Introduction
The question of whether defendants receive equal treatment before the law is complicated because equal treatment can have various definitions. At the most basic level, equal treatment means that defendants who are accused and convicted of statutorily similar offenses have engaged in similarly serious crimes. It has long been recognized, however, that similar acts or behaviors may be reacted to differently depending on the characteristics of the people involved and the circumstances surrounding the crime. One key dimension theorized to predict varying legal responses to similar acts is the degree of intimacy that exists or existed between victims and defendants (Reference BlackBlack 1976, Reference Black1993; Reference Gottfredson and GottfredsonGottfredson & Gottfredson 1988; Reference HorwitzHorwitz 1990). Typically, sociolegal and feminist theorists argue that law varies inversely with the degree of intimacy that exists between people (e.g., Reference BlackBlack 1976; Reference Rapaport, Fineman and MykitiukRapaport 1994). “Law is most likely to become involved, to proceed aggressively, and to be penal in style when the parties are strangers; it is least likely to become involved and most likely to be lenient and conciliatory when they are intimates” (Reference BlackBlack 1976:40).
A systematic examination of findings from relevant research, however, does not allow for any conclusive statement regarding the role of intimacy in law. Moreover, in the last several decades, the social construction of intimacy within the legal environment has undergone and continues to experience fundamental changes. For example, family violence—historically constructed as “private” acts for which legal intervention was not appropriate—has increasingly been labeled deviant or criminal. Therefore, the traditional assumption that intimacy results in criminal justice leniency may be too simplistic in contemporary societies. The central objectives of this study, then, are to examine the extent to which the degree of intimacy that exists between victims and defendants may affect criminal justice responses to violent crime and to determine if this association has changed over time.
Prior Research
Why Might Intimacy Matter?
Current and traditional perspectives in the sociology of law and deviance have drawn attention to the use of screening devices by social control agents as important instruments of social control that take into account factors that are not directly related to the deviant act itself. This work highlights how criminal court personnel may develop certain expectations about the nature of an offense based on characteristics of the individuals involved or concerns that arise over time in a community (Reference BeckerBecker 1963; Reference BlackBlack 1976; Reference EmersonEmerson 1983; Reference Erikson and BeckerErikson 1964; Reference HorwitzHorwitz 1990; Reference Kitsuse and CicourelKitsuse & Cicourel 1963; Reference Rubington and WeinbergRubington & Weinberg 1978; Reference SchurSchur 1971; Reference SudnowSudnow 1965; Reference Swigert and FarrellSwigert & Farrell 1977). These expectations can shape public attitudes and behavior toward deviants that, in turn, help identify those individuals who are to be defined as deviant or criminal (Reference Farrell and SwigertFarrell & Swigert 1986). In short, social control agents may rely on stereotypes or commonsense assumptions about crime and criminals that lead them to focus on some offenses and offenders more than others.
The degree of intimacy that exists between a defendant and a victim is one characteristic that has been shown to generate stereotypical images in cases of interpersonal violence, leading to more lenient punishments in some cases (Reference MietheMiethe 1987; Reference RapaportRapaport 1991, Reference Rapaport, Fineman and Mykitiuk1994; Reference WaegelWaegel 1981). For example, research has demonstrated that violence between intimates is often perceived to involve strong emotion such as desperation or rage because of the intensity of intimate relationships and the accompanying interactions (Reference LoftinLoftin 1986; Reference MaxfieldMaxfield 1989; Reference Messner and TardiffMessner & Tardiff 1985; Reference Parker and SmithParker & Smith 1979; Reference Rojek, Williams and WilsonRojek & Williams 1993; Reference SampsonSampson 1987; Reference Smith and ParkerSmith & Parker 1980). The presence of strong emotion can also act to decrease defendant culpability in law because it reduces the perceived presence of premeditation or intent. In contrast, violence between strangers is often presumed to occur in the context of violence committed for gain, thereby often lacking strong emotion (see Reference BlockBlock 1981; Reference RiedelRiedel 1987; Reference Rojek, Williams and WilsonRojek & Williams 1993). This increases defendant culpability and, consequently, the severity of punishment.
Similarly, some research has shown that, while images of provocation may occasionally be invoked in cases that involve strangers, crimes between intimates are more often perceived to involve some degree of victim responsibility or participation (Reference RapaportRapaport 1991; Reference RiedelRiedel 1987). Within the criminal process, the legal notion of provocation generally mitigates the culpability of the defendant, leading to more lenient punishments (see Reference MietheMiethe 1987; Reference Williams and McDonaldWilliams 1976; Reference WolfgangWolfgang 1957). Based on these and other stereotypes, particularly about intimate violence, it is commonly assumed that the degree of intimacy shared by victims and defendants will affect criminal justice outcomes.Footnote 1
Does Intimacy Matter?
Research on determinants of criminal justice outcomes has seldom examined the role of victim-defendant relationship—the best available measure of intimacy—as a key variable of interest.Footnote 2 Instead, the majority of studies have included it as a control variable in analyses that examine the effect of social status variables such as race, gender, and social class on criminal justice decisionmaking (see exceptions in Reference MietheMiethe 1987; Reference SimonSimon 1996a, Reference Simon1996b). This research has shown that the association between victim-defendant relationship and law varies by the type of analysis conducted, the stage of the criminal process examined, or the type of crime being sanctioned. For instance, studies using bivariate analyses have consistently found that violence between intimates is treated more leniently by criminal justice officials than violence that occurs between strangers (Reference Ferraro, Boychuk, Buzawa and BuzawaFerraro & Boychuk 1992; Reference HickmanHickman 1995; Reference LundsgaardeLundsgaarde 1977; Reference PalmerPalmer 1999; Reference Rapaport, Fineman and MykitiukRapaport 1994; Vera Institute of Justice 1977).
In contrast, the effect of victim-defendant relationship on court outcomes is less clear in more rigorous multivariate analyses that enable researchers to control for the effects of other legal (e.g., prior criminal record, offense seriousness) and extralegal (e.g., race, age) factors. For example, several studies have shown that defendants who victimize intimates are treated more leniently at some stages of the criminal justice process, but not at all stages (Reference AdamsAdams 1983; Reference Erez and TontodonatoErez & Tontodonato 1990; Reference Horney and SpohnHorney & Spohn 1996; Reference MietheMiethe 1987; Reference SimonSimon 1996a; Reference Spohn and SpearsSpohn & Spears 1997; Reference Williams and McDonaldWilliams 1976). Other studies found no association between victim-defendant relationship and the allocation of criminal sanctions (Reference AlbonettiAlbonetti 1991; Reference SimonSimon 1996b; Reference MyersMyers 1979a, Reference Myers1979b). The role of relationship type in determining legal outcomes has also been found to vary for personal and property crimes generally or, within the category of violent crime, by the type of violence examined (Reference Steffensmeier, Ulmer and KramerSteffensmeier, Ulmer, & Kramer 1998; Reference Williams and McDonaldWilliams 1976).
There are various reasons why findings about the relationship between intimacy and law appear to be contradictory. First, victim-defendant relationship is traditionally conceptualized as either intimate or nonintimate, with no allowance for within-group variations.Footnote 3 Within these broad categories, however, are distinct relationship types that may differ in ways that lead to varying criminal justice responses. For example, of all social relationships, none is more intense than that of intimate partners. From a sociological point of view, the relevant characteristics of this social arrangement that produce such intensity are sexual intimacy and physical proximity (Silverman & Reference Silverman and KennedyKennedy 1993). In fact, violent acts that occur between intimate partners are often seen as synonymous with “crimes of passion.” However, these unions are often grouped with arguably “less intense” intimate relationships such as friends and other family members, precluding an examination of their potential variation in legal treatment. Consequently, when criminal justice outcomes for the broader categories of “intimate” and “nonintimate” are compared, the results may depend, in part, on the proportion of intimate partners, family, and friends that constitute the larger group of intimates.
Second, researchers have begun to recognize that, due to the sequential nature of the criminal justice process, analyses need to consider the impact of earlier decisions on later outcomes (Reference Bernstein, Kelly and DoyleBernstein, Kelly, & Doyle 1977; Reference Bernstein, Kick, Leung and SchultzBernstein et al. 1977; Reference Bernstein, Cardascia, Ross, Alvarez and LuttermanBernstein, Cardascia, & Ross 1979; Reference HaganHagan 1974; Reference Miethe and MooreMiethe & Moore 1986; Reference PetersiliaPetersilia 1983; Reference SchurSchur 1971; Reference Swigert and FarrellSwigert & Farrell 1977; Reference Thomson and ZingraffThomson & Zingraff 1981). At the same time, outcomes at earlier stages of the legal process should be examined as important decisionmaking points in and of themselves, not only because of their influence on later outcomes, but also because the sequential nature of the process means that different criminal justice actors make decisions at different stages. For instance, the specific concerns of a prosecutor at the charging stage may be substantively different from judicial concerns at sentencing. However, studies have rarely tracked cases through the system, allowing for the possibility that criminal justice leniency may occur only at specific stages and considering each decision point as a potential predictor of later outcomes.
In addition to the above conceptual and methodological issues, it may also be that the association between intimacy and law is not invariant over time as has been traditionally assumed (see Reference BlackBlack 1976, Reference Black1993). Instead, the effect of intimacy on law may vary temporally, but no study to date has examined its effects over a significant amount of time. In fact, most research has examined study periods less than three years in duration and primarily from the mid-1970s, a time when the role of intimacy in law began to be vigorously challenged by feminist researchers. As a result, during the past three decades, major legislative and policy initiatives have significantly changed the way in which criminal justice actors respond to intimate violence. As already noted, though, few studies have focused exclusively on victim-defendant relationship as the key variable of interest. I argue that this represents a significant gap in criminal justice research given that the relationship between a defendant and his/her victim has traditionally been perceived as one of the most critical variables in research on violent crime and criminal justice (Reference DeckerDecker 1993; Reference BlackBlack 1976, Reference Black1993; Reference Gottfredson and GottfredsonGottfredson & Gottfredson 1988; Reference HorwitzHorwitz 1990).
The Present Study
Using data on the total population of homicides in one urban jurisdiction over a 23-year period, I examine the legal treatment of defendants who vary in the type of relationship they shared with their victims prior to the killing.Footnote 4 Two research questions are central to my investigation. First, to what extent does the relationship between a defendant and a victim affect criminal justice responses to homicide? Second, does the effect of intimacy on criminal justice outcomes in homicide cases vary over time? With respect to the first question, drawing from the above research, I hypothesize that:
H1: Cases involving defendants and victims who share a close relationship will be treated more leniently by the courts than cases that involve defendants and victims who share more distant relationships.
However, I argue that if criminal justice actors do use screening devices to respond to cases, they are likely to do so as soon as cases enter the criminal justice system. As a result, due to the sequential nature of criminal justice decisionmaking, the role played by intimacy may have the greatest impact at earlier stages of the process. In other words, if decision makers at later stages (i.e., sentencing) recognize that criminal justice leniency has occurred earlier in the process (i.e., charging), additional leniency may not be perceived as appropriate. Therefore, I hypothesize that:
H2: Cases involving defendants and victims who share a close relationship will be treated more leniently than those who share more distant relationships at earlier stages of the criminal justice process rather than at the later stages.
Finally, as discussed above, there is reason to expect that changes have occurred over time in the construction of intimacy in law. Intimates are increasingly taking their disputes before the courts, and recognition has grown that intimate violence, in particular, should attract the same degree of concern among criminal justice decision makers as violent crime between nonintimates (Reference Fineman, Fineman and MykitiukFineman 1994). During the past several decades, then, concern about intimate violence has taken “hold in the shifting moods of the community” (Reference Erikson and BeckerErikson 1964:11). Social and legal changes that have arisen in response to this growing concern may have lead to a reduction in criminal justice leniency toward these types of crimes over time. Therefore, I hypothesize that:
H3: Any effect of intimacy on criminal justice decisionmaking should attenuate over time. In particular, any evidence of the more lenient treatment of intimate violence compared to other types of violence should be less evident in more recent years.
In testing the above hypotheses, my analysis builds on previous research in this area in several important ways. First, my work moves beyond the traditional conceptualization of broader relationship categories common in criminal justice research, using more refined distinctions among relationship types (discussed in more detail below). Second, in response to research that has emphasized the sequential nature of the criminal process, I track more than 1,000 homicide cases through the criminal justice system, considering each stage as an important decision point as well as examining the effects of early decisions on later outcomes.
Finally, because intimate relationships and attitudes toward intimacy have changed fundamentally in recent decades, I assess the effect of intimacy on law over almost a quarter of a century. Beginning in the early 1970s, the period I examine parallels the beginning of a social movement that led to increasing public awareness of intimate and family violence as a serious social problem. Below, I describe the data, measures, and analytic techniques adopted to examine these hypotheses.
Data
The data for this study are drawn from the total population of homicides from 1974 to 1996 that were resolved through the courts in Toronto, Ontario.Footnote 5 Data collection began with a review of summary police reports on homicides in Toronto. These reports provide an outline of the circumstances immediately surrounding the cases and demographic and other personal information on the defendants and the victims. To address gaps in the data or to confirm information, I examined a variety of other official and unofficial sources.Footnote 6 Because my study examines court outcomes, I also collected more detailed information on the circumstances surrounding the homicides and the criminal justice outcomes for each case from prosecutors' files and from reports published in various daily newspapers.
From 1974 through 1996, a total of 1,533 homicides were recorded in the city of Toronto. During this 23-year period, a total of 1,003 homicides were dealt with through the adult criminal court system.Footnote 7 Using these data, I identify patterns in criminal justice responses to homicide for various victim-defendant relationship categories. Below, I discuss the measurement of key independent, control, and criminal justice outcome variables as well as the analytic techniques used in the analysis (see Table A1 for coding and descriptive statistics for all variables).
Independent Variables
Victim-Defendant Relationship
Drawing from work by Reference BlackBlack (1976, Reference Black1993) and others (Reference DeckerDecker 1993; Reference HorwitzHorwitz 1990; Reference Silverman and KennedySilverman & Kennedy 1987, 1993), I construct a five-category measure of victim-defendant relationship to use as a proxy for intimacy.Footnote 8 The five relationship types that I examine here are intimate partners, family members (not including spouses), friends, acquaintances, and strangers. Underlying my measure of intimacy is the assumption that intimate partners have more frequent interactions, more common social characteristics, and so on than do family members and others (see Figure 1).
Figure 1. Continuum of victim-defendant relationships: Arrows indicate that increasing intimacy is hypothesized to correspond to increasing leniency for defendants
The intimate partner category includes current and former legal spouses, common-law partners, and dating couples.Footnote 9 Family members include those defendants who killed parents, children, and other kin, such as grandparents. Stepchildren, stepparents, common-law kin, and foster children are also included in this category. Friends include housemates, roommates, drinking buddies, and other types of relationships in which the defendant and victim were determined to have shared a relatively close or amicable relationship prior to the killing. The acquaintance category includes coworkers, neighbors, business relationships (both legal and illegal), and adversaries. Finally, the stranger category includes defendants who had no prior contact or minimal contact (i.e., met on the day of the incident) with their victims.Footnote 10 I entered each relationship type into the model as a dummy variable (0,1) and used the category “stranger” as the reference group.
Time Period
The second key independent variable is a dichotomy that captures the time period in which the case entered the court system. This dichotomy distinguishes between those cases that entered the courts in 1984 or earlier (coded 0) and those that entered the courts in 1985 or later (coded 1). I split the study period in this way because the majority of legislative and policy changes occurred in Canada in the early 1980s or later.Footnote 11 For example, mandatory charging and prosecution (“no-drop”) policies were introduced in Canada in the early 1980s, beginning with federal guidelines issued to the Royal Canadian Mounted Police as well as federal and territorial Crown prosecution offices in 1983. By 1985, then, some form of spousal assault policy was in place in most of the provinces in Canada (Reference BrownBrown 2000).Footnote 12 These new practices had a significant effect on the number of common assault charges laid in incidents of intimate partner violence and how they were processed through the criminal justice system. In addition, the new legislation served as an impetus for change in both public and professional attitudes toward intimate violence. Furthermore, in the early 1990s, following several highly publicized homicides of women by their estranged male partners, criminal harassment provisions in the Canadian Criminal Code came into force. The rationale behind this new legislation was to protect victims of criminal harassment by criminalizing otherwise lawful acts that cumulatively amounted to stalking.Footnote 13 An examination of these two time periods, then, allows for a comparison of the criminal justice treatment of cases resolved in distinct legal and social environments.
Control Variables
I also include a number of control variables for victim, defendant, and case characteristics that prior research has found to be associated with criminal justice decisionmaking. I group these independent variables into legal factors and extralegal factors, including defendant and victim characteristics as well as situational characteristics of the homicide.Footnote 14
Legal Factors
I include seven legal variables in my analyses:Footnote 15 defendant's criminal history, defendant's role in the incident, number of defendants involved in the homicide, number of victims killed, severity of initial prosecution charge, mode of conviction, and severity of conviction. First, I capture the criminal history of the defendant with a dichotomous measure indicating whether the defendant had a prior criminal record (coded 1) or had no prior history of convictions (coded 0).Footnote 16 I expect that the existence of a prior criminal record will significantly affect the severity of the sentence as documented by prior research (Reference Blumstein, Cohen, Martin and TonryBlumstein et al. 1983; Reference Gottfredson and GottfredsonGottfredson & Gottfredson 1988; Reference Hagan, Bumiller, Blumstein, Cohen, Martin and TonryHagan & Bumiller 1983; Reference Klepper, Nagin, Tierney, Blumstein, Cohen, Martin and TonryKlepper, Nagin, & Tierney 1983) and may also have some effect on early outcomes such as initial charge and mode of conviction.
The second legal variable pertains to the defendant's role in the homicide. The defendant's role may mitigate the perceived blameworthiness of a defendant if it is determined that he or she was a follower rather than a leader or organizer (Reference Steffensmeier, Ulmer and KramerSteffensmeier, Ulmer, & Kramer 1998). For example, an individual who aids and abets a homicide is legally indistinguishable from the primary defendant. However, the degree of involvement in the offense may be relevant in determining the sentence (Reference Grant, Chunn and BoyleGrant, Chunn, & Boyle 1998). Thus, I include a dichotomous variable that captures whether the defendant was the primary offender (coded 1) or a secondary offender (coded 0).
The number of defendants involved in a homicide is the third legal variable, capturing whether there was a single defendant (coded 0) or multiple defendants (coded 1). Furthermore, the killing of more than one person at one time may also lead to more severe criminal justice responses (Reference Huang, Wilson, Finn, Barry and FriedmanHuang et al. 1996; Reference MyersMyers 1980). Because the majority of homicides involve one victim only, I include a fourth legal variable that designates whether the case involved one victim (coded 0) or more than one victim (coded 1). The final three legal variables—initial prosecution charge, mode of conviction, and conviction severity—are outcome variables that are used as controls in later decisionmaking stages. They are described in more detail below.
Extralegal Factors
I include gender, age, race, and employment status of both the defendant and the victim in all models. Research has shown that these are important indicators of social status in Western society and that they have implications for criminal justice decision-making (e.g., Reference Steffensmeier, Ulmer and KramerSteffensmeier, Ulmer, & Kramer 1998).Footnote 17 For example, while results are mixed, research has shown that court outcomes vary for males and females (see reviews in Reference Bickle and PetersonBickle & Peterson 1991; Reference Daly and BordtDaly & Bordt 1995; Reference OdubekunOdubekun 1992; Reference Steffensmeier, Kramer and StreifelSteffensmeier, Kramer, & Streifel 1993), young and old defendants (Reference Steffensmeier, Kramer and UlmerSteffensmeier, Kramer, & Ulmer 1995), whites and nonwhites (see reviews in Reference KleckKleck 1985; Reference Kramer and SteffensmeierKramer & Steffensmeier 1993), and employed and unemployed defendants (Reference BorisBoris 1979; Reference Reskin and VisherReskin & Visher 1986). I also include controls for gun use and location of killing. Gun use is often used as a measure of offense seriousness, predicting more severe court outcomes (Reference Hagan, Nagel and AlbonettiHagan, Nagel, & Albonetti 1980; Reference LaFreeLaFree 1980; Reference LizotteLizotte 1978), and the public nature of some crimes may be perceived as posing a greater threat to the maintenance of social order (Reference LundsgaardeLundsgaarde 1977).
Criminal Justice Outcomes
I use five dependent variables to capture the sequence of decisionmaking in the criminal process: initial prosecution charge, mode of conviction, verdict at trial, severity of conviction, and length of sentence.Footnote 18
Initial Prosecution Charge
Research has shown that this is an important decision point for defendants because the seriousness of the initial charge, under most circumstances, will be an important determinant of the conviction and sentence (Reference Brereton and CasperBrereton & Casper 1981–82). In my analysis, I distinguish between defendants charged with first-degree murder (coded 1) and those charged with a less serious crime (coded 0).Footnote 19 The primary justification for a first-degree murder charge, relative to a less serious charge, is evidence of premeditation—a factor that makes lethal acts more serious in the eyes of the law (Reference Grant, Chunn and BoyleGrant, Chunn, & Boyle 1998). For my sample as a whole, 37% of the defendants were initially charged with first-degree murder.
Mode of Conviction
Criminal defendants have a constitutionally guaranteed right to trial by judge or jury. However, the majority of criminal charges in Canada (and many other Westernized countries) are disposed of by guilty plea rather than at trial (e.g., Reference RubyRuby 1999). While guilty pleas may not be as frequent in homicide cases as they are for other types of violence, they still represent a significant proportion of cases that result in convictions. Two reasons for this may be the high risks perceived to be associated with trials by defense and prosecution or the expense involved (Reference MatherMather 1979). To capture the mode of conviction, I use a dichotomy that distinguishes between cases that were resolved at trial (coded 1) and those that were resolved through guilty pleas (coded 0). In the total sample, 58% of the cases were resolved at trial.Footnote 20
Verdict at Trial
Some research has shown that when cases of violent crime are resolved at trial, judges are less likely to convict if the defendant allegedly victimized a stranger, whereas juries are more likely to convict in cases of stranger violence (Reference MyersMyers 1981). Beyond this, research has rarely examined how victim-defendant relationship may affect a defendant's likelihood of conviction at trial. In my analysis, I coded cases that resulted in acquittal as 0 and cases in which defendants were found guilty at trial as 1. In the Toronto data, of those cases resolved at trial, 60% of the defendants were found guilty.
Severity of Conviction
The distinction between a murder and a manslaughter conviction is of great significance for a defendant in Canada because murder carries a mandatory penalty of life imprisonment, whereas manslaughter convictions carry no minimum mandatory sentence (Reference Grant, Chunn and BoyleGrant, Chunn, & Boyle 1998). Generally, murder is distinguished from manslaughter by the existence of malice aforethought on the part of the defendant (i.e., intent to commit a crime). Thus, the presence of certain extraneous or mitigating factors such as provocation, intoxication, or diminished responsibility may reduce an offense from murder to manslaughter (Reference MitchellMitchell 1991). In my sample as a whole, almost 40% of the defendants were convicted of either first- or second-degree murder (coded as 1) and the remainder were convicted of less serious charges (coded as 0).Footnote 21
Length of Sentence
The sentencing of a defendant involves two separate decisions: whether to imprison the defendant and, if imprisonment is imposed, what length of sentence is appropriate for the crime. Due to the seriousness of the offense being examined, only 20 cases in the Toronto data set did not result in a term of imprisonment; thus, I examine only the second decision: the length of sentence imposed.Footnote 22 The sentencing of convicted offenders is based, in part, on materials presented to the court after guilt has been determined, such as pre-sentence reports that describe, among other things, the criminal history of a defendant. However, sentencing decisions are also dependent upon the type of conviction and, within each offense category, the range of possible sentences. For example, defendants convicted of first-degree murder in Canada receive an automatic life sentence with no chance of parole for 25 years. In contrast, while a conviction of second-degree murder also carries a mandatory life sentence, the period of parole ineligibility may range from 10 to 25 years, thus allowing for greater judicial discretion. An interval-level variable captures the length of sentence imposed. The average sentence length in my sample of cases was about nine years.
Analytic Procedures
Because variables at different levels of measurement impose particular restrictions on the types of analyses that can be conducted, I employ two types of regression procedures. I use logistic regression to estimate those models with a dichotomous dependent variable and ordinary least-squares to estimate the model predicting length of sentence. In addition, because earlier decisions are expected to affect later outcomes, unmeasured variables that affect one stage of the process may be correlated with unmeasured variables that affect a later outcome, producing a correlation between error terms on the two dependent variables. As a result, without including information on how cases are screened into the sentencing stage (i.e., the conviction process), estimates of the effect of variables included in the sentencing equation may be biased (Reference BerkBerk 1983; Reference Berk and RayBerk & Ray 1982; Reference Klepper, Nagin, Tierney, Blumstein, Cohen, Martin and TonryKlepper, Nagin, & Tierney 1983). To correct for this, I follow a two-equation estimation procedure to control for incidental selection bias in the coefficients of the variables affecting various decision points (Reference BerkBerk 1983; Reference Berk and RayBerk & Ray 1982; Reference HeckmanHeckman 1976, Reference Heckman1979; Reference Klepper, Nagin, Tierney, Blumstein, Cohen, Martin and TonryKlepper, Nagin, & Tierney 1983).
Findings
Bivariate Patterns
Table 1 shows the results from the bivariate analysis that distinguishes among the five relationship types to compare their treatment across the various criminal justice stages. The table also shows differences in the characteristics of those involved and the circumstances surrounding the crime by relationship type. There are significant differences in the criminal justice treatment of defendants based on the type of relationship they shared with their victims.Footnote 23 Specifically, family members are treated more leniently at all stages of the process except for mode of conviction (i.e., cases sent to trial); friends receive lenient treatment at the initial charging stage and at sentencing; acquaintances are treated more severely at the charging stage; and, finally, strangers receive more severe treatment at the latter three stages. There are no significant differences in the treatment of intimate partners compared to other relationships at the bivariate level.
Table 1. Bivariate Patterns for Five-Category Victim-Defendant Relationship Variable by Criminal Justice Outcomes, Time, Legal, and Extralegal Variables in Homicide Cases in Toronto, Ontario, 1974–1996a
a All numbers are percentages except for mean age for victims and defendants and for sentence length.
b The category “family member” does not include spouses; this relationship is included in the “intimate partner” category.
Note: Asterisks indicate significant variation between that relationship type and a comparison group that includes the four remaining relationships;
* p<0.05;
** p<0.01;
*** p<0.001.
Table 1 also reveals that there are distinct differences in the characteristics of the cases, depending on the type of victim-defendant relationship. With respect to situational factors, for example, my results show that killings between strangers and acquaintances are significantly more likely to occur in a public setting than homicides involving intimate partners, family, or friends. As noted above, public violence may lead to more serious criminal justice responses because these acts are perceived to threaten public order (Reference LundsgaardeLundsgaarde 1977). Therefore, it may be the public nature of these events that lead to more severe sanctions in these cases rather than the type of relationship that existed between the victim and the defendant. Based on these and other differences shown in Table 1, then, the more important question for this analysis is whether defendants who share various types of relationships with their victims are treated differently once controls are introduced for factors that distinguish among these crimes and may also affect criminal justice decisionmaking.
Multivariate Analysis
Table 2 presents unstandardized coefficients for a series of logit models in which the four dichotomous criminal justice outcomes are regressed on each of the victim-defendant relationships, the time period in which the case entered the courts, and the control variables.Footnote 24 Ordinary least-squares regression coefficients for length of sentence imposed are shown in the final column. As noted above, I use the “stranger” category as the reference category. I discuss findings for each stage of the process below.
Table 2. Multivariate Analysis Incorporating the Five-Category Victim-Defendant Relationship Variable by the Five Criminal Justice Outcomes, Toronto, Ontario, 1974–1996
Note: Standard errors in parentheses;
* p<0.05;
** p<0.01;
*** p<0.001.
First-Degree Murder Charge
The results of Model 1 show the regression coefficients for those variables that had net effects on the seriousness of the initial prosecution charge. Confirming Hypothesis 1, my results demonstrate that defendants closest to their victims—intimate partners—were significantly less likely to be charged with first-degree murder than other types of defendants. Contrary to my expectations, however, the defendants most likely to be charged with first-degree murder were not those who killed strangers, but those who killed acquaintances.Footnote 25 Defendants who killed family members and friends were treated no differently at this stage than those who killed strangers, despite the tendency in research to conceptualize these relationships as similar to that of intimate partners.
With respect to time, the second key independent variable, I found that defendants whose cases entered the courts during the more recent period—1985 to 1996—were more likely to be charged with first-degree murder than defendants whose cases were heard during the earlier period. Other factors that increased the likelihood that a defendant would be charged with first-degree murder included the death of more than one victim, the involvement of more than one defendant, a female victim, and the use of a gun. In contrast, if the victim was unemployed, a first-degree murder charge was less likely, consistent with previous research (Reference BorisBoris 1979).
Case Sent to Trial
With respect to Model 2, mode of conviction, I found that defendants who killed intimate partners were significantly less likely to have their cases resolved at trial than defendants who killed strangers. In other words, those defendants who killed intimate partners were more likely to enter guilty pleas for their crimes. Again, family members and friends, traditionally viewed as intimates, were treated no differently than strangers at this stage in the process.
Looking at the effect of time, defendants whose cases entered the courts in the later period were less likely to have their cases resolved at trial. The only legal factor that was significantly associated with mode of conviction was the number of victims killed in the incident: Defendants who killed more than one victim were more likely to have their cases sent to trial than those who killed a single victim. In contrast, if defendants were unemployed, their cases were less likely to go to trial, more often being resolved through guilty pleas. Two victim characteristics were also relevant at this stage: Cases involving female victims were more likely to go to trial than those with male victims, and cases that involved unemployed victims were less likely to be resolved at trial than those with employed victims. Finally, defendants who used a gun in the killing were more likely to have their cases sent to trial than those who used other methods of lethal violence, such as strangulation, stabbing, or beating.
Found Guilty
There were no differences across victim-defendant relationship categories in the likelihood that a defendant was found guilty at trial (Model 3).Footnote 26 However, time period was significant: If a case was tried in 1985 or later, a defendant was more likely to be found guilty at trial than those defendants whose cases were heard in 1984 or earlier. Various legal and extralegal factors also increased the likelihood that a defendant would be found guilty at this stage. A guilty verdict was also more likely if the defendant had a prior criminal record, if he or she was the primary offender, or if multiple defendants were involved. The likelihood that a defendant was found guilty at trial also increased as the severity of the initial charge increased. Only one defendant characteristic was significantly related to the trial verdict: Defendants who were unemployed were less likely to be found guilty at trial than those who were employed. Similarly, if the victim was unemployed, the likelihood of a guilty verdict decreased (see also Reference BorisBoris 1979; Reference Williams and McDonaldWilliams 1976).
Murder Conviction
Similar to the previous stage, there were no differences across relationship types in the likelihood that a defendant would be convicted of murder (Model 4).Footnote 27 With respect to the second focal variable, time period was associated with more punitive responses to violent crime: Cases that were resolved in 1985 or later were more likely to result in a murder conviction than earlier cases. Five of the six legal variables were also associated with conviction severity. Defendants who had prior criminal records or who were primary offenders were more likely to be convicted of murder. Those cases that involved the death of more than one victim also increased the likelihood of a murder conviction and, as expected, the severity of the initial charge increased the likelihood of a murder conviction. The only legal factor that significantly decreased the likelihood of a defendant being convicted of murder was a guilty plea. If the defendant pled guilty to his or her crime, a murder conviction was less likely. With respect to extralegal factors, defendants were less likely to be convicted of murder if they were nonwhite or if their victims were unemployed. In contrast, the likelihood of a murder conviction increased if the defendant was unemployed or if the victim was female.
Length of Sentence
Finally, contrary to my second hypothesis, even when earlier decisions were controlled for, there were still significant differences across victim-defendant relationships in the length of sentence that defendants received. Defendants who shared the closest relationships with their victims—intimate partners and family members—received significantly lighter sentences than other defendants (see Model 5).Footnote 28 Specifically, defendants who killed intimate partners received sentences that were about one year shorter than those imposed on defendants who killed strangers, and those who killed family members received sentences that were close to two-and-a-half years shorter than for those who killed strangers. Killers of friends and acquaintances were treated no differently than those who killed strangers at this last stage in the process.
The importance of time was again evident at this stage of the process. Sentencing decisions made in 1985 or later resulted in slightly longer sentences than those made in the earlier period. With respect to legal factors, cases that involved multiple victims and multiple defendants also increased the length of sentence and, as expected, more serious convictions resulted in longer sentences. In contrast, defendants who pled guilty were sentenced to shorter periods of imprisonment. Surprisingly, none of the defendant characteristics were significantly related to the sentencing decision. However, with respect to victim characteristics, if the case involved a female victim, the defendant received a longer sentence, and if the victim was unemployed, the length of sentence imposed upon the defendant decreased. These latter findings underscore the importance of examining both defendant and victim characteristics in criminal justice research.
In summary, with respect to my first research question, my results demonstrate that the degree of intimacy between a defendant and a victim does affect criminal justice responses to interpersonal violence. However, the association between intimacy and law is contingent upon the stage of the criminal process and the type of relationship being examined. However, when leniency is evident, my analyses showed that it is those defendants who share the closest relationship to their victims—intimate partners—who are most likely to be the recipients of such treatment.
I suggest above, however, that the contradictory findings about the role of intimacy in law evident in prior research may also stem from the fact that this association is not invariant over time. Below, I examine the independent effects of time on criminal justice decisionmaking, comparing cases resolved during the earlier period to cases that entered the system in more recent years—1985 or later.Footnote 29
The Independent Effects of Time
Intimate violence has been the subject of various legislative and policy changes in the past several decades. I examine the effect of time to determine if the lenient treatment of intimate violence as demonstrated above has abated somewhat over time, paralleling this increased attention. Overall, as demonstrated in the above analyses, time was significantly and positively associated with four of the five outcome variables examined. This means that, in general, as time has passed, criminal justice responses to violence appear to have become more punitive (see Table 2). To examine whether the role of intimacy in criminal justice decisionmaking has also varied over time, I compare cases that entered the criminal justice system in 1984 or earlier to cases that were resolved through the courts in 1985 or later for each relationship type.
Examining the earlier period, Table 3 demonstrates that defendants who killed intimate partners were less likely to be charged with first-degree murder and less likely to have their cases resolved at trial than those who killed strangers. Moreover, those who killed intimate partners and family members were significantly less likely to be convicted of murder than defendants who shared more distant relationships with their victims. Finally, defendants who killed intimate partners, family members and friends received shorter sentences in the earlier years, compared to those who killed strangers. In contrast, an examination of the later period reveals no evidence of leniency at any stage of the criminal justice process for any victim-offender relationship type. In fact, the only significant association during this period shows that acquaintance killers were more likely to be charged with first-degree murder than those who killed strangers. These results, then, support the third hypothesis of my analysis: The effect of intimacy on criminal justice decisionmaking has attenuated somewhat over time. That is, any evidence of criminal justice leniency in cases of intimate violence compared to other types of violent crime is less evident in the later period—1985 or later. I turn to a discussion of this and others issues below.
Table 3. An Examination of Changes Over Time in the Criminal Justice Response to Violent Crime for Each of the Criminal Justice Outcomes by Victim-Defendant Relationship, Toronto, Ontario, 1974–1996
a Stranger homicide is the reference category for both time periods.
Note: All control variables are included in the above models. Standard errors are in parentheses.
* p<0.05;
** p<0.01;
*** p<0.001.
Discussion and Conclusion
This article addresses two specific research questions. First, do criminal justice responses to violent crime vary, depending on the degree of intimacy that exists between victims and defendants? Second, does the association between intimacy and law vary over time? My results demonstrate that criminal justice responses to violent crime do vary depending on the degree of intimacy between victims and defendants as captured by the best available proxy: victim-defendant relationship. However, I show that the association between intimacy and law is more complex than traditionally perceived because it depends on the type of relationship as well as the stage of the criminal process being examined. My results also show that criminal justice leniency toward intimate violence was more evident in the earlier period than in the later period, suggesting that there may be some reason for optimism about the potential for changing criminal justice attitudes toward intimate violence. I expand on these issues in more detail below.
Variation Within Intimate and Nonintimate Categories
Summarizing prior literature on intimacy and law, I suggest that one reason research findings have remained equivocal might stem from the way in which intimacy has been conceptualized. That is, the majority of research has relied upon a dichotomous measure that distinguishes between intimate and nonintimate relationships. Underlying this approach is the assumption that all intimate violence will be treated similarly and all nonintimate violence will be treated the same by criminal justice actors. However, I argue that, within these broad categories, there are distinct victim-defendant relationships that may also lead to variation in criminal justice responses to interpersonal violence. Replacing the traditional dichotomy with my five-category relationship measure, I show that intimate violence is not always treated the same by criminal justice officials and, further, that there is variation in how nonintimate violence is sanctioned.
For example, previous research has included “friends” in the intimate category (e.g., Reference AlbonettiAlbonetti 1991; Reference Erez and TontodonatoErez & Tontodonato 1990; Reference MyersMyers 1980; Reference SimonSimon 1996a, Reference Simon1996b). However, my analysis demonstrates that overall, in the case of homicide at least, defendants who killed friends were not treated significantly different from those who kill acquaintances or strangers. I also found that cases involving victims and defendants perceived to be closest in relational distance—intimate partners and family members—were not always subject to the same treatment. When intimacy was associated with less law, those defendants who killed intimate partners were most often subject to criminal justice leniency. But what is it about intimate partner relationships that leads to leniency in cases of violence? Why are crimes between intimate partners treated differently from crimes between family members or friends? One possible answer is that intimate partners share a type of sexual intimacy, over and above the physical proximity that is common to all intimate relationships. As a result, it may be that this sexual component leads criminal justice actors, and the general public, to assume that there is an increased intensity in such relationships that will frequently lead to strong emotions and loss of control during their interactions. As noted above, loss of control or passionate rage often acts to reduce defendant culpability in law. However, the existence of this and other stereotypes and their association with particular relationship types has yet to be systematically examined. I will return to this issue below when discussing the effect of intimacy on law over time.
In short, given the importance placed on victim-defendant relationship in research on violent crime, my findings suggest that researchers should consider how more distinct types of intimate and nonintimate relationships are treated. Before we can begin to understand why differential legal treatment might occur for victims and defendants who share different types of relationships, we need to know who is most often subject to criminal justice leniency and, as discussed next, where in the process such leniency is most likely to occur.
Where Intimacy Matters Most
This article highlights a second reason why research findings might be inconsistent with respect to the role of intimacy in law: an inability to capture the sequence of decisionmaking that characterizes the criminal justice process. By examining five separate stages of the process as important decisionmaking points and considering the effect of earlier decisions on later outcomes, I found that three criminal justice stages appear to matter most: the charging stage, mode of conviction, and sentencing. Because previous research is predominantly characterized by studies that focus on one or two outcomes, it is difficult to compare my findings to previous research. My results suggest, though, that criminal justice leniency is more evident in these early stages and again at sentencing. These stages involve different decision makers—prosecutors and judges—and different substantive concerns. Thus, why intimacy matters more at these decision points than at intervening stages is a pertinent question and should be the focus of future research.Footnote 30
The Effect of Intimacy on Law Over Time
Finally, and perhaps most important, my results suggest that those skeptical of the impact of legislative and policy changes on criminal justice attitudes toward intimate violence may be too pessimistic. My findings demonstrate that criminal justice leniency toward intimate violence is less evident in more recent years than in the earlier period of my study. As a result, it may be that amendments to or the implementation of various programs and policies during the past few decades have served as an impetus for change in the attitudes of criminal justice officials and members of the public, generally, toward violence within intimate relationships. Indeed, it may be that an increasing awareness about the issues and complexities that surround intimate partner violence, in particular, has lead to a concern, as Erikson argues, that has taken hold “in the shifting moods of the community” (1964:11).
While my analysis was not able to determine whether there is a direct relationship between the increasing recognition of the seriousness of intimate violence and the absence of criminal justice leniency toward this type of violence in the latter period of my study, it does suggest that such an association is possible and warrants further investigation. For example, as already noted, the introduction of mandatory charging and “no-drop” prosecution policies during the 1980s led to significant changes in the way criminal justice officials respond to violence between intimate partners. Designed to counter the notion that such incidents are private affairs and to recognize this type of violence as a serious societal problem, these initiatives are believed to have served as an impetus for change in both public and professional attitudes toward violence as well as to ongoing awareness-building among criminal justice actors and elected decision makers (Health Reference HealthCanada 1999). It is interesting, then, that my results showed that during the same time period—1985 and later—criminal justice leniency in cases of intimate violence appears to have abated, suggesting the possibility that there may be some connection between these parallel trends.
Surprisingly, there has been little research in Canada or elsewhere that has systematically examined whether criminal justice attitudes toward intimate violence have changed during this period and whether such changes are the result of the changing legal and social environments within which they respond to such crimes. One Ontario study found growing police support for the mandatory charging policy over the course of its implementation in one jurisdiction, demonstrating that 50% of the police officers felt that the policy was effective in helping battered women and reducing family violence, compared with only 33% in 1985 (London Family Court Clinic 1991). Moreover, two-thirds of the police officers agreed that the policy sent an important message to the community. With respect to prosecution, some researchers have demonstrated that the introduction of specialized courts and prosecution units that accompanied “no-drop” policies has led to “domestics” being redefined from “low-profile, messy cases with poor prospects for conviction” to “high-priority cases requiring skilled and sensitive lawyers” (Reference BrownBrown 2000:10).
While no Canadian research to date has evaluated judicial opinion on charging and prosecution policies or judicial attitudes toward intimate violence, it is reasonable to expect that judicial decisionmaking may also be responding to this changing social climate. Research has shown that legislators, policymakers, judges, and other criminal justice actors frequently cite public opinion when formulating and implementing policy and, in particular, criminal justice policy (Reference Roberts and HoughRoberts & Hough 2002), and that public today, in Canada at least, perceives family violence as a serious social problem that needs to be addressed. For example, a recent survey found that the majority of Canadians believe family violence has become a more serious problem during the past 10 years and thus perceive it as a high-priority issue for government and for their own community (EKOS 2002). It is important to remember as well that criminal justice professionals are also members of this same general public who indicate they believe intimate violence is a serious issue and a priority concern. Given that similar social changes with respect to the problem of intimate violence have occurred in the United States and other westernized countries, it is possible that similar criminal justice trends in responding to violence may be evident in other countries as well.
Accompanying the increased recognition of intimate violence as a societal issue that requires attention have been ongoing challenges to the traditional stereotypes associated with these types of crimes. As such, an examination of stereotypes about different victim-defendant relationship types—and, in particular, intimate partner relationships—may help explain why criminal justice leniency is afforded to some defendants and not others. Furthermore, such challenges to the validity of these stereotypes may help explain why criminal justice responses to intimate partner violence appear to have become more punitive over time. For example, as noted above, research has shown that violent crime between intimate partners has traditionally been treated as synonymous with “crimes of passion.” In other words, hot-blooded acts that do not involve premeditation or planning are acts that are perceived to occur between intimate partners (for a more detailed discussion, see Reference Rapaport, Fineman and MykitiukRapaport 1994). Recent work by feminists and other researchers, however, has shown that intimate partner violence more often involves control and domination rather than passion (e.g., Reference Dobash and DobashDobash & Dobash 1979, Reference Dobash and Dobash1984, Reference Dobash and Dobash1992; Reference Saunders, Yllo and BogradSaunders 1988; Reference Wilson, Johnson and DalyWilson, Johnson, & Daly 1995) and is often premeditated or, at the very least, intentional and purposeful (Reference DawsonDawson 2001; Reference PolkPolk 1994).
Similarly, in one of the few studies that examined victim-defendant relationship as a focal variable, Reference MietheMiethe (1987) found that offenders who were known to their victims were slightly less likely than strangers to receive a prison sentence. He argued that this could be explained, in part, because factors perceived to be indicative of “dangerousness” (e.g., gender, race, weapon use) were more important in cases involving strangers than those involving victims and offenders known to each other. Again, this finding is consistent with the stereotypical notion that incidents of intimate violence are isolated crimes of passion that have little probability of recurring and, as such, pose little future danger to members of the public or to the maintenance of social order in general (Reference LundsgaardeLundsgaarde 1977). As such, court actors may believe that they need to concentrate on violence by strangers because such offenders are perceived to pose the most danger to the public and are feared most by members of society (Reference Davis and SmithDavis & Smith 1981). However, the increasing recognition of intimate violence as a public rather than a private concern has resulted, in part, from the gradual process of redefining this type of violence as dangerous to both the public at large and to social order, not just to individual victims (Reference Pleck, Ohlin and TonryPleck 1989). Accompanying this redefinition may be more severe sanctions for these types of crimes.
Sociologists have long emphasized the importance of studying the values and attitudes of people responding to deviant or criminal behavior when attempting to explain criminal justice processing (Reference BeckerBecker 1963; Reference SchurSchur 1971). In general, though, little systematic research has examined the reasoning practices behind decisions made by legal actors (Reference DalyDaly 1994). And despite the recent increase in public and professional attention focused on intimate violence, even less research has examined contemporary attitudes and assumptions about intimacy inherent in the criminal justice process. Therefore, it is not yet obvious how information about a prior relationship between a victim and a defendant is processed and interpreted by criminal justice officials in different situations nor how dominant stereotypes and assumptions may affect their decisions. As part of this future examination, though, it will be important for researchers to conceptualize intimacy as existing on a continuum rather than as a monolithic category—intimacy is not necessarily something that either exists or does not exist; rather, it exists in varying degrees that can have implications for both societal and legal responses to violence.
The analysis that I have presented here does not resolve the debate about the role played by intimacy in law. My results are based on the processing of homicide cases in one Canadian urban jurisdiction. It may be that the processing of homicides will differ from that of other types of violent crime, and this may have implications for the generalizability of my findings. Moreover, because I examined criminal justice decisionmaking in only one jurisdiction, it is also possible that my findings may not hold true in other Canadian courts or in courts in other countries. However, given the parallel legal and social changes that have occurred in westernized countries to address the problem of intimate violence and violence against women, I argue that other jurisdictions and other countries may also be witnessing increasingly punitive criminal justice responses to various types of intimate violence similar to what I have shown here. However, support for my argument awaits further investigation on the role of intimacy in law over time.
Appendix
Table A1. Coding and Description of Variables, Toronto Homicide Data, 1974–1996 (N=1,003)
