): 281–307.

Beaulieu

Messner

(1999) Race, gender, and outcomes in first degree murder cases. Journal of Poverty 3(1): 47–68.

Belknap

(2001) The Invisible Woman: Gender, Crime and Justice, 3rd edn. Belmont, CA: Wadsworth/Thompson.

Berk

(1983) An introduction to sample selection bias in sociological data. American Sociological Review 48: 386–398.

Black

(1976) The Behavior of Law. New York: Academic Press.

Bond

Jeffries

(2014) Similar punishment? Comparing sentencing outcomes in domestic and non-domestic violence cases. British Journal of Criminology 54(5): 849–872.

Boyce

(2013) Adult criminal court statistics in Canada, 2011/2012. Juristat Catalogue No. 85-002-X. Ottawa: Statistics Canada.

10.

Campbell

Runyan

(1998) Femicide. Guest editors’ introduction. Homicide Studies 2(4): 347–352.

11.

Cohen

(1975) Applied Multiple Regression/Correlation Analysis for the Behavioral Sciences. Hillsdale, NJ: Lawrence Erlbaum.

12.

Cotter

(2014) Homicide in Canada, 2013. Juristat Catalogue No. 85-002-X. Ottawa: Statistics Canada.

13.

Crenshaw

(1991) Mapping the margins: Intersectionality, identity politics, and violence against women of color. Stanford Law Review 43(6): 1241–1299.

14.

Curry

(2010) The conditional effects of victim and offender ethnicity and victim gender for sentences for non-capital cases. Punishment and Society 12(4): 438–462.

15.

Curry

Lee

Fernando Rodriguez

(2004) Does victim gender increase sentence severity? Further explorations of gender dynamics and sentencing outcomes. Crime and Delinquency 50(3): 319–343.

16.

Daly

Tonry

(1997) Gender, race and sentencing. Crime and Justice 22: 201–252.

17.

Dawson

(2006) Intimacy, violence and the law: Exploring stereotypes about victim–defendant relationship and violent crime. Journal of Criminal Law and Criminology 96(4): 1417–1450.

18.

Dawson

(2012) Intimacy, homicide, and punishment: Examining court outcomes over three decades. Australian and New Zealand Journal of Criminology 45(3): 400–422.

19.

Dawson

Gartner

(1998) Differences in the characteristics of intimate femicide: The role of relationship state and relationship status. Homicide Studies 2(4): 378–399.

20.

Demaris

(1992) Logit Modeling: Practical Applications. London: Sage.

21.

Doob

(2011) The unfinished work of the Canadian Sentencing Commission. Canadian Journal of Criminology and Criminal Justice 53(3): 279–297.

22.

Doob

Webster

(2006) Countering punitiveness: Understanding stability in Canada’s imprisonment rate. Law and Society Review 40(2): 325–367.

23.

Farrell

Swigert

(1986) Adjudication in homicide: An interpretative analysis of the effects of defendant and victim social characteristics. Journal of Research in Crime and Delinquency 23(4): 349–369.

24.

Fox

(1997) Applied Regression Analysis, Linear Models and Related Methods. Thousand Oaks, CA: Sage.

25.

Franklin

(2008) Women offenders, disparate treatment, and criminal justice: A theoretical, historical, and contemporary overview. Criminal Justice Studies 21(4): 341–360.

26.

Franklin

Fearn

(2008) Gender, race and formal court decision-making outcomes: Chivalry/paternalism, conflict theory or gender conflict? Journal of Criminal Justice 36(3): 279–290.

27.

Gartner

Dawson

Crawford

(1999) Woman killing: Intimate femicide in Ontario, 1974–1994. Resources for Feminist Research 26(3/4): 151–173.

28.

Gillespie

Loughran

Smith

. (2013) Exploring the role of victim sex, victim conduct, and victim–defendant relationship in capital punishment sentencing. Homicide Studies 18(2): 175–195.

29.

Glaeser

Sacerdote

(2000) The Determinants of Punishment: Deterrence, Incapacitation and Vengeance (Discussion Paper 1894). Harvard Institute of Economic Research. Cambridge, MA: National Bureau of Economic Research.

30.

Grant

(2010) Intimate femicide: A study of sentencing trends for men who kill their intimate partners. Alberta Law Review 47: 779–822.

31.

Grant

Chunn

Boyle

(1998) The Law of Homicide. Scarborough, Ont.: Carswell.

32.

Hindson

Potter

Radelet

(2006) Race, gender, region and death sentencing in Colorado, 1980–1999. University of Colorado Law Review 77: 549–594.

33.

Holcomb

Williams

Demuth

(2004) White female victims and death penalty disparity research. Justice Quarterly 21(4): 877–902.

34.

Hotton Mahony

(2010) Women and the Criminal Justice System. Ottawa: Statistics Canada.

35.

Hotton Mahony

Turner

(2012) Police-reported clearance rates in Canada, 2010. Juristat Catalogue No. 85-002-XPE. Ottawa: Statistics Canada.

36.

IACHR (Inter-American Commission on Human Rights) (2014) Missing and Murdered Indigenous Women in British Columbia, Canada. Washington, DC: IACHR.

37.

Johnson

Dawson

(2011) Violence Against Women in Canada: Research and Policy Perspectives. Toronto: Oxford University Press.

38.

Johnson

Van Windgerden

Nieuwbeerta

(2010) Sentencing homicide offenders in the Netherlands: Offender, victim and situational influences in criminal punishment. Criminology 48(4): 981–1018.

39.

Klepper

Nagin

Tierney

(1983) Discrimination in the criminal justice system: A critical appraisal of the literature. In: Blumstein

Cohen

Martin

Tonry

(eds) Research on Sentencing: The Search for Reform, Vol. II. Washington, DC: National Academy Press, pp. 55–128.

40.

Laurent

Platzer

Idomir

(2013) Femicide: A Global Issue that Demands Action. Vienna: Academic Council on the United Nations (ACUNS) Vienna Liaison Office.

41.

Mahoney

(1991–1992) Legal images of battered women: Redefining the issue of separation. Michigan Law Review 90(1): 1–94.

42.

Manson

(2001) The Law of Sentencing. Toronto: Irwin Law.

43.

Miller

(2005) Victims as Offenders: The Paradox of Women’s Violence in Relationships. New Brunswick, NJ: Rutgers University Press.

44.

Mujica

Tuesta

(2014) Femicide penal response in the Americas: Indicators and the misuses of crime statistics, evidence from Peru. International Journal of Criminology and Sociological Theory 7(1): 1–21.

45.

Nowak

(2012) Femicide: A Global Problem. Small Arms Survey. Vienna: Academic Council on the United Nations System (ACUNS).

46.

NWAC (Native Women’s Association of Canada) (2010) What Their Stories Tell Us: Research Findings from the Sisters in Spirit Initiative. Ottawa: NWAC.

47.

Orme

Reis

(1991) Multiple regression with missing data. Journal of Social Service Research 15(1–2): 61–91.

48.

Pearce

(2013) An awkward silence: Missing and murdered vulnerable women and the Canadian justice system. Dissertation, University of Ottawa.

49.

Phillips

Haas

Coverdill

(2012) Disentangling victim gender and capital punishment: The role of the media. Feminist Criminology 7(2): 130–145.

50.

Pierce

Radelet

(2002) Race, region, and death sentencing in Illinois, 1988–1997. Oregon Law Review 81: 39–96.

51.

Rapaport

(1994) The death penalty and the domestic discount. In: Fineman

Mykitiuk

(eds) The Public Nature of Private Violence: The Discovery of Domestic Abuse. New York: Routledge, pp. 224–254.

52.

RCMP (Royal Canadian Mounted Police) (2014) Missing and Murdered Aboriginal Women: A National Operational Overview. Ottawa: RCMP.

53.

Richards

Jennings

Smith

. (2014) ‘Female victim effect’ in capital punishment: An examination of victim sex-specific models of juror sentence decision-making. Crime and Delinquency (Online) 18(2): 1–24.

54.

Royer

Hritz

Hans

. (2014) Victim gender and the death penalty. Cornell Law School, Legal Studies Research Paper Series.

55.

Russell

DEH

Harmes

(2001) Femicide in Global Perspective. London: Teacher’s College Press.

56.

Silverman

Kennedy

(1993) Deadly Deeds: Murder in Canada. Scarborough: Nelson.

57.

Songer

Unah

(2006) The effect of race, gender, and location on prosecutorial decisions to seek the death penalty in South Carolina. South Carolina Law Review 58: 161–205.

58.

Statistics Canada (2006) Victimization and Offending among the Aboriginal Population in Canada. Ottawa: Minister of Industry.

59.

Statistics Canada (2014) Canada’s population estimates: Age and sex, 2014. The Daily, 26 September.

60.

Stauffer

Smith

Cochran

. (2006) The interaction between victim race and gender on sentencing outcomes in capital murder trial: A further exploration. Homicide Studies 10(2): 98–117.

61.

Steffensmeier

Ulmer

Kramer

(1998) The interaction of race, gender and age in criminal sentencing: The punishment cost of being young, black, and male. Criminology 36(1): 763–798.

62.

Ulmer

(2012) Recent developments and new directions in sentencing research. Justice Quarterly 29(1): 1–40.

63.

United Nations (2011) Report of the Special Rapporteur on Violence against Women, its Causes and Consequences, Rashida Manjoo. New York: United Nations.

64.

UNODC (2011) Homicide statistics. Available at: www.unodc.org/unodc/en/data-andanalysis/homicide.html.

65.

WHO (World Health Organization) (2005) WHO Multi-Country Study on Women’s Health and Domestic Violence Against Women: Initial Results on Prevalence, Health Outcomes and Women’s Responses. Geneva: WHO.

66.

Widyono

(2008) Conceptualizing femicide. In: Strengthening Understanding of Femicide: Using Research to Galvanize Action and Accountability. Seattle, WA: PATH, pp. 7–25.

67.

Williams

Holcomb

(2004) The interactive effects of victim race and gender on death sentence disparity findings. Homicide Studies 8(4): 350–376.

68.

Williams

Demuth

Holcomb

(2007) Understanding the influence of victim gender in death penalty cases: The importance of victim race, sex-related victimization, and jury decision-making. Criminology 45(4): 865–891.