Anda di halaman 1dari 19

Public Perceptions of the Seriousness of Victim-Precipitated Homicides: An Empirical Analysis of the Consensus Model of the Law on Homicide in Israel

Sergio Herzog Department of Sociology and Anthropology University of Haifa, Israel

July, 2005

Suggested Running Head: Seriousness of victim-precipitated homicides

Address for correspondance: Dr. Sergio Herzog Department of Sociology and Anthropology University of Haifa, Haifa 31905, Israel Tel: 972-4-824-0993 Fax: 972-4-824-0819 E-mail: sherzog@soc.haifa.ac.il Personal site: http://soc.haifa.ac.il/~sherzog

Public Perceptions of the Seriousness of Victim-Precipitated Homicides: An Empirical Analysis of the Consensus Model of the Law on Homicide in Israel

Abstract Criminal statutes on homicide widely support the consensus model of the criminal law. The present study attempted to ascertain whether the typical lack of recognition of victim-precipitated homicide as a separate, less serious category of homicide by criminal statutes reflects public views on this issue. To this end, a representative sample of the Israeli population was requested to evaluate the seriousness of various precipitated and non-precipitated homicide scenarios. The findings show that the role played by the victims in their own lethal victimization had a limited effect on respondents perceptions, thus providing additional support for the consensus model of the criminal law on homicide.

2 A pivotal question in the sociology of law is the extent to which the criminal law reflects the values prevailing in society. Briefly, confronted with the conflict model, the theoretical consensus model of the criminal law (henceforth, consensus model) postulates a close match between the attitudes of various social groups toward both the definition and the relative seriousness of criminal offenses and, in turn, between such attitudes and their expression in the criminal law. In this frame, criminal statutes on homicide are widely cited as supporting the principles of the consensus model (see Jacoby & Cullen, 1999). First, crime seriousness studies consistently show that regardless of social and cultural variation, many forms of homicide, unlike other offenses, are almost universally ranked by the public as the most serious of criminal crimes (e.g., Herzog, 2003; Rossi et al., 1974; Sellin & Wolfgang, 1964; Wolfgang et al., 1985), demanding the most serious of criminal punishments (e.g., Applegate et al., 1993; Jacoby & Cullen, 1999; Mitchell, 1998). These views are widely reflected in criminal codes and their determination that the most severe punishments long terms of imprisonment, mandatory life imprisonment without parole, and even the death penalty should be handed down in the case of homicide (Ashworth, 1999). Second, although homicide is a heterogeneous criminal phenomenon, criminal statutes often categorize it very similarly, on the basis of the criminal intent of the killer at the time of the offense mens rea. As a result, homicides are usually subdivided into three separate categories of seriousness: murder (intentional and planned homicide), manslaughter (intentional but unplanned homicide), and negligent (involuntary and reckless) homicide (Ashworth, 1999). Several studies (e.g., Applegate et al., 1993; Mitchell, 1998; Rossi et al., 1974) have found that criminal intent constitutes an important factor in public evaluations of the seriousness of homicides.

3 An additional characteristic of formal codes dealing with homicide (and most offenses) is the principle of individual autonomy (Ashworth, 1999): In general, individuals are expected to have both the capacity for self-control and sufficient free will to make meaningful choices; hence, they should be seen as responsible for their behavior. Thus, a person who selects a criminal response such as an unlawful killing from a range of various behaviors should be held individually accountable for the consequences of that response. A salient implication of this principle is the almost exclusive focus of the homicide law on the side of the killer, both on his/her behavior (actus reus) and especially on his/her state of mind while committing the killing. Actually, this principle separates killer from victim almost totally, blaming only the former for the lethal consequence of his/her individual act. A clear exception to this rule is provided by the recognition of provocation by the victim (by his/her words and/or acts) as a possible mitigating circumstance in the frame of the killers mens rea. This recognition allows some intentional but undeliberated homicides to be treated as manslaughter unlawful killings resulting from an instantaneous, unthinking reaction to a particular provocation, in the frame of a sudden and temporary loss of temper or heat of passion. 1 This distinction means that on conviction, unlike murder, which usually carries a mandatory sentence of life imprisonment, the sentence is at the discretion of the judge (Ashworth, 1999).

Victim-precipitated homicides
Von Hentig (1948) and Wolfgang (1958), the most notable critics of this traditional offender-oriented nature of criminology and the criminal law, proposed a more dynamic interactive approach to the offender-victim relationship. In their opinion, the formally sharp division between killer and victim is often archaic and sometimes even

4 completely unrealistic: It is based on the stereotypical perception of victims as vulnerable, passive, innocent bystanders, shrinking from assaultive situations, and of killers as brutal, strong aggressors, seeking out their victims. In fact, this separation often constitutes an artificial distortion of the reality: In certain instances it may be unclear who is acting and who is being acted upon. Indeed, the eventual victim may actually precipitate the criminal act that ultimately leads to his/her lethal victimization. They postulated that sometimes it is the victim rather than the killer who determines whether there will be a homicide. They also argued that in some cases, mere chance determines which person becomes the victim and which the killer (von Hentig, 1948; Wolfgang, 1958). Based on these views, Wolfgang (1958: 252-4) introduced the concept of victim- precipitated (henceforth, VP) homicide to refer to those homicide cases in which the victim was a direct and positive precipitator of his/her own death. Within the broad concept of provocation, VP homicide is usually defined (and measured) as a homicide in which the victim is the first in the homicide drama to use physical force against his/her subsequent slayer.2

The present study


Formally, VP homicides, despite the active and determinant role of the victims in their own demise, are not differentiated from manslaughter cases in general: VP exists when killers are shown to have had no deliberated intention to kill and victims clearly provoked them to commit the lethal act. Despite their similarity with non-VP cases, the literature contains no evidence specifically relating to public perceptions of the seriousness of VP homicides. It is not clear yet whether the lay public shares the belief that VP homicides, resulting from an extreme, aggressive provocation by the deceased victims, should be considered an additional mitigating circumstance.

5 Pursuant to the consensus model, the aim of the present study was to examine empirically whether the formal lack of recognition of VP homicides as a separate, less serious category in homicide law reflects the publics view on this issue. To this end, public perceptions regarding the seriousness of VP homicides were assessed and compared with the perceived seriousness of non-VP homicide cases.3 Concretely, respondents from a large national sample were asked to evaluate the seriousness of hypothetical, multidimensional scenarios entailing both VP and nonVP homicide events. Because VP homicides are not differentiated by the law from other manslaughter events, all the evaluated homicide events in this study were in fact manslaughter cases, in which provoked persons killed their victims with intention but without premeditation, under a sudden loss of temper. However, unlike non-VP cases, VP scenarios described the victim as explicitly being the first in using physical force against the killer. Operationally, the research goal was to assess whether significant differences in the seriousness scores assigned to the scenarios would be found between VP and non-VP cases. Based on the rationale that within the concept of provocation, prior use of force by the victim may constitute an additional justification for the homicide, rendering the provoked, assaulted killer less culpable or responsible in the public eye, it was hypothesized that VP will constitute an important factor in public evaluations of the seriousness of the homicide. Operationally, it was expected that respondents would assign significantly lower seriousness scores to VP cases than to non-VP cases. The present study has various implications, both theoretical and practical. On the theoretical level, the wide literatures on crime seriousness, homicide in general, and particularly on VP cases, contain no evidence on public perceptions of VP homicides; thus the present study may be perceived as filling a gap in this literature in various

6 theoretical issues. Moreover, this study adds to the common applicability of the consensus model to the case of homicide, by analyzing whether the lack of formal recognition of VP as a separate, less serious category of homicide does indeed reflect the view of the public on the issue. If the formal consideration of VP homicide as an additional form of manslaughter is based on principles of public consensus, then no significant differences should be found among respondents with regard to the perceived seriousness of VP and non-VP events. Alternatively, if the consensus model is not applicable to the homicide law in VP situations, then significant differences should emerge in this comparison. On the practical level, it has often been argued that differences between public attitudes and the position of the law may adversely affect public trust in the legal system, which, in turn, may undermine public willingness to comply with the law and to rely on the criminal justice system (Roberts, 1992). Moreover, the consensus model is based on the rationale that democracies are founded on the principle of elected representation and popular control of the law by the citizenry; hence, the law should reflect public views.

The research design


To overcome the limitations of direct poll data (see Roberts, 1992), social scientists developed the simple scenario method. This approach has been used widely in assessing public perceptions of the seriousness of a variety of offenses (see Rossi et al., 1974; Sellin & Wolfgang, 1964; Wolfgang et al., 1985), including homicide (see Mitchell, 1998). One of the primary weaknesses of this approach, however, is its inability to take into account multiple factors surrounding the scenario that may influence public perceptions of it (Applegate et al., 1993; Jacoby & Cullen, 1999; Roberts, 1992). This problem was overcome in this study by applying a factorial-

7 design methodology. This factorial-design methodology uses multidimensional hypothetical scenarios presented in a form that combines the benefits of controlled, randomized experimental designs and conventional surveys (see Rossi & Anderson, 1982). The scenarios are created by randomly selecting values (levels) from each of several variables (dimensions; one level per dimension per scenario) until each dimension is represented and a complete scenario is formed. Thus, all scenarios considered represent a random sample of all possible scenarios available from the universe of all levels across the chosen dimensions. Rossi and Anderson (1982) note that by permitting multiple dimensions of a crime scenario to vary randomly across scenarios (in this case, VP and non-VP cases), and by controlling the personal characteristics of the respondents, this technique allows for exploration of the effects of several variables simultaneously, while permitting unbiased estimates of the contributions of each of them to the overall judgment of the respondent.

Method
The findings of crime seriousness studies conducted in Israel (e.g., Herzog, 2003), and the formal reference of Israeli criminal law to homicide and manslaughter events (based on English common law) are very similar to those described previously; hence the suitability of Israel as a research location for the analysis of seriousness perceptions of VP homicides. The research data were collected from a random sample of the adult Israeli population (n = 805). Recent Israeli telephone directories (2001), covering all geographical regions, provided the sampling framework, and the application of a systematic random sampling method assured identical probability of inclusion of all households listed.4 The research sample was 55 percent female, 73 percent Jewish, 61 percent secular, and 81 percent were veteran citizens. Two thirds

8 of the sample reported monthly familial incomes higher than 5,000 NIS (about $1,000 US). These data show a very close fit with recent official national data on the Israeli population (CBS, 2003). Finally, the sample was highly heterogeneous in terms of age (mean age: 35; SD = 13.7; range: 15-81), and the educational level of the respondents was relatively high (13 years, i.e., some post-secondary education). Compared with the general population, there is a bias in the direction of older and consequently more educated respondents due to the exclusion of very young respondents. Respondents seriousness perceptions of hypothetical manslaughter scenarios (n = 2,288) were collected by anonymous questionnaires,5 administered by means of a telephone survey (response rate = a high 68 percent; interview length: 7-10 minutes). Data collection took place during January and February 2002. A content analysis of Israels major newspapers revealed no coverage of irregular homicide events committed around the time of the survey, which might have affected respondents attitudes.

Research variables
Respondents were requested to evaluate the seriousness of the manslaughter scenarios by choosing one value on a Likert scale ranging from 1 = Not serious at all to 11 = Very serious. This measure constituted the dependent variable of the research. To increase the uniformity of the evaluative task, respondents were told at the beginning of the interview that their responses should be based on their personal evaluation of the scenarios and not on their knowledge of the Israeli criminal law (see Herzog, 2003; Rossi et al., 1974; Sellin & Wolfgang, 1964). The 17 independent and control variables for this study were derived from two sources: 10 from the randomly assigned factorial dimensions within the scenarios (see Appendix 1), and seven from the respondent characteristics (as detailed in Table 1),

9 obtained at the end of the questionnaire, after the evaluation of the scenarios. Based on the multidimensional characterization of VP homicides (see Wolfgang, 1958), all scenarios contained 10 dimensions: two related to the homicidal act (homicide motive, modus operandi; dimensions A and B, respectively), five to the killers characteristics (gender, ethnicity, age, criminal record, and occupational status; dimensions C to G, respectively), and three to the victims characteristics (gender, VP, and criminal record; dimensions H to J, respectively). As stated, to construct a given scenario, one level from within each of the 10 dimensions was selected, with each given equal probability within each dimension (see two sample scenarios in Appendix 1).6 Despite the use of the factorial-design methodology, some characteristics were kept uniform across all of the evaluated scenarios: First, they were described in such a way that there could be no question as to the responsibility of the killers (and in VP cases also of the victims), and the lethal consequences of their acts. Second, all of the scenarios involved a single killer and a single victim. Of the scenario dimensions, VP (I) constituted the independent variable in this study. Thus, in addition to the seven personal variables of the respondents, the remaining nine random dimensions in the scenarios constituted control variables. Scenarios representing alternatively VP and non-VP cases were randomly presented to the respondents for evaluation. The measure of VP was a dichotomous variable coded 1 if the victim provoked the killing by his/her own physical attack against his/her subsequent slayer; it was also stated that the victims resort to violence occurred immediately prior to the lethal attack (see Appendix 1). To avoid influencing the respondents, the phenomenon of VP homicide was not specifically mentioned in the scenarios.

10

Results
The descriptive statistics of the data show that both non-VP and VP manslaughter cases were considered serious acts by the whole sample of respondents (means of 9.86 and 9.31, SD = 1.94 and 2.30, respectively; maximum possible value = 11). However, a statistically significant difference (t-test) was found between them: VP homicides were perceived as significantly less serious than non-VP homicides (p < 0.01). To interpret effects of other scenario dimensions or of respondent characteristics on the seriousness evaluations of the scenarios, the influence of the independent variable on the dependent variable was also analyzed using a multivariate OLS regression model. Table 1 presents the unstandardized regression coefficients and standard errors of all the scenario dimensions and the respondent variables included in the model for the whole sample of respondents. Due to the large number of scenarios included in these analyses, the significance level was reduced to 0.01. [Insert Table 1 about here] Table 1 shows that when scenario dimensions and respondent characteristics were taken into account, the independent variable of this study VP homicide (I) still significantly influenced the seriousness perceptions of the scenarios: As previously detailed, VP homicides, in which the victim was a direct precipitator of his/her own death by means of the use of physical force, were considered significantly less serious than non-VP homicides, in which the victim did not use physical force against the offender. Despite this significant effect, note that some levels of the control variables also showed significant coefficients in this analysis, at similar or even higher levels of significance. On the one hand, for the scenario dimensions, significant coefficients were found for manslaughter scenarios involving prolonged domestic abuse (variable A), and/or by and against women (variables C and H, respectively). Generally,

11 domestic abuse homicides and/or homicides committed by female offenders were perceived as significantly less serious than homicides committed against women, which were perceived as significantly more serious than homicides committed against men or when the gender of the victim was not mentioned. Note also that young and/or non-Jewish respondents assigned significantly lower seriousness scores to the scenarios, compared to their counterparts.

Discussion
The aim of the present study was to assess public perceptions of the seriousness of VP manslaughter scenarios, and to compare these perceptions to parallel seriousness perceptions of non-VP manslaughter scenarios. The study chose the consensus model as its theoretical frame, and analyzed whether the formal lack of recognition of VP homicides in criminal codes reflected public views on this issue. The hypothesis of this study was supported by the findings: When evaluating VP scenarios, varying randomly on a number of dimensions and representing the victim in the homicide drama as being the first to use physical force against his/her killer, Israeli respondents perceived them to be significantly less serious than non-VP scenarios, also varying randomly on a number of dimensions. Moreover, this finding remained stable even when two groups of control variables additional scenario dimensions and respondents personal characteristics were taken into account in a regression analysis (Table 1). Thus, these findings seem to suggest that, unlike most Western criminal statutes, the public tends to recognize VP, that is, the previous use of physical force by the victim against the killer, as a special form of provocation, and consequently, as an important factor affecting the seriousness of homicides. In other words, the Israeli public seems to see some forms of manslaughter more closely

12 linked to self-defense than the law formally recognizes. That is, VP homicide cases seem to constitute, at least for the public's eyes, a less serious type of manslaughter than non-VP homicide cases. It should be added here that despite its statistical significance, the difference in seriousness means between non-VP and VP cases was far from being important. Although when compared with non-VP events, a significantly lower mean was obtained for the perceived seriousness of VP cases, both are basically high-order means, separated by only half a point (9.86 and 9.31). Taking into account the range of the seriousness scale (1-11), this decrease in perceived seriousness, although statistically significant, should be seen as minimal rather than substantial. Table 1 reinforces this interpretation: It shows that in the regression analysis, the most influential variable on respondents evaluations of seriousness was not VP but rather additional specific characteristics of the scenarios (a homicide motive - domestic abuse -, and killers and victims gender). Assuming that the statistical significance of VP (and of a few personal variables) was mainly obtained due to the relatively high number of cases considered (more than 2,000 scenarios), it can be concluded that these public perceptions were hardly affected by VP (and social variation). Hence, these findings seem to indicate that because the effects of homicide are always irreversible and absolute, the Israeli public, irrespective of social divisions, simply perceives non-VP and VP cases quite similarly, and experiences a similar level of revulsion against similar manslaughter cases, committed as a heated reaction to prior use of physical force by the victim as it does for other less serious forms of loss of temper (e.g., altercations). This implies that the current use of the concept of provocation, and the formal lack of recognition of VP homicides as a separate less serious category of manslaughter in criminal statutes seem to reflect the general

13 public will on this issue, thus providing additionally empirical support for the consensus model. Finally, despite the effort invested in this study to overcome theoretical and methodological obstacles, its limitations need to be taken into account when analyzing its conclusions. Due to the application of the factorial survey approach and other methodological considerations (such as the use of a telephone survey), this research was based on short hypothetical scenarios depicting typical manslaughter situations. In this context, it may be assumed, on the one hand, that other factors not considered in these scenarios, such as the former relationship between offender and victim, and other factors associated with intent and self-defense, might influence the respondents' seriousness judgments. Hence, further analysis of the question and hypothesis raised by the present study using other survey techniques, in other national contexts, and with more extensive descriptions of crime situations is highly recommended.

14

Notes
1. According to English law, it has to be proved that the provocation was of such a nature or degree that the offender lost self-control and killed the victim during the period of sudden rage and that any reasonable person would have reacted in this way. 2. Studies based on formal police data provided differing levels of VP homicides; Wolfgang (1958) found that 22 percent of all homicide cases were VP. Note that this narrow definition of VP excludes passive VP (the victim exhibits a personal characteristic that unknowingly either threatens or provokes the attacker), suicidal events (a suicidal individual engages in a calculated life-threatening criminal incident in order to force another individual, usually a police officer, to kill him/herself), and role reversal (the eventual victim sets out to commit a crime and is foiled by a resisting person who is sometimes charged with a crime). 3. This study was based on the (usually incidental) finding that despite the consensually extreme perceived seriousness of homicide acts, perceptions of seriousness differed for different types of homicide (e.g., see Mitchell, 1998). 4. According to formal data, 98 percent of households in Israel have telephones, and the percentage of people unlisted in the directories is fairly low. 5. Originally, each respondent evaluated 10 different homicide situations (n = 8,050 scenarios), including not only manslaughter cases. As stated, because of the present research question, only the manslaughter cases were analyzed in this study. 6. The construction of the scenarios deviated slightly from complete randomization, introducing no important problems of multicollinearity. For example, when the killing was committed due to romantic infidelity and domestic abuse (dimension A), the gender of both the offender and the victim were stated (dimensions C and H, respectively).

15

References
Applegate, B., Wright, J., Dunaway, R., Cullen, F. & Wooldredge, J. (1993). Victim-offender race and support for capital punishment: A factorial design approach. American Journal of Criminal Justice, 18 (1): 95-115. Ashworth, A. (1999). Principles of criminal law. Oxford: Oxford University Press. Central Bureau of Statistics (2003). Statistical abstract of Israel. Jerusalem: CBS. Herzog, S. (2003). Religiosity and perceptions of crime seriousness by Jewish and Muslim respondents in Israel. Deviant Behavior, 24 (2): 153-174. Jacoby, J. & Cullen, F. (1999). The structure of punishment norms: Applying the Rossi-Berk model. The Journal of Criminal Law and Criminology, 89 (1): 245-307. Mitchell, B. (1998). Public perceptions of homicide and criminal justice. British Journal of Criminology, 38 (3): 453-472. Roberts, J. (1992). Public opinion, crime and criminal justice. In M. Tonry (Ed.), Crime and Justice: An Annual Review of Research, vol. 16. Chicago: University of Chicago Press. Rossi, P. & Anderson, A. (1982). The factorial survey approach: An introduction. In P. Rossi & S. Nock (Eds.), Measuring social judgments: The factorial survey approach. Beverly Hills, CA: Sage. Rossi, P., Waite, E., Bose, C. & Berk, R. (1974). The seriousness of crimes: Normative structure and individual differences. American Sociological Review, 39 (2): 224-237. Sellin, T. & Wolfgang, M. (1964). The Measurement of Delinquency. New York: Wiley. von Hentig, H. (1948). The criminal and his victim. New Haven: Yale University

16 Press. Wolfgang, M. (1958). Patterns in criminal homicide. New York: Wiley. , Figlio, R., Tracy, P., & Singer, S. (1985). The National Survey of Crime Severity. Washington, D.C.: USGPO.

17 Table 1: Unstandardized Regression Coefficients and Standard Errors for the Seriousness of Manslaughter Scenarios (n = 2,288) by Scenario Dimensions and Respondents Personal Details, for the Whole Sample of Respondents (n = 805) 1
Variables A. Homicide motive 2 Values Financial dispute Romantic infidelity Domestic abuse Shooting Stabbing Strangling Male Female Jewish Arab 25 years old 50 years old 0 = No ; 1 = Yes Unemployed Working in a stable job Male Female 0 = No ; 1 = Yes 0 = No ; 1 = Yes 0 = Female ; 1 = Male Interval (in years) Interval (in schooling years) 0 = Less than 5,000 NIS ; 1 = More 0 = Jewish ; 1 = Not Jewish 0 = Secular ; 1 = Not secular 0 = Veteran ; 1 = New immigrant Valid n R2 Constant d.f. Coefficient (SE) -0.13 (.17) -0.30 (.21) -0.76* (.20) -0.04 (.13) 0.03 (.13) -0.14 (.18) -0.16 (.20) -0.58* (.30) -0.15 (.12) 0.10 (.13) 0.19 (.11) 0.14 (.14) 0.35 (.12) 0.06 (.11) -0.07 (.13) -0.20 (.20) 0.40* (.20) -0.40* (.10) 0.25 (.13) -0.11 (.09) 0.01* (.00) -0.02 (.01) 0.02 (.10) -0.36* (.12) -0.03 (.10) -0.12 (.12) 2,112 .085* 9.40* (.33) 24

B. Modus operandi 3

C. Killers gender 3 D. Killers ethnicity 3 E. Killers age 3 F. Killers criminal record G. Killers occupational status 3 H. Victim's gender 3 I. Victim-precipitated homicide J. Victims criminal record K. Respondent's gender L. Respondents age M. Respondents education N. Respondent's income O. Respondent's ethnicity P. Respondent's religiosity Q. Respondent's status Model data

* p < 0.01 1. Due to the skewed distribution of the dependent variable, the same analyses were conducted using the natural logs of the variable scores. Moreover, due to possible within-respondent variability, the regression analyses were also conducted by HLM procedures. Generally, both regression variations produced very similar results to those presented. 2. Dummy variable, Political dispute is the reference group. 3. Dummy variable, Not stated is the reference group.

18

Appendix 1
A. Homicide motive 1. Romantic infidelity / 2. Financial dispute / 3. Political dispute / 4. Domestic abuse B. Modus operandi 1. Shooting (weapon) / 2. Stabbing (knife) / 3. Strangling (hands) / 4. (Not stated) C. Killers and H. Victims gender 1. Male / 2. Female / 3. (Not stated) D. Killers ethnicity 1. Jewish / 2. Arab / 3. (Not stated) E. Killers age 1. 25 years old / 2. 50 years old / 3. (Not stated) F. Killers and J. Victims criminal record 1. Yes / 2. (Not stated) G. Killers occupational status 1. Unemployed / 2. Working in a steady job / 3. (Not stated) I. Victim-precipitated homicide 1. Yes / 2. No

Some sample scenarios: 1. Two men enter into a furious political argument during which one of them, a man with a criminal record, hits the other and, in response, in the heat of the dispute, the latter, a 50-year-old, unemployed Jewish man, strangles the former to death.

2. A man and a woman enter into a furious argument concerning suspicion of romantic infidelity during which, in the heat of the dispute, one of them, a 25-year-old Arab man with a criminal record, shoots the woman to death.

Anda mungkin juga menyukai