xml version 1.0 encoding UTF-8 standalone no
record xmlns http:www.loc.govMARC21slim xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.loc.govstandardsmarcxmlschemaMARC21slim.xsd
leader nam 22 Ka 4500
controlfield tag 007 cr-bnu---uuuuu
008 s2010 flu s 000 0 eng d
datafield ind1 8 ind2 024
subfield code a E14-SFE0003493
Is there an "innocent female victim" effect in capital punishment sentencing?
h [electronic resource] /
by Amelia Kirkland.
[Tampa, Fla] :
b University of South Florida,
Title from PDF of title page.
Document formatted into pages; contains X pages.
Thesis (M.A.)--University of South Florida, 2010.
Includes bibliographical references.
Text (Electronic thesis) in PDF format.
Mode of access: World Wide Web.
System requirements: World Wide Web browser and PDF reader.
ABSTRACT: Disparities in the administration of capital punishment are a prominent social and political issue. While the focus of death penalty disparity research initially lay with the defendant and how the defendant's race or ethnicity affects sentencing outcomes, only marginal support for offender effects has been found. A consistent finding, however, is that victim race has a significant effect on capital sentencing outcomes. Recent examinations of the joint effects of victim characteristics indicate that victim gender also has some influence in capital sentencing decisions. While these prior studies have examined the interactive effects of victim gender and victim race the current study proposes that victim-related variables other than race may be important components in understanding the female victim effect. This analysis is focused on understanding the joint effects of victim gender in terms of identifying an "innocent female victim" effect. Based on prior studies and theoretical perspectives, three hypotheses are proposed and tested here using a sub-population of capital cases in North Carolina between the years 1990 and 2007: 1. Cases with a female victim and male defendant will be more likely to result in the death penalty than other defendant-victim gender dyads, 2. Cases with a female victim and stranger defendant will be more likely to result in the death penalty than other dyads, and 3. Cases with a female victim who was not involved in illegal activity at the time of her victimization will be more likely to result in the death penalty than other dyads. The results indicate that victim conduct (illegal activity) and victim gender both play a role in jury sentencing recommendations, but regardless of victim conduct, cases with a female victim are the most likely to result in the death penalty. Therefore, this study finds marginal support for an "innocent female victim" effect in jury decisions to recommend the death penalty, but consistent support for a "female victim" effect. Conclusions and implications of the findings are discussed.
Advisor: M. Dwayne Smith, Ph.D.
t USF Electronic Theses and Dissertations.
Punishment Sentencing? by Amelia Lane Kirkland A thesis submitted in partial fulfillment of the requirements for the degree of Master of Arts Department of Criminology College of Behavioral and Community Sciences University of South Florida Major Professor: Dwayne Smith, Ph.D. Thomas Loughran, Ph.D. Sondra Fogel, Ph.D. Date of Approval: April 15 2010 Keywords: homicide, death penalty, sentencing, gender, jury decision making Copyr ight 2010, Amelia Lane Kirkland
Acknowledgements This thesis would not have been possible without the support of my thesis committee with whom I had the fortunate opportunity to work. Dr. Smith, thank you for your faith in me throughout this process. As always, your editorial eye and knowledge on this subject were invaluable. The willingness to help me that you have expressed, including the use of the data for this project, is greatly appreciated. Dr. Loughran, I sincerely thank you for your patience with me as well as your guidance through all stages of this project. I would also like to thank you for introducing me to the concept of brevity and know that in the future I will try not to use seven words when four will do Dr. Fogel, thank you for your insights and support. I am grateful that this thesis provided an opportunity for me to work with you. I appreciate the time and effort you have all put into assisting me with this thesis as well as my general educational development.
i Table of Contents List of Tables ................................ ................................ ................................ ...................... ii Abstract ................................ ................................ ................................ .............................. iii Chapter 1 Introduction ................................ ................................ ................................ ......... 1 Chapter 2 Review of Literature ................................ ................................ ............................ 3 The Role of the Victim in Capital Sentencing ................................ ......................... 3 Theoretical Frameworks ................................ ................................ .......................... 7 Focal concerns theory ................................ ................................ .................. 8 Chivalry/paternalism hypothesis ................................ ................................ 10 The Role of Gender and the Innocent Female Victim ................................ ........... 11 Chapter 3 Current Study ................................ ................................ ................................ .... 13 Hypotheses ................................ ................................ ................................ ............. 13 Victim gender and offender gender ................................ ........................... 13 Victim gender and victim offender relationship ................................ ........ 14 Victim gender and victim conduct ................................ ............................. 14 Chapter 4 Methods ................................ ................................ ................................ ............. 16 Data ................................ ................................ ................................ .................... 16 Dependent variable ................................ ................................ .................... 19 Independent vari able and control variables ................................ ............... 19 Analytic Plan ................................ ................................ ................................ .......... 21 Chapter 5 Results ................................ ................................ ................................ ............... 24 Difference in Proportions Analysis ................................ ................................ ........ 25 Logistic Regression Analysis ................................ ................................ ................. 28 Chapter 6 Discussion ................................ ................................ ................................ ......... 33 References ................................ ................................ ................................ .......................... 40
ii List of Tables Table 1 Descriptive Statistics of Variables (N=821) ................................ ......................... 20 Table 2 Difference in Proportions of Death Sentence Recommendations Tes ting for Main Effects (N=821) ................................ ................................ ......... 2 7 Table 3 Difference in Proportions of Death Sentence Recommendations Main and Interaction Effects (N=821) ................................ ................................ ... 28 Table 4 Logistic Regression Analysis Testing for Main Effects (N=821) ....................... 30 Table 5 Logistic Regression Analysis Testing for Interaction Effect of Female Victim and Stranger Defendant (N=821) ................................ .................. 32
iii Punishment Sentencing? Amelia Lane K irkland ABSTRACT Disparities in the administration of capital punishment are a prominent social and political issue. While the focus of death penalty disparity research initially lay with the entencing outcomes, only marginal support for offender effects has been found. A consistent finding, however, is that victim race has a significant effect on capital sentencing outcomes. Recent examinations of the joint effects of victim characteristics in dicate that victim gender also has some influence in capital sentencing decisions. While these prior studies have examined the interactive effects of victim gender and victim race the current study proposes that victim related variables other than race may be important components in understanding the female victim effect. This analysis is focused on understanding the joint effects of victim gender in theoretical perspectiv es, three hypotheses are proposed and tested here using a sub population of capital cases in North Carolina between the years 1990 and 2007: 1. Cases with a female victim and male defendant will be more likely to result in the death penalty than other defe ndant victim gender dyads, 2. Cases with a female victim and stranger defendant will be more likely to result in the death penalty than other dyads, and 3. Cases with a female victim who was not involved in illegal activity at the time of her
iv victimization will be more likely to result in the death penalty than other dyads. The results indicate that victim conduct (illegal activity) and victim gender both play a role in jury sentencing recommendations, but regardless of victim conduct, cases with a female v ictim are the most likely to result in the death penalty. Therefore, this study finds and implications of the findings are discussed.
1 Chapter 1 Introduction Disparities in the administration of capital punishment are a prominent social and political issue. The question researchers and policy makers are faced with is the extent to which capital sentencing disparities are a product of extra legal factors that can be identified and addressed. The existing death sentencing disparity literature has concentrated largely on the influence of race and ethnicity on sentencing outcomes (s ee Kavanaugh Earl, Cochran, Smith, Fogel, & Bjerregaard, 2008). Examining the influence of race, gender, and the socio economic status of defendants has been a focus of death penalty disparity research. Recent studies, however, have expanded the scope of c apital sentencing disparity research to include examinations of the role of victim gender and race (Hindson, Potter, & Radelet, 2006; Holcomb, Williams, & Demuth, 2004; Stauffer, Smith, Cochran, Fogel, & Bjerregaard, 2006; Williams & Holcomb, 2004; William s, Demuth, & Holcomb, 2007). Findings from these studies and other qualitative works (e.g. Sundby, 2003) have suggested the presence of female victim effect as well as an decisi ons. Qualitative examinations of the extent to which jurors draw distinctions Project (Sundby, 2003); however little research has been conducted that looks empirically at vi with victim gender.
2 In their 2007 study, Williams, Demuth, and Holcomb claim that there is an unfortunate lack of interest in understanding the relationship between vi ctim gender and capital sentencing outcomes. Therefore, the current study intends to further explore the role of victim gender in jury decisions to recommend death. In order to understand the importance of the current study the research conducted thus far on victim characteristics and the death penalty will first be discussed; empirical and theoretical support for the examination of victim gender and gender related variables will be presented; and the findings of the current study will be explicated and dis cussed.
3 Chapter 2 Review of Literature The Rol e of the Victim in Capital Sentencing The death penalty is one of the most polarizing criminal justice issues in social, political and academic forums. The history of the death penalty in the United States has had an influential effect on its polarizing quality. Two U.S. Supreme Court decisions have raised grave concerns about the equality in the administration of capital punishment, Furman v. Geor g i a (1972) and McClesk e y v Kemp (1987). Furman v. Georgia (1 972) addressed questions about whether the death penalty was being carried out in a cruel and unusual manner thus constituting violations of the eighth and fourteenth amendments. At ication, particularly where non White defendants were concerned. This decision led to a moratorium on death sentencing in the United States, a moratorium that ended in 1976 with Gregg v. Georgia and a set of accompanying decisions. However, concerns remain ed that Gregg had not served to diminish the perceived inequality that served as the basis of Furman The issue came to a head in McCleskey v. Kemp (1987) whereby a s post Furman in McCleskey rested on research conducted with the intent of determining the pervasiveness of racial bias in capital sentencing, later reported in Baldus, Woodworth, and Pulaski (1990). However, the McCleskey court cited several methodological issues
4 with the study, concluding that the findings did not carry enough weight to provide convincing evidence race based disparity (Kavanaugh Earl et al., 2008). Nevertheless, t remain. The Baldus et al. (1990) study added to an existing body of research on use of the death penalty, but has also served to spur much subsequent research. In the same vein as the Baldus et al. research, many seminal death penalty studies have focused on offender characteristics, specifically offender race. A consistent finding of studies examining race and the death penalty is that perpetrator race is only marginally re lated to receiving a death sentence (Williams & Holcomb, 2004); however, a secondary finding in many disparity studies is that race of the victim, specifically White victims is commonly a significant predictor of receiving the death penalty, particularly when joined with defendant race (Baldus, Woodworth, Zuckerman, Weiner, & Broffitt, 1998; Hindson et al., 2006; Paternoster, 1984 ; Williams & Holcomb, 2004). The examination of multiple victim characteristics in capital jury sentencing recommendations has t hus become a sub focus of death penalty research. Due to extensive concerns about the correlation between victim race and death outcomes, a finding which had become well established by the 1990s (Hindson et al., 2006), four primary studies in the past dec ade have attempted to expand this area of inquiry by exploring the interactive effects of victim characteristics (see Holcomb et al., 2004; Williams & Holcomb, 2004; Williams et al., 2007; Stauffer et al., 2006). The ikely that decision makers consider the race or
5 Meaningful differences may exist when multiple characteristics are examined interactively and these interactions provide a mo re accurate understanding of decision omes. Using Ohio Supplementary Homicide Reports (SHR) data from the years 1981 through 1994, as well as data from multiple state sources on the homicides resulting in the death penalty Williams and Holcomb (2004) examined interactive victim effects with a sample of 5,320 cases, 271 of which resulted in the death penalty. The results indicated that homicides with a White female victim were significantly more likely to result in the death penalty than cases with other victim gender race dyads. Specifically, with White female victims as the reference category, there was a 65.8 % decrease in the odds of receiving the death penalty for cases with a White male victim, a 62.4% decrease in the odds for cases with a Black female victim, and a 73.7% decrease in the odds for cases with a Black male victim (Williams & Holcomb, 2004). These findings were replicated by Holcomb et al. (2004) with the same Ohio SHR data extended through 1997. Additional analysis conducted in the Holcomb et al. study indicated that differen t factors were associated with death sentences for cases with White male victims, Black male victims, and Black female victims, but no additional variables included in the analysis were significant in predicting death sentences for homicides with White fem ale victims. As an extension of the work conducted by Williams and Holcomb (2004), Stauffer et al. (2006) sought to determine if the White female victim effect would emerge in a sample of death penalty cases in North Carolina. A sample of 953 jury decisio ns in
6 capital murder trials from the years 1979 through 2002 were the object of the analysis. In addition to including variables used in previous Williams and Holcomb works, Stauffer et al. also included additional variables deemed relevant in the sentenci ng literature. When using the variables modeled by Williams and Holcomb, Stauffer et al. found that compared to cases with White female victims, cases with Black male victims and cases with White male victims were significantly less likely to result in the death penalty. However, departing victims were not found to be significantly more likely to result in the death penalty compared to cases with Black female victims. Furthermore, when expanding upon the model of variables introduced by Williams and Holcomb to include prior criminal behavior, the involvement of rape in the crime, attorney type, victim involvement in illegal activity, and the number of aggravators accepted by the jury, the interact ions effects were no longer statistically significant. Stauffer et al. concluded that there are nuances in empirically assessing the interactive effects of victim characteristics. Relevant to the current analysis, they also found that other characteristics related to the victim, such as involvement in illegal activity and the relationship between victim and offender, were significant factors in death sentencing. In an attempt to further explore the interactive effects between victim gender and victim race, Williams et al. (2007) included an analysis of sex related victimization. Using the Baldus et al. (1990) study data, which includes a stratified random sample of 1,066 core weighted cases in which the defendant was indicted for murder or voluntary manslau ghter and convicted in Georgia between the years 1973 and 1979, Williams et al. utilized logistic regression to examine the role of victim gender within this seminal
7 dataset. The results indicated joint effects of victim gender and race and that specifical ly cases with White female victims were treated the most harshly and cases with Black male victims the most leniently. Additionally, they found that when controlling for sex related variables the female victim effect was minimized concluding that the sexua l nature of some female victimization may explain the female victim effect. These recent examinations of the joint effects of victim characteristics indicate that victim gender has some influence in capital sentencing decisions. Specifically, cases with f emale victims are more likely to result in harsher punishments (the death penalty compared to life in prison). While these prior studies have examined the interactive effects of victim gender and victim race the current study proposes that victim related v ariables other than race may be important components in understanding the female victim effect. The current analysis is focused on understanding the joint effects of victim l approaches from prior research as a guide, the role of victim gender in capital punishment is explored by examining potential mechanisms (beyond race) that are hypothesized to influence the female victim effect in jury decision making. Theoretical Framew orks Limited theoretical explanations exist that explain victim gender effects in sentencing outcomes; however potential theoretical frameworks have been posited. Focal concerns theory and the chivalry/paternalism hypothesis have both been discussed in the existing literature examining victim characteristics in sentencing. The development of the current study is based on some of the basic tenets of these theories. It is important to note, however, that this study does not purport to test either of these the ories but rather
8 employees these theories as a guideline for identifying victim related variables that are hypothesized to have interactive effects with victim gender. Focal c oncerns theory. Literature examining the relationships between race, gender, and sentencing has concerns theory is often utilized to explain judges decision making based on offender s characteristics, some of the concepts are equally applicable to un derstanding potential interactions between victim characteristics. The three focal concerns that may influence blameworthi ness, protection of the community, and practical implicati ons of sentencing decisions (Steffensmeier, Ulmer, & Kramer, 1998). While these focal concerns were initially directed at the offender they can be used to explain relationships between victim characteristics and sentencing as well (see Baumer et al., 2000; Williams & Holcomb, 2004). The first focal concern, blameworthiness, is most relevant to the current study however all three focal concerns are addressed in terms of how they may explain victim effects in sentencing outcomes. The first of these concerns, blameworthiness, includes analysis of offense offenders viewed as being more blame worthy should receive harsher sentences. Our focus here, however, is on the victim and how victim characteristics can influence perceptions of offender blameworthiness. If the victim wa s engaged in illegal or improper activity at the time of the incident, jury members may view the defendant as less
9 conduct (Baumer et al., 2000; Rye Greatrix, & Enright 2006; Sundby, 2003). Juries recommending life sentences have been foun d to be more likely to discuss victim characteristics during their deliberations than death juries, and data suggests a correlation between perception of the victim and jury recommendations of a life sentence (Sundby, 2003). Qualitative findings such as th ese support the idea that victim conduct and characteristics may operate in a similar fashion as defendant conduct and characteristics when juries are making sentencing decisions. The second focal concern, protection of the community, is concerned with th e harm the offender has caused or may continue to cause in the community (Steffensmeier community may be seen as deserving of a longer sentence (or death as opposed to lif e) Offenders who are strangers to the victim may be seen as a larger threat to the community as a whole apital Jury Project indicate that jurors identify more with victims murdered by strangers and an individual who preys upon randomly chosen victims poses the starkest image of the dangerous individual, and future dangerou sness (Sundby, 2003, pp. 359). The third focal concern, practical constraints and consequences, is more a consideration prior to jury sentencing recommendations (t he legal outcome examined in the current study). This focal concern refers to considerations of the workings of the
10 justice system such as cost of prosecution, utility of plea bargaining, and considerations effensmeier et al., 1998). One aspect of this focal concern is consideration of how incarceration will affect others in terms of disruption to ties with children and family members (Steffensmeier et al., 1998). While meant to pertain to how incarceration w family, murder of a female victim may be seen as disrupting a family unit causing additional harm to other members of society and thus resulting in increased puni tiveness (Curry Lee, & Rodriguez, 2004). Chival ry/paternalism h ypothesis Previous research examining the role of gender in sentencing has often discussed the chivalry or paternalism hypothesis as an explanation for gender differences in sentencing outcomes. Traditional gender beliefs viewed women as t he weaker sex, more passive, innocent, and dependent than men (Franklin & Fearn, 2008). The chivalry hypothesis has often been utilized to explain lighter sentences given to female offenders compared to their male counterparts, but this concept can also be applied to understand victim gender effects in sentencing (Curry et al., 2004; Franklin & Fearn, 2008). While masculine norms may be less concerned with men who victimize men, men who victimize women may incur harsher punishments for acting against the no rm (Curry et al., 2004). In terms of chivalry, jurors would be more likely to condemn those who bring violence against women particularly women viewed as truly innocent (Franklin & Fearn, 2008; Rye et al., 2006). appeared repeatedly in research on rape and sexual assault but less so in examinations of capital sentencing. As indicated in the
11 literature on offender characteristics, not all women benefit from the lofty ideals of chivalry (Franklin & Fearn, 2008). For example, female minority offenders are less likely to positively benefit from notions of chivalry compared with female W hite offenders (Franklin & Fearn, 2008). This concept can also be applied to victims. Those victims that were in the wrong place at the wrong time may be seen as more deserving of paternalistic and chivalrous protections than those victims whose characteristics or actions deem them less worthy of retaliation by actors in the justice system (Rye et al., 2006; Sundby, 2003). The Role of Gen der and the Innocent Victim Prior research illustrates the utility and importance of examining multiple victim characteristics and the interactions between victim characteristics more in depth. Few studies have looked specifically at the interactive effec ts of multiple victim characteristics in terms of capital sentencing and those that have are limited to exploring joint effects of race and gender. Although the emphasis on racial disparities in capital sentencing is a valid concern, especially in light of race relations, the interaction between victim race and gender does not fully account for disparities in capital sentencing. Understanding how gender may operate as a stratification device beyond its potential interacti on with race can help shed light on criminal justice decision Based on current research examining multiple victim characteristics and theoretical approaches to understanding gender in jury decisio n making, there are multiple gender related characteristics that may have an effect on sentencing recommendations which have not been quantitatively explored. The current study attempts to blend the concept of en studied extensively in the non capital
12 sentencing literature with the capital sentencing literature that has analyzed the role of victim gender.
13 Chapter 3 Current Study Hypotheses ses and justifications for their use in the current study are explained below. Victim gender and offender g ender The desire to protect female victims may manifest in harsher punishments for those who offend against them. Based on the chivalry hypothesis men who victimize women should receive the harshest sanctions of all given that men are suppose to be the protectors of women (Curry et al., 2004; Franklin & Fearn, 2008). Though minimal research has been done on gender interactions in capital sentencing studies of non capital sentencing outcomes have shown that male offenders who victimize females receive significantly longer sentences than any other gender combination (Curry et al., 2004; Farrell & Swigert, 1986; Franklin & Fearn, 2008; Rye et al., 200 6). Therefore the first hypothesis predicts the interaction between victim gender and perpetrator gender: 1. Cases involving a female victim will be more likely to result in the death penalty than cases involving a male victim. Specifically, cases with a m ale defendant and female victim will be more likely than any other gender combination (male defendant and male victim, female defendant and male victim, female defendant and female victim) to result in the death penalty.
14 Victim g ender and victim offender r elationship As suggested by the chivalry hypothesis and the blameworthiness and community protection aspects of focal concerns theory, those victims who need the most protection (or those offenders who deserve the harshest punishments) are those with no control over most harshly to those defendants that chose their victims randomly. Jurors identified with because they could put would predict that the victims of stranger violence would be seen as less blameworthy and more deserving of protection than victims of non stran ger violence. Specifically, those women who are victims of stranger violence should be perceived as more innocent by jurors because of their lack of control over their own victimization (Sundby, 2003). Therefore the hypothesis regarding the interaction bet ween victim gender and victim perpetrator relationship is as follows: 2. Cases involving a female victim whose perpetrator was a stranger (versus a non stranger) will be more likely to result in the death penalty than cases with a female victim whose perpe trator was not a stranger, or cases with a male victim whose perpetrator was a stranger or non stranger. Victim gender and victim c onduct Prior research on the role of the victim in sentencing has indicated that victim conduct at the time of the incident can influence sentencing outcomes (Baumer et al., 2000; Rye et al., 2006; Spohn & Spears, 1996; Sundby, 2003). Killings of disreputable or stigmatized victims have been found to result in more lenient punishments for their
15 perpetrators potentially indicat ing that these offenders do not pose as great a threat to the community as those who victimize victims who are not stigmatized (Baumer et al., 2000). We predict that those female victims who were engaged in illegal activity at the time of their murder shou ld be perceived as less innocent resulting in decreased odds of the death penalty for their perpetrators. Therefore, the third hypothesis concerns the interaction between victim gender and victim illegal activity: 3. Cases involving a female victim who wa s not involved in illegal activity at the time of the incident will be more likely to result in the death penalty than cases involving a female victim who was involved in illegal activities at the time of the incidence, or cases with a male victim who eith er was or was not involved in illegal activity at the time of the incident.
16 Chapter 4 Methods Data The data utilized in this study are from the North Carolina Capital Sentencing Project (see Kavanaugh Earl et al., 2008 for a discussion of this datase t). Cases included in this dataset are homicide cases in which (a) a first degree murder conviction was secured, (b) the state sought the death penalty, and (c) the trial advanced to the sentencing phase whereby the jury recommended either a life sentence or the death penalty (Stauffer et al., 2006). Information about each case was derived from reviews of trial documents contained in North Carolina Supreme Court and Court of Appeals cases and/or from public records obtained from the counties in which the tr ial were held. Information about offenders was obtained from the North Carolina Department of Corrections website while information about victims was provided by the North Carolina medical examiners office or through use of a commercial CD, North Carolina Vital Records: Deaths 1968 1996 (Ancestry.com, 2000). Within the data, cases involving multiple offenders tried for the murder of one victim are treated as separate cases and, likewise, instances involving one offender and multiple victims are treated as s eparate cases. The population consists of 1,338 cases for the years 1977 (the year North Carolina resumed capital punishment after Gregg) through 2007 (the last year for which full data has been collected). Not only is there detailed information about the victim and the circumstances of the homicide, but the
17 data also contains significant information about specific legal factors such as aggravators and mitigators submitted and accepted in each of the cases. Following Gregg North Carolina adopted a bifurcat ed trial procedure whereby a sentencing phase is conducted if the defendant is found guilty of first degree murder. At the sentencing phase, the prosecution must prove the existence of one or more aggravating factors for the defendant to be eligible for ca pital punishment. In response, the defense is allowed to present a set of mitigating factors that are designed to argue against the imposition of a death sentence. Following this deliberation, the jury retires to deliberate the sentence. Although termed a the sentence is binding unless deemed by the trial judge to be improperly assessed. For the purposes of this study, cases from the years 1990 2007 serve as the focus of the analyses. In 1990, the U.S. Suprem e Court decision McKoy v. North Carolina changed guidelines regarding the acceptance of mitigating circumstances at trial. Prior to McKoy jurors had to unanimously decide to accept a mitigating factor; following the McKoy decision, the acceptance of miti gators no longer has to be unanimous (for a detailed discussion see Kremling, Smith, Cochran, Bjerregaard, & Fogel, 2007). In effect, prior to McKoy that the mitigator was not ac cepted by the jury. Following McKoy and still in effect today, a single juror accepting a mitigator leads to the mitigator being recorded as having been accepted. Consequently, if analyses include mitigating factors from the dataset, they are comparable a s a group for trials conducted prior to or after the McKoy decision. Because post McKoy trials are more contemporary and far more numerous, they constitute the data of this study.
18 There are 917 cases that can be analyzed for the period between April 1990 and December 2007. Of these 917 cases, 821 contained complete information on all variables. Attrition analysis was conducted on the 96 cases that were eliminated from the final sample. Comparing the cases to be eliminated to the cases that are retained ind icated that 95 of the 96 excluded cases resulted in a life sentence. Subsequent analyses revealed that this finding is considered to be a function of the capital sentencing process. As mentioned earlier, the sentencing phase begins after the guilt phase of the trial. During the sentencing phase, jurors recommend either life in prison or death. Jurors are provided an Issues and Recommendations for Punishment sheet that enumerates the sentencing decision process. Aggravating factors are presented and voted up on first and if an aggravating factor(s) is accepted, the jury moves on to consideration of the mitigating factors before providing a sentencing recommendation. If no aggravating factors are accepted, the defendant is automatically sentenced to life in pri son without the possibility of parole. Forty six of the 96 cases that are eliminated were cases in which no aggravating factors were accepted, and therefore no mitigating factors were considered. The remaining 50 cases were eliminated because data on the aggravating and mitigating factors (specifically, the Issues and Recommendation for Punishment sheets) were missing from case files. While it is possible that some of these documents were simply lost, it was discovered that judges have the discretion to d iscard the Issues and Recommendations form if a life sentence is assessed because the document will not be a part of any subsequent appeals decisions. It was discovered that a substantial number of the cases with missing Issues and Recommendation for Punis hment were those in which the jury became deadlocked, leading to speculation that the forms may not have been
19 completed by the jury, and therefore considered disposable by the trial judge. Therefore, to reiterate, attrition analysis determined that the mis sing cases did not yield concerns about the scientific differences between eliminated and retained cases for the current study. Dependent v ariable In North Carolina capital jurors have only two sentencing options: life in prison without the possibility of parole or the death penalty Therefore the dependent variable, sentence outcome, is dichotomous. Independent and control v ariables The main focus of this research is interaction between gender and victim characteristics. The independent variables of interest are victim gender, defendant gender, victim defendant relationship, and victim involvement in illegal activity. Additional control variables were chosen based on their use in previous studies using this dataset to examine the interaction of multi ple victim characteristics on capital sentencing decisions (see Stauffer et al., 2006). These include demographic characteristics of victims and offenders, and other extra legal factors that could impact sentencing decisions. As well, Stauffer et al. (2006 ) noted the importance of including as legal factors the variable that level of aggravation is a powerful predictor of death sentencing, as well as the levels of mitiga tion accepted by the jury These variables and their distributions can be found in Table 1, including the note accompanying that table.
20 Table 1 Descriptive Statistics of Variables (N=821) Variable N (%) Received Death Sentence N (%) Sentence Life 420 (51.2) Death 401 (48.8) Defendant Sex Male 792 (96.5) 391 (49.4) Female 29 (3.5) 10 (34.5) Victim Sex Male 465 (56.6) 194 (41 .7) Female 356 (43.4) 207 (58.1) Victim Illegal Activity No 676 (82.3) 346 (51.2) Yes 145 (17.7) 55 (37.9) Victim Offender Relationship Non stranger 540 (65.8) 267 (49.4) Stranger 281 (34.2) 134 (47.7) Defendant Race Whit e 330 (40.2) 169 (51.2) Non W hite 491 (59.8) 232 (47.3) Victim Race White 477 (58.1) 256 (53.7) Non White 344 (41.9) 145 (42.2) Homicide in Urban Area Non Urban 432 (52.6) 222 (51.4) Urban 389 (47.4) 179 (46.0) Attorney Type Private Attorney 33 (4.0) 9 (27.3) Public Defender 788 (96.0) 392 (49.7) Multiple Victim Homicide? No 490 (59.7) 234 (47.8) Yes 331 (40.3) 167 (50.5) Weapon Other 331 (40.3) 196 (59.2) Gun 490 (59.7) 205 (41.8) Defendant Prior Record No Prior Record 549 (66.9) 252 (45.9) Prior Record 272 (33.1) 149 (54.8) Offense involved Rape No rape 753 (91.7) 347 (46.1) Rape 68 (8.3) 54 (79.4) Note: Defendant age: M = 28.21; Range = 16 to 68. Victim age: M = 38 .66; Range = 0 to 100. The total number of aggravators accepted: M = 2.19; Range = 1 to 9. The total number of mitigators accepted: M = 12.31; Range = 0 to 111.
21 Analytic Plan First, descriptive statistics are examined to determine if victim gender, defen dant gender, victim defendant relationship, or victim illegal activity are associated with sentencing outcomes. Second, the differences in the proportions of death sentence recommendations for the post McKoy subpopulation are examined. The occurrence of ce rtain scenarios in capital trials, such as jury knowledge that the victim was involved in illegal activity at the time of their victimization, indicates the utility in examining the conditional means and probabilities prior to conducting regression analysi s. Third, logistic regression analysis is employed to examine the interactive effects of victim gender and victim related characteristics, controlling for the effects of a set of other independent variables. Logistic regression is utilized for multivariat e analyses when the dependent variable is dichotomous. The dependent variable in the current study is dichotomous (0 = jury recommendation of a life sentence, 1 = death sentence) and the interaction and control variables are categorical and continuous. Mod eling interaction effects in logistic regression is the most appropriate way to examine the hypotheses because we want to know if the effect of a certain independent variable (victim gender) on a binary outcome variable (jury recommendation ) differs based on a third variable or moderator (defendant gender, victim defendant relationship, victim illegal activity involvement). The three interactions used to test the hypotheses will be presented as three different models. If the effect of victim gender varies s ignificantly by defendant gender, the relationship between victim and defendant, or victim involvement in illegal activity then there are interaction effects.
22 The independent variables of interest have nominal (as opposed to interval or ordinal) values th at cannot be rank ordered so dummy variable coding must be utilized to examine the effects of these variables on the outcome (McClendon, 1994). The following series of dummy variables are used to test the hypotheses. For the first hypothesis the male defen dant and female victim combination is used as the reference category because this interaction is predicted to receive the harshest penalty (death) relative to the other combinations (male defendant and male victim, female defendant and female victim, femal e defendant and male victim). To test the second hypothesis the reference category for analyses is female victim and stranger defendant because this category is predicted to receive the harshest punishment compared to all other combinations (female victim and non stranger defendant, male victim and stranger defendant, male victim and non stranger defendant). For the third hypothesis the reference category is female victim and no victim illegal activity because this interaction is predicted to have the harsh est sentencing outcome (death) when compared with other combinations (female victim and illegal activity, male victim and illegal activity, male victim and no illegal activity). The independent variables included in this analysis are represented by dummy v ariables assigning a 1 to the categories of interest. The results of the regression analysis are indicated in the form of odds ratios. The odds ratios will be presented in the results section as the percent odds. The odds ratio indicates the odds of the ev ent occurring compared to chance. An odds ratio of 1 indicates that the likelihood the event will occur is perfect chance. By subtracting 1 from the odds ratio and multiplying by 100 the percent likelihood is calculated. The odds for dichotomous variables (all key independent variables) can be interpreted, for example, as
23 the odds of receiving a death sentence for the category of the independent variable coded 1 compared to the odds for the category coded 0 while holding all other variables constant. Interp retations of the interaction terms will be reported in the results; examples interpreted in the following way: The percent likelihood of receiving the death penalt y for a male perpetrator with a female victim is greater/or less than the odds of getting the death penalty for any other perpetra tor/victim gender combination. The second likelihood of receiving the death penalty for a female victim with a stranger perpetrator is greater/or less than the odds of getting the death penalty for any other gender/relationship ed in the following way: The percent likelihood of receiving the death penalty for female victims who were not involved in illegal activi t y at the time of the incident is greater/or less than the odds of getting the death penalty for any other gender/acti vity involvement combination.
24 Chapter 5 Results An important note must be made concerning the first hypothesis prior to explicating the results. As indicated in Table 1, the number of female defendants is less than thirty ( n =29). The interaction ter m for the first hypothesis (male perpetrators with female victims) is highly correlated with the victim gender variable indicating a collinearity issue. While this small sample size means there is not enough power to test the hypothesis empirically, there are scientific implications of this base rate. The data utilized for this study are a population of capital cases indicating that data on female offenders (and cases that are not male defendant female victim dyads) are not missing but rather are extremely rare in capital sentencing in North Carolina. The interaction between victim and defendant gender may be an extra legal factor in sentencing decisions, but the occurrence of female perpetrated capital offenses is so infrequent that there is not enough data to examine the influence of gender interactions on capital jury decision making. Therefore, the results reported here are in terms of the second and third hypotheses examining the influence of the innocent female victim in jury decisions to recommend deat h. The results of analyses examining the second and third hypothesis are presented in two parts: first, the main effects and interaction effects of the independent variables of interest are examined for the data using difference in proportions analysis; s econd, the
25 hypotheses are examined using logistic regression, controlling for legal factors and demographic characteristics. Difference in Proportions Analysis Hypothesis 2 examines the relationship between victim gender and victim offender relationship i n terms of jury sentencing recommendations (see Tables 2 and 3). D ifference in proportions analysis (Blalock, 1960) indicates that there is a main effect of victim gender such that cases with female victims are significantly more likely to result in death than cases with male victims ( t =4.72, p <.01). We find no main effect, or no statistically significant differences in the proportions of death sentences, for victim offender relationship (stranger compared to non stranger). The main effect of victim gender remains when controlling for victim offender relationship. There are statistically significant differences in the proportions of death sentences recommended for female victims murdered by a stranger or non stranger compared to male victims murdered by a st ranger or non stranger. Cases (regardless of victim offender relationship) with a female victim are significantly more likely to result in death than cases (regardless of victim offender relationship) with a male victim ( t =2.82, p <.01; and t =3.73, p <.01 re spectively). Interaction effects are also found for the second hypothesis indicating that the difference in the proportions of death sentences for female victim stranger defendant cases is significantly different than the proportion of death sentences in male victim non stranger defendant cases ( t =2.93, p <.01). There is also an interaction effect indicated by the significant difference in proportions of the jury recommending death between cases with a female victim non stranger defendant and male victim st ranger defendant ( t =3.42, p <.01). Cases with a female victim stranger defendant are significantly more likely to
26 result in the jury recommending death than cases with a male victim non stranger defendant; and cases with a female victim non stranger defenda nt are significantly more likely to result in the jury recommending death than cases with a male victim stranger defendant. Hypothesis 3 examines the relationship between victim gender and victim involvement in illegal activity (see Tables 2 and 3). D iffe rence in proportions analysis indicates that there are main effects for both victim gender and involvement in illegal activity in terms of jury recommendations of death. Cases with female victims are significantly more likely to result in a recommendation of death than cases with a male victim ( t =4.72, p <.01); and cases where the victim was not involved in illegal activity are more likely to result in a recommendation of death than cases where the victim was involved in illegal activity ( t =2.98, p <.01). The main effect of victim gender remains when involvement in illegal activity is controlled for, such that female victims involved in illegal activity are significantly more likely to receive a jury recommendation of death than cases with male victims involve d in illegal activity ( t =1.70, p <.10). The main effect of victim gender also remains when controlling for the victim not being involved in illegal activity such that cases with a female victim who was not involved in illegal activity is more likely to resu lt in a jury recommendation of death than cases with a male victim who was not involved in illegal activity ( t =3.94, p <.01). The main effect of victim involvement in illegal activity does not remain significant when comparing female victims but there are statistically significant differences in the proportion of cases resulting in jury recommendation of death for male victims. Cases involving male victims who were not involved in illegal activity were
27 significantly more likely to result in death than cases where a male victim was involved in illegal activity ( t =1.95, p <.10). Finally there is an interaction effect indicated between victim gender and victim involvement in illegal activity. Cases with a female victim who was not involved in illegal activity ar e significantly more likely to result in a jury recommendation of death than cases with a male victim who was involved in illegal activity ( t =4.75, p <.0 1 ) Table 2 Difference in Proportions of Death Sentence Recommendations Testing for Main Effects (N=821) Total Stranger Defendant Non Stranger Defendant Difference in Proportions Victim Illegal Activity Victim No Illegal Activity Difference in Proportions Female Victim .582 (n=356) .585 (n=106) .580 (n=250) t=0.09 .500 (n=36) .591 (n=320) t=1.04 Male Victim .417 (n=465) .414 (n=175) .421 (n=290) t=0.15 .339 (n=109) .441 (n=356) t=1.95* Total .477 (n=281) .494 (n=540) .379 (n=145) .512 (n=676) p <.10
28 Table 3 Differences in Proportions Recommended Main and Interaction Effects (N=821) Total Male Victim* Stranger Defendant Male Victim* Non Stranger Defendant Total Male Victim* Victim No Illegal Activity Male Victim* Victim Illegal Activity Female Victim* Stranger Defendant .585 (n=106) t=2.82** t=2.93** Female Victim* Non Stranger Defendant .580 (n=250) t=3.42** t=3.73** Total .414 (n=175) .421 (n=290) Female Victim* Victim No Illegal Activity .591 (n=320) t=3.94** t=4.75** Female Victim* Victim Illegal Activity .500 (n=36) t=0.68 t=1.70* Total .441 (n=356) .339 (n=109) p <.10, ** p <.01 Logistic Regression Analysis As indicated by the above analyses there are significant differences in proportions of death sentence recommendations in the dataset. The relationships between the independent variables and the d ependent variable are further tested by controlling for legal factors, demographic characteristics, and additional variables deemed relevant in capital sentencing literature using logistic regression analysis. The main effects of the independent and contro l variables on jury sentence recommendations are presented in
29 Table 4. As discussed previously the first hypothesis cannot be empirically examined because of the lack of variability in the defendant gender variable; therefore, it is not surprising that the re is not a main effect of defendant gender in the main effects model. However, there is a significant main effect of victim gender such that there is a 41.9% increase in the likelihood of the jury recommending death when the case involves a female victim compared to a male victim (95% CI: [.996, 2.023], p <.10). There also is a main effect of victim involvement in illegal activity and the effect is in the direction hypothesized. There is a 35.8% decrease in the likelihood of the jury recommending death when the victim is involved in illegal activity compared to cases where the victim is not involved in illegal activity (95% CI: [.410, 1.006], p <.10). There is, however, no main effect for the victim offender relationship variable a finding also shown in the difference in proportions analysis. It should be noted that, while not statistically significant, the direction of effect of victim offender relationship is not as hypothesized. The hypothesis was that defendants who are strangers will be sentenced more se verely than defendants who are not strangers because the perceived public threat of stranger perpetrators is greater. The data indicates that there is a 4.5% decrease in the likelihood of the jury recommending death in cases where the victim and defendant are strangers. Implications for this finding will be addressed in the Discussion chapter.
30 Table 4 Logistic Regression Analysi s Testing for Main Effects (N=821) Variable Odds Ratio p CI (lower, upper) % change Female victim 1.419 0.053 (0.996, 2.023) 41.9 Male defendant 0.514 0.136 (0.215, 1.232) 48.6 Victim illegal activity 0.642 0.053 (0.410, 1.006) 35.8 Stranger relationship 0.955 0.815 (0.651, 1.401) 4.5 Non W hite victim 0.712 0.094 (0.478, 1.060) 28.8 Non W hite defendant 0.852 0. 427 (0.573, 1.266) 14.8 Victim age 0.988 0.004 *** (0.979, 0.996) 1.2 Defendant age 1.040 0.000 *** (1.019, 1.062) 4.0 Urban county 0.671 0.019 ** (0.481, 0.937) 32.9 Public defender 3.919 0.003 *** (1.586, 9.686) 291.9 Total victims 1.110 0. 571 (0.774, 1.591) 11.0 Gun used as weapon 0.696 0.060 (0.477, 1.015) 30.4 Prior record 0.891 0.552 (0.610, 1.303) 10.9 Rape aggravator accepted 1.837 0.101 (0.888, 3.800) 83.7 Total aggravators accepted 1.935 0.000 *** (1.618, 2.314) 93. 5 Total mitigators accepted 0.914 0.000 *** (0.894, 0.933) 8.6 Constant 0.167 0.005 p <.10, ** p <.05, *** p <.01 Several of the control variables have statistically significant main effects including victim race, victim age, defend ant age, type of attorney, weapon, county type (urban/rural), rape accepted as an aggravator, total aggravators accepted, and total mitigators accepted. The likelihood of the jury recommending death decreases 28.8% when the victim is non White compared to White victims (95% CI: [.478, 1.060], p <.10). The likelihood of the jury recommending death decreases 1.22 % with every one year increase in victim age (95% CI: [.979, .996], p <.001) and increases 4.00% with every one year increase in defendant age (95% CI: [1.019, 1.062], p <.001). The jury is about 4 times more likely to recommend death in cases where the defendant is represented by a public defender compared to cases where the defendant is represented by a private attorney (95% CI: [1.586, 9.686], p <.01). The likelihood of the jury recommending death
31 are decreased 30.4% when the weapon used is a gun as opposed to an alternative weapon (95% CI: [.477, 1.015], p <.10). The likelihood of the jury recommending death decreases 32.9% for cases in urban counties in comparison to cases in rural counties. The likelihood of the jury recommending death are increased 83.7% when rape is accepted as an aggravating factor (95% CI: [.888, 3.800], p =.10). Finally, the likelihood of the jury recommending death increases 93.5% for every one aggravator accepted (95% CI: [1.618, 2.314], p <.001), and decreases 8.6% for every one mitigator accepted (95% CI: [.894, .933], p <.001). The logistic regression analysis for the second hypothesis examines the relationship between victim gen der and victim offender relationship, and, as shown in Table 5, indicates that the interaction effect is not statistically significant. However, we again find that in cases with female victims and stranger defendants the likelihood of the jury recommending death decreases 38.3% compared to other victim gender and victim offender relationship dyads. This finding indicates that the direction of the interaction between victim gender and victim offender relationship is not in the direction hypothesized (explana tions for this finding will be considered in the Discussion chapter). In this model, the main effects of victim gender and victim involvement in illegal activity remain significant. The likelihood of the jury recommending death increases 64.7% in cases wit h a female victim compared to cases with a male victim (95% CI: [1.083, 2.504], p <.05) controlling for the interaction between victim gender and victim offender relationship. The likelihood of the jury recommending death increases 53.2% when the victim is not involved in illegal activity controlling for the interaction term (95% CI: [.416, 1.024], p <.10). Also, all of the control variables that had significant main effects
32 remain statistically significant when the interaction term female victim stranger def endant is including in the analysis. Table 5 Logistic Regression Analysi s Testing for Interaction Effect of Female Victim and Stranger Defendant (N=821) Variable Odds Ratio p CI (lower, upper) % change FV*SD 0.617 0.187 ( 0 .302, 1.263) 38.3 Female v ictim 1.647 0 .020 ** (1.083, 2.504) 64.7 Male defendant 1.905 0.15 0 ( 0 .218, 1.262) 90.5 Victim illegal activity 1.532 0 .063 ( 0 .416, 1.024) 53.2 Stranger relationship 1.155 0.552 ( 0 .718, 1.856) 15.5 Non W hite victim 0.717 0.102 ( 0 .481, 1.068) 28.3 Non W hite defendant 0.838 0.382 ( 0 .564, 1.246) 16.2 Victim age 0.988 0 .005 *** ( 0 .980, 0 .996) 1.2 Defendant age 1.040 0 .000 *** (1.019, 1.062) 4.0 Urban county 0.667 0 .018 ** ( 0 .478, 0 .932) 33.3 Public defender 3.983 0 .003 *** (1.604, 9.887) 298. 3 Total victims 1.112 0.563 ( 0 .775, 1.595) 11.2 Gun used as weapon 0.700 0 .065 ( 0 .479, 1.022) 30.0 Prior record 0.894 0.563 ( 0 .611, 1.307) 10.6 Rape aggravator accepted 1.939 0 .077 ( 0 .932, 4.037) 93.9 Total aggravators accepted 1.958 0 .000 *** (1.635, 2.345) 95.8 Total mitigators accepted 0.913 0 .000 *** ( 0 .893, 0 .932) 8.7 Constant 0.051 0 .003 p <.10, ** p <.05, *** p <.01 The third hypothesis was intended to examine the interaction effect of cases with a female victim who was not involved in illegal activity. Analysis of the correlation matrix for the third hypothesis indicates that, like the variables in Hypothesis 1, the interaction term is highly correlated with victim gender ( R =.913, p <.01). This indicates a collineari ty issue. There are not enough cases ( n = 36) of female victim involved in illegal activity scenarios t o test the hypothesis empirically using logistic regression. Therefore, conclusions concerning the third hypothesis drawn from the results will focus on th e difference in proportion analysis discussed earlier
33 Chapter 6 Discussion There are several important findings that can be identified from the results of the current study. First, there were not enough cases to empirically examine the first hypothes is because, as discovered, the number of victim and defendant gender dyads with a female defendant is rare. However, because the data used is a sub population of cases this inability to conduct empirical analyses is not a limitation but a finding itself. I n North Carolina there are very few female perpetrated homicides that are tried capitally suggesting that gender of the defendant may also be influential and this influence may present at earlier stages of the criminal justice process. Therefore, empirical ly examining the relationship between victim and defendant gender may benefit from analysis of earlier phases in the criminal justice process or an examination of non capital in addition to capital homicide cases. The remaining hypotheses are first discuss ed in terms of the difference in proportions analysis and then in terms of the logistic regression analysis. Conclusions, limitations, and suggestions for future research follow. The second and third hypotheses were both examined using difference in propo rtions analysis. The findings indicate varying support for these two hypotheses. When examining the difference in the proportions of death recommendations there is a significant main effect of victim gender; however, no main effect for victim offender rela tionship is found. The interaction effects indicate that regardless of victim offender relationship, cases with a female victim are significantly more likely to result in a jury
34 recommendation of death than cases with a male victim. This finding suggests t hat victim gender has a more salient relationship with death sentence recommendations than victim offender relationship; and that the female victim effect is not enhanced when joined with relationship between the victim and offender. In terms of the third hypothesis, examining the relationship between victim conduct and victim gender, there is a significant main effect of victim gender even when controlling for victim involvement in illegal activity. The significant main effect of victim involvement in illegal activity remains when controlling for male victims. The interaction between cases with a female victim not involved in illegal activity and cases with a male victim involved in illegal activity supports the hypothesis. Cas es with female victims who are seen as innocent are more likely to result in a recommendation of death relationship between victim gender and victim conduct while contr olling for other variables in a regression model was not possible, the results of the difference in proportions analysis indicate that victim conduct (especially when joined with victim gender) has an influential effect on jury sentencing recommendations i n capital cases. The findings from the difference in proportions analysis are similar to the findings of the logistic regression analysis. The main effects logistic regression model indicates that victim gender and victim involvement in illegal activity a re significantly related to jury recommendations of death even when controlling for other legally and non legally relevant variables. Cases with female victims and cases where the victim was not involved in illegal activity are more likely to result in a d eath sentence. Victim offender relationship does not have a
35 statistically significant main effect when controlling for other variables. Though no main effect of victim offender relationship was found, logistic regression was utilized to examine the potenti al interaction effect of victim offender relationship with victim gender. As indicated in the results, the interaction term is not found to be statistically significant. Based on the findings, one possible conclusion is that no interactive effect is presen t because at the sentencing phase the legal factors, such as aggravators and mitigators, are more influential than this interactive relationship. Although the interaction term in the logistic regression model for the second hypothesis is not statistically significant, victim gender does remain significantly related to jury decisions to recommend death. There also remains a marginally significant main effect for the victim involvement in illegal activity variable indicating that victim conduct does play som e role in sentencing decisions. Sundby (2003) indicates that jurors recommending life sentences were more likely to engage in discussions about the victim and victim related variables such as conduct ( specifically, involvement in illegal activity). This fi nding appears to be supported through the results of the study reported here. Even after controlling for legally relevant variables and variables commonly controlled for in sentencing studies (e.g., victim and defendant age, attorney type, prior record), w e find that victim gender and victim conduct (in the form of involvement in illegal activity) are significantly related to the jury sentence recommendations. As noted, the relationship between victim offender relationship and sentencing outcome was not in the hypothesized direction, a finding which is deserving of further discussion. Cases where the victim and defendant are strangers, even when interacted with victim gender, are less likely to result in a death sentence than cases where the
36 victim and offe nder are not strangers. While this finding is directionally inverse to the hypothesis, as well as some previous research (Sundby, 2003), there are potential explanations. The direction of the original hypothesis is based on the concept of community protect ion and perceived future dangerousness of the offender, which suggests that jurors are more likely to reserve the harshest penalties for offenders who are strangers to their victims. In capital cases in North Carolina there are only two sentencing options once the defendant has been found guilty, life in prison without parole and the death penalty. Both of these options may be perceived by the jury as providing permanent incapacitation in terms of community protection and future dangerousness. Therefore th e direction of the relationship between stranger homicides and sentencing outcome may not be based on concerns of future dangerousness but on other factors. Jury decisions to recommend death instead of life in prison may be based on other considerations s uch as the harm caused to the primary victim and secondary victims (surviving family members, friends, the community) or qualitative interpretations of the cruelty of the crime. Homicides that occur between individuals who are not strangers, and particular ly familial homicides, may result in more violence than homicides between strangers. As indicated in the regression models, weapon type is significantly related to sentencing outcome such that cases involving a gun are less likely to result in a jury recom mendation of death. This finding indicates that cases where the weapon was a likely to result in a death recommendation. There may be a relationship between victim of fender relationship and weapon such that strangers are more likely to use a gun while non strangers are more likely to use another form of violence which may be more readily
37 available. In other words, stranger homicides may be more utilitarian while homici des occurring between individuals who know each other may be more emotional and thus finding may not be generalizable, but future research should consider that victim offender relationship may operate differently for various crimes and sentencing processes. Examining the interaction between victim gender and other gender related variables at different stages in the criminal justice decision making process may provide a more explicit description of how gender affects capital cases. Several control variables (in addition to weapon type) exhibit significance in both the main effects model and the interaction model and these findings warrant some discussion. Due to their r ole in the capital sentencing process it is not surprising that the total number of aggravators accepted substantially increases the likelihood of receiving the death penalty (since one aggravator must be accepted in order for a defendant to be sentenced t o death) or that the total number of mitigators accepted decreases the likelihood of a jury recommending death. Furthermore, the acceptance of rape as an aggravating factor increasing the likelihood of receiving death is also in the direction expected. As indicated by Williams et al. (2007) sexual victimization of female victims may account for the increased likelihood of the death penalty in those cases. The current rape as an accepted aggravating factor, and rape plays a marginally significant role in jury sentencing decisions. In addition to these legally relevant factors several extra legal control variables, including victim age, defendant age, victim race, and l awyer type, are significantly related to jury sentencing recommendations in North Carolina.
38 The older the defendant and the younger the victim the more likely the case is to result in a death sentence recommendation. This finding is in line with the exist ing effect. Younger victims, particularly children, may be viewed as more innocent by jurors and thus more deserving of the protections of the criminal justice syste m, particularly when their victimizer is an adult. In addition to age, a victim race effect remains in the with a non W hite victim are less likely to result in a jury rec ommendation of death than cases with a W hite victim. Particularly striking is the relationship between type of lawyer and capital sentencing outcome, however this finding is expected to be an artifact of the data. There are only 33 cases with a private att orney out of the total sub population. Of these cases less than one third resulted in the death penalty, however about half of the cases with a public defender resulted in the death penalty. It is concluded that this finding does not necessarily raise con cern about the influence that attorney type (and by extension, defendant socio economic status) has in sentencing decisions because of the rarity of private attorney cases. femal making in post McKoy capital murder trials in North Carolina. As hypothesized, female victims who are not involved in illegal activity may be seen as more innocent or deserving of protection than other victim gender victim conduct dyads. In opposition to expectations, the sub population examined in this study indicates that cases with non stranger defendants are more likely to result in the jury recommending death than cases with stranger defendants. While this study does f ind that
39 victim conduct plays a role in jury sentencing recommendations, the female victim effect is not exponentially increased when joined with a measure of victim conduct. In fact the difference in proportions analysis indicates that victim conduct has a significant effect in male victim cases but has no significant impact in comparisons of female victim cases. Regardless of victim conduct, cases with a female victim are the most likely to result in the death penalty. Therefore, the prominent finding of the current analysis is that there is a consistent victim gender effect in capital cases in North Carolina. t, the concept of innocence or worth should not be abandoned in examinations of gender and sentencing. will not be evident in other data. Future research should consider how variables relating adds to an understanding of the role that victim gende r plays in capital sentencing beyond the focus prior studies have maintained on victim race. The multidimensional influence of the female victim should be further explored in future capital sentencing research.
40 References Ancestry.com. (2000). North Carolina vital records: Deaths 1968 1996 CD ROM. Provo, UT: MyFamily.com, Inc. Baldus, D. C., Woodworth, G., & Pulaski, C. A. (1990). Equal j ustice and the d eath p enalty: A l egal and e mpirical a nalysis Boston, MA: Northeastern University Press. Baldus, D C., Woodworth, G., Zuckerman, D., Weiner, N. A., & Broffitt, B. (1998). Racial d iscrimination and the d eath p enalty in the p ost Furman e ra: An e mpirical and l egal o verview, with r ecent f indings from Philadelphia. Cornell Law Review 83 1638 1770. Baumer E. P., Messner, S. F., & Felson, R. B. (2000). The r ole of v ictim c haracteristics in the d isposition of m urder c ases. Justice Quarterly 17 281 307. Blalock, H. (1960). Social statistics New York: McGraw Hill. Curry, T. R., Lee, G., & Rodriguez, S. F. (2004). Does v ictim g ender i ncrease s entence s everity? Further e xplorations of g ender d ynamics and s entencing o utcomes Crime & Delinquency 50 319 343. Farrell, R. A., & Swigert, V. L. (1986). Adjudication in h omicide: An i nterpretive a nalysis of the e ff ects of d efendant and v ictim s ocial c haracteristics. Journal of Research in Crime and Delinquency 23 349 369.
41 Franklin, C. A., & Fearn, N. E. (2008). Gender, r ace, and f ormal c ourt d ecision m aking o utcomes: C hivalry/ p aternalism, c onflict t heory or g end er c onflict? Journal of Criminal Justice, 36 279 290. Hindson, S., Potter, H., & Radelet, M. L. (2006). Race, g ender, r egion and d eath s entencing in Colorado, 1980 1999. University of Colorado Law Review 549 Holcomb, J. E., Williams, M. R., & Demuth, S. (2004). White f emale v ictims and d eath p enalty d isparity r esearch. Justice Quarterly 21 877 902. Kavanaugh Earl, J., Cochran, J. K., Smith, M. D., Fogel, S. J., & Bjerregaard, B. (2008). Racial Bias and the Death Penalty. In M. J. Lynch, E. B. Patterson, & K. K. Childs (Eds.) Racial d ivide: Racial and e thnic b iases in the c riminal j ustice s ystem (pp. 147 196). Monsey, NY: Criminal Justice Press. Kremling, J., Smith, M. D., Cochran, J. K., Bjerregaard, B., & Fogel, S. J. (2007). The r o le of m itigating f ac tors in c apital s entencing b efore and a fter McKoy v. North Carolina Justice Quarterly 24 357 381. McClendon, M. J. (1994). Multiple r egression and c ausal a nalysis Long Grove, IL: Waveland Press. Paterno ster, R. (1984). Prosecutorial d iscretion in r eque sting the d eath p enalty: A c ase of v ictim b ased r acial d iscrimination. Law & Society Review 18 437 478. Rye, B. J., Greatrix, S. A. & Enright, C. S. (2006). The c ase of the g uilty v ictim: The e ffects of g ender of v ictim and g ender of p erpetrator on a ttr ibutions of b lame and r esponsibility. Sex Roles 54 639 649.
42 Spohn C., & Spears, J. (1996). The e ffect of o ffender and v ictim c haracteristics on s exual a ssault c ase p r ocessing d ecisions. Justice Quarterly 13 649 679. Stauffer, A. R., Smith, M. D., Coch ran, J. K., Fogel, S. J., & Bjerregaard, B. (2006). The i nteraction b etween v ictim r ace and g ender on s entencing o utcomes in c apital m urder t rials: A f urther e xploration. Homicide Studies 10 98 117. Steffensmeier, D., Ulmer, J., & Kramer, J. (1998). The i nteraction of r ace, g ender, and a ge in c riminal s entencing: The p unishment c ost of b eing y oung, Black, and m ale. Criminology 36 763 798. Sundby, S. E. (2003). The c apital j ury and e mpathy: The p roblem of w orthy and u nworthy v ictims. Cornell Law Review 88 343 376. Williams, M. R., Demuth, S., & Holcomb, J E. (2007). Understanding the i nfluence of v ictim g ender in d eath p enalty c ases: The i mportance of v ictim r ace, s ex r elated v ictimizat ion, and j ury d ecision m aking. Criminology 45 865 891. Williams, M. R. & Holcomb, J. E. (2004). The i nteractive e ffects of v ictim r ace and g ender on d eath s entence d isparity f indings. Homicide Studies 8 350 36