a. \v 12.3.,“ . q !‘ iv‘rytlriv... I it . $11.... an. s. {- t2 . 12:! .. .ru... 6:. lal ._ rug-.13.. . . ...$~p:$£...4n.. . :5 an?! azmfln, floyhjy‘ h). v. a .. . ”mg: #uumwmnf ..§.,: . ..... :3 “aw . 323V“... . 35$?! , 33; ’ {.1} cos}: _ pang. ammmmh ‘ .. u. .‘hanu . . . , ,. . . u. flu-3“: TI 33. 39)."... A. It... zmnunrz 1.! A». , 3.1.7.?! .flfiJaf; ( . 1. . 35...! ...fl.....v;:..p3.ev=u. :1. . - .‘i. ..clt.§x«n...y 3.. . I . 9.2L! » {IsarIQ’pa 2.... . .341. $9.40“ ‘thgu ‘l\el. .0 .. v\.lvthOIr I» n 2 .l‘dv’fi‘n I! inai .Is 2 EDT! . .t I. .3. .u:no ‘fillfiv! v4.1. 3111‘: ‘ , .(ilg U3 1‘ v. V. .l‘ hl. hvo’ I... la: ld‘sbnohuldli‘ tan”...- .2.( 31»!B.Dn‘l’. {It‘ll Buli."‘1“‘fz .n'l'v WW 4 CC LIBRARY Michigan State University PLACE IN RETURN BOX to remove this checkout from your record. TO AVOID FINES return on or before date due. MAY BE RECALLED with earlier due date if requested. DATE DUE DATE DUE DATE DUE (330? 07.1 FEBOlZW . moo Cfllflmm.“ GENDER AND JUSTICE IN SENTENCING DECISIONS: AN ANALYSIS OF THE IMPACT OF TRADITIONAL GENDER EXPECTATIONS ON SENTENCING OUTCOMES FOR FELONY FEMALE OFFENDERS IN THE STATE OF MINNESOTA BY Barbara Ann Koons A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY School of Criminal Justice 2000 by v... ‘v us .~ ~ g 9-. V. L» .\~ ABSTRACT GENDER AND JUSTICE IN SENTENCING DECISIONS: AN ANALYSIS OF THE IMPACT OF TRADITIONAL GENDER EXPECTATIONS ON SENTENCING OUTCOMES FOR FELONY FEMALE OFFENDERS IN THE STATE OF MINNESOTA BY Barbara Ann Koons The research was designed to enhance understanding of the impact of sentencing reforms on women offenders. The key hypothesis is that prior to the implementation of sentencing guidelines in Minnesota, women who fulfill traditional gender roles are more likely than men are to receive alternatives to incarceration. After sentencing guidelines, no significant differences in the likelihood of receiving a prison sentence are expected between men and women. A related hypothesis is that sentencing reforms and the “war on drugs” are disproportionately related to women’s higher odds of going to prison. The analysis examines three time periods and compares the sentencing outcomes for men and women for drug and property offenses. The data come from the Minnesota Sentencing Guidelines Commission and Ramsey and Hennepin Counties pre-sentence investigation reports for 4,076 ,. ur- IA) " -_ II, I. convicted offenders. Several theories and explanations such as social construction feminist theory, multiracial feminist theory, and explanations based on chivalry suggested several independent variables including gender, race, dependent children, and offense type. These and legal independent variables were examined in relation to two dependent variables, incarceration and sentence length. The results suggest mixed support for feminist explanations of sentencing disparity. For the first sentence decision, gender was influential in decisions made both before and after sentencing guidelines. Women were more likely than men to receive an alternative to incarceration. Additionally, tests of interactions between gender and race, dependent children, and offense type showed only one significant interaction. At Time 2, white women were more likely to be incarcerated than nonwhite women. For the sentence length decision, gender was influential only at Time 2 when women received significantly shorter prison sentences than men, controlling for other predictors of sentence length. Finally, the findings suggest that sentencing reforms and the “war on drugs” in the state have not disproportionately affected women in a negative way. Copyright by BARBARA ANN KOONS 2 0 0 0 Dedicated in loving memory of my grandmothers, Anna Reichman Koons & Elizabeth Umstead Walton .Ai-A‘ -.‘-v1 - ' u..a.a. u i . “F '- ‘vo .- . Inn... - _ " v- ‘ an- . _ ...-,_ -- _ _ , ~- ' h... u '- ...- .- ~V : ...~ ‘ . ~‘~~- “. fi-‘- . . 2v- .1.“ . 6“ h .73:- I (I! I ACKNOWLEDGEMENTS This remarkable accomplishment would not have been possible without the encouragement and support of my mentors, friends, and family. I would like to first extend a special note of thanks to Dr. Merry Morash, the chair of my committee, for her unwavering support of my work and professional development. I hope that my future endeavors and successes are a reflection of her faith and support in my skills and talent. In addition, I would like to acknowledge the assistance of Dr. Christina DeJong for her advice and guidance in helping me to complete my dissertation. To my final two committee members, Dr. Marilyn Aronoff and Dr. Cris Sullivan, I would like to thank the both of them for their unique insights and contributions. I would also like to thank Anne Wall at the Minnesota Sentencing Guidelines Commission for her help and assistance with the data. Of course I cannot forget to acknowledge my wonderful friends from the graduate program at Michigan State University who I deeply respect. John Burrow, Pam Schram, Tracy O'Connell, Sean‘Varano, Gwen Bramlet- Hecker, Kevin Gray, Don Hummer, and Vic Bumphus-- a BIG vi ' A nu _v- -..-r an. - ‘ . ~o< a .4 u 9,. n r M uu- - . ‘n-Y- ~0-‘. _ Iva. h 0....“ u... I q \ ~ - I..‘. thank you to each of them. I am proud to be their friend, colleague, and fellow MSU alumnus. I would also like to thank my new friends and colleagues at the University of South Carolina, especially Angela Gover and John MacDonald, for helping me to see “the light at the end of the tunnel.” To my parents, brothers Bob and Mark, and sister Sharon, I extend my deep appreciation for all of the love and support they have shown me over the years. I greatly respect and appreciate the fact that my parents instilled in me at an early age the importance of being an independent thinker, being committed to goals, and having determination. Finally, to Ron Witt who came into my life during the writing of this dissertation, I am so fortunate to share this wonderful accomplishment with him. Words cannot express what his love, encouragement, strength, and arm twisting has meant to me. vii o .- --~ -q V- v.... 'I- .0. - .- ‘ U - o .— O - ~ ~ — u o .- H TABLE OF CONTENTS LIST OF TABLES xi CHAPTER 1 INTRODUCTION 1 Statement of the Problem 1 The Debate Over Equal vs. Different Treatment of Women Offenders 5 Crime, Arrest, and Incarceration Patterns of Women Offenders 12 The Female Drug Offender 18 Sentencing 94 Background 94 Determinate Sentencing: Getting Tough on Crime 95 Research Study 49 Organization 90 CHAPTER 2 REVIEW OF THE LITERATURE S2 Gender and Sentencing 83 Legal Factors S8 Extra-Legal Factors 62 Drug Offenses 82 Studies on Gender and Sentencing Guidelines 87 Criticisms of the Gender and Sentencing Research 96 Summary 100 Research Questions 103 CHAPTER 3 METHODOLOGY 107 Hypotheses 107 The Study Site 108 The Data 110 Sample 113 Specification of Variables of Interest 118 Data Analysis 123 Sample Selection Bias 132 Conclusion 136 \riii 0». - ».~1 ~- CHAPTER 4 ANALYSIS OF DATE 137 Sample Characteristics 137 Full Sample 137 Time Periods 143 Offense Type 152 Conclusion 156 Bivariate Correlations Between Variables of Interest 158 Conclusion 159 Analysis of Sentence Outcome: In vs. Out 165 Full Model 165 Time 1 Analysis 171 Time 2 Analysis 176 Time 3 Analysis 184 Analysis for Sentence Length Decisions 191 Sentence Length Decisions at Time 1 191 Sentence Length Decisions at Time 2 700 Sentence Length Decisions at Time 3 709 Analysis of Predicted Probabilities of Receiving a Prison Sentence 918 Time 1 Analysis 918 Time 2 Analysis 919 Time 3 Analysis 921 Conclusion 921 Analysis of Equivalence of Regression Coefficients 923 Conclusion 926 CHAPTER 5 CONCLUSIONS 928 Summary and Discussion 928 Decision to Incarcerate 928 Sentence Length Decision 931 Predicted Probabilities of Receiving a Prison Sentence 732 Sentencing Reforms and the “War on Women” 933 Limitations of Present Research 935 Implications for Theory 939 Chivalry 939 Implications for Policy 945 Minnesota Sentencing Guidelines 945 Justice for Women APPENDIX A MINNESOTA SENTENCING GUIDELINES GRID 747 951 APPENDIX B OFFENSE SEVERITY LEVEL 754 APPENDIX C CALCULATION OF CRIMINAL HISTORY INDEX SCORE 756 APPENDIX D FACTORS TO BE EXCLUDED IN MAKING DEPARTURE DECISIONS APPENDIX E FACTORS TO BE INCLUDED IN MAKING DEPARTURE DECISIONS 759 761 APPENDIX F HISTORY OF MINNESOTA’S CONTROLLED SUBSTANCES LAW 766 APPENDIX G OFFENSE SEVERITY REFERENCE TABLE 975 REFERENCES 978 Table Table Table Table Table Table Table Table Table Table Table Table Table Table Table 10: 11: 12: 13: 14: 15: LI ST OF TABLES Total Arrests for Women, 1998 16 Percent Change in Felony Convictions of Women in State Courts, 1990-1996 Felony Convictions of Women in State Courts, 1990-1996 Independent and Dependent Variables .............. Expected Relationships Between Independent and Dependent Variables ............ Descriptive Information for the Overall Sample and Bach Time Period Descriptive Information for Men and Women, Overall Sample Descriptive Information for Each Time Period by Gender Descriptive Information by Offense TYPe Descriptive Information by Offense Type and Gender Bivariate Correlations for Variables in Full Model and Time 3 Model Bivariate Correlations for Variables in Time 1 and Time 2 Models Logistic Regression Results for Full Model and Gender Models Logistic Regression Results for Time 1 Model and Gender Models Logistic Regression Results for Time 1 Model with Interaction Effects xi 72 73 .m121 ”122 140 142 150 153 157 161 163 167 173 175 Table Table Table Table Table Table Table Table Table Table Table Table Table Table Table 16: 17: 20: 21: 22: 23: 24: 25: 26: 27: 28: 29: 30: Logistic Regression Results for Time 2 Model and Gender Models Logistic Regression Results for Time 2 Model with Interaction Effects Predicted Probability of Incarceration for Gender and Race Interactions, Time 2 ..... Logistic Regression Results for Time 3 Model and Gender Models Logistic Regression Results for Incarceration Decisions, Final Models ........... Descriptive Information for Time 1, Sentence Length Decision Test for Multicollinearity, Initial and Corrected Models, Time 1 Linear Regression Results for Sentence Length Decision, Time 1 Model Descriptive Information for Time 2, Sentence Length Decision Test for Multicollinearity, Initial and Corrected Models, Time 2 Linear Regression Results for Sentence Length Decision, Time 2 Model Descriptive Information for Time 3, Sentence Length Decision Test for Multicollinearity, Initial and Corrected Models, Time 3 Linear Regression Results for Sentence Length Decision, Time 3 Model Predicted Probabilities for Receiving a Prison Sentence by Time Period xii 179 182 183 187 189 193 197 199 903 705 908 712 715 917 720 Table 31: Table 32: Equivalence for Men and Within Time Equivalence for Men and Periods of Regression Coefficients Women Drug Offenders Periods of Regression Coefficients Women Across Time xiii 924 925 CHAPTER 1: INTRODUCTION Chapter one provides an introduction and discussion of the female offender population and reforms in the sentencing area over the last three decades. First, the statement of the problem highlights the need for and importance of studying the impact of gender on sentencing decisions. Second, the debate over equal versus different treatment for women offenders is considered. This section also discusses the criminal arrest, conviction, and sentencing patterns of women and a brief description of women drug offenders. Next, an overview of the sentencing area, including background information on recent changes from indeterminate sentencing systems to determinate sentencing systems is presented. In addition, Minnesota’s sentencing guidelines system is described and a discussion of recent state drug legislation is provided. Finally, the research study is presented. Statement of the Problem The decision to examine gender and sentencing decisions was based in part on the belief that an important reciprocal relationship between gender and decision-making in the criminal justice system exists. Gender impacts the decisions of officials working in the system and the decisions made by officials shape perceptions of gender and .- or. a 4 n~ . a construct its meaning within the legal arena. Amid the calls for increasing penalties and neutrality in sentencing, women have been neglected in the process. Little attention to and understanding of women's offending patterns and the circumstances surrounding their crimes, or how they become involved in the criminal justice system in the first place, has characterized the sentencing reform movement. Under current determinate-based sentencing systems (e.g., sentencing guidelines) there is no regard for the realities or experiences of the women, and for that matter, difference among women. After all, women and men do not start on equal footing in society, yet sentencing changes and reforms assume that they do. The philosophical underpinning of recent sentencing reforms presupposes that “like-situated offenders” exist and can be neatly categorized so that proportionality of punishment can be achieved (Tonry, 1995). Because women have received more leniency under indeterminate sentencing schemes, one could make the argument that women are disproportionately impacted by attempts to remove all consideration for non-legal factors and rely on narrowly defined legally prescribed one. Prior to the introduction of determinate sentencing, under an indeterminate system, the courts enjoyed a broad level of discretion in making sentencing decisions. Critics have maintained that discretion resulted in discrimination and disparate treatment. Since the 19703 a significant amount of research has explored the issues of disparities in sentencing decisions due to gender, race, and class (Armstrong, 1977; Kruttschnitt, 1980; Mann, 1984; Spohn, Welch, & Gruhl, 1985; Steffensmeier, Kramer, & Streifel, 1993; Zingraff & Thomson, 1984). While findings have been mixed at the sentencing stage with regard to gender effects, early research generally found evidence that women were treated more leniently when compared to men. Many have attributed this to paternalistic or chivalrous views of judges and other actors working within the courts. Research conducted more recently suggests that the relationship between gender and sentencing decisions is more complex. More recent studies find that not all women are treated in a lenient manner, but that sentencing is influenced by other factors related to the defendant’s gender. Changes in sentencing policies around the United States over the last three decades also have raised questions about whether women continue to be treated more leniently then men. Toughened legislation in the form of “three strikes,” mandatory minimums, sentencing guidelines and truth in sentencing have intensified the punitive response to crime. While proponents of these sentencing changes intended to address crimes of the predominantly male, violent criminals, critics contend that the changes in sentencing laws have unduly punished women offenders. For example, feminists such as Chesney-Lind (1997) have warned that the toughened D. :o ._. response to drug violations in particular has impacted the lives of women offenders more significantly than the lives of men and as a result she and others (Feinman, 1994) have labeled the “war on drugs” a “war on women.” The “war on drugs” campaign has had a significant impact on the entire criminal justice system, but particularly the courts and corrections areas. Arrest rates over the last two decades suggest that the “war on drugs” strategy has been a successful one in identifying illicit drug users. Arrests for drug law violations between 1984 and 1998 increased 168% (580,900 to 1,559,100 total estimated arrests for drug law violations)(FBI, 1998). The courts have carried the burden of processing the greater number of drug offenders. In 1990, 33% of all felony convictions in state courts were for drug-related crimes. In an effort to handle the volume of drug cases more efficiently, Florida in 1989 established the first drug court. Today there are more than 200 drug courts around the country and many more in the planning stages (National Association of Drug Court Professionals, 1997). These statistics suggest that a significant number of people entering the system, who end up being convicted and sentenced to some type of punitive sanction, have a drug dependency problem. Therefore it becomes important to understand the impact that the new sentencing legislation has had on women who are sentenced for drug-related offenses. Are women being sentenced any differently from men as researchers have demonstrated in the past, or have changes in sentencing laws reduced this disparity based on gender? Few prior studies have considered this question. Despite the limited research on gender and sentencing, many suggest that changes in drug enforcement and sentencing have been the catalyst behind the large numbers of women entering the corrections system (Chesney—Lind, 1995; Feinman, 1994). They argue that women who would have previously been sentenced to probation are now being sent to correctional institutions instead. Thus, any chivalrous or lenient treatment that women may have enjoyed at the sentencing phase of the criminal justice system, it is argued, no longer exists. The increasing numbers of women being imprisoned in this country, along with the belief that women offenders have different and unique needs and thus require special services, contributes to the debate over whether women should be handled in the justice system as equal to men or as different. The Debate Over Equal vs. Different Treatment of Women Offenders There is considerable debate about whether women benefit more from being treated as equal to men or with recognition of special gender-linked needs and circumstances than men. Over time, women involved in the criminal justice system have endured a mix of equal and differential A'd' .uv n 3“. ’l treatment1 or handling.‘2 Given the complex history, several opposing positions currently exist for explaining the handling of women who are processed within the criminal justice setting (Belknap, 1996). Three major explanations of how gender affects crime processing have been set forth and debated: 0 Equal treatment - women and men are sanctioned equally and equivalently by the courts 0 Chivalry — women are sanctioned more leniently compared to men by the courts 0 Evil woman - Women are sanctioned more severely compared to men by the courts According to the first position, the equal treatment perspective, men and women are treated similarly, and any occurrence of differential decision making is not related to their gender per se. For example, research has indicated that men receive harsher sentences than women, but this is due in large part to the fact that men have more extensive criminal histories and/or commit more serious offenses than women. Thus, the type of sentence received relates to the severity of the current offense, criminal history and other legally relevant factors. Many feminist legal scholars have argued that the only 1 Treatment in this dissertation refers to the handling or response to women by criminal justice officials, including judges, etc. 2 See the work of Chesney-Lind and Pollock (1995) for a discussion about the use of both differential and equal strategies that have been used at different points in history to shape the response to women under correctional supervision. IN I" 'V H) II! way to rid the system of gender discrimination is to advocate for equality under the law and equal treatment by criminal justice officials. This is the only way to protect women from discrimination (Chesney-Lind & Pollock, 1995). [T]he equalization proponents feel that given legal and social realities, differential treatment for women will always mean unequal treatment; by accepting different definitions and treatments, women run the risk of perpetuating the stereotype of women as “different from” and “less than” male (Chesney-Lind & Pollock, 1995: 156). Those who are critical of this view argue that women are different from men and when assertions are made that they are equal to men, women will lose every time since the standard by which equal is measured is a male standard. Policies under the law are often developed and administered specifically to deal with the criminality of men involved in the legal system, without attention to their appropriateness for women. The remaining two positions, the chivalry explanation and the “evil woman” explanation, both focus on differential treatment. On the one hand, the chivalry position maintains that women are treated more leniently by the criminal justice system (Belknap, 1996). Additionally, it argues that woman who come into contact with the criminal justice system are treated in a protective manner, resulting in a selective application of the law (Armstrong, 1977; Bickle & Peterson, 1991). On the other hand, the “evil woman” position maintains that women are treated more severely than their male counterparts at the sentencing stage. Chivalrous attitudes and behaviors toward women have a long history. Chivalry, the notion that women are in need of protection, first surfaced during the middle ages in Europe (Moulds, 1980).3 The worship of women was a key principle of the chivalry period, a time that emphasized duty of noble service, courage and obedience of rule (Cornish, 1980: 27-28). While this type of service disappeared over time, some remnants of this regard for women continue to be evident in our social world even today. According to Moulds (1980) chivalry is revealed in contemporary society by the way in which appropriate behavior is defined along gender lines and the relationship between men and women. For example, women are expected to 3 Chivalry involved sentiment, practices, laws and customs that abounded among dominant classes in Europe between the 11th and 16th centuries. It can be defined as according to Cornish (1908: 13), The moral and social law and custom of the noble and the gentle class in Western Europe during the Middle Ages, and the results of that law and custom in action. It applies, strictly speaking, to gentleman only. Its three principal factors are war, religion, and love of ladies: its merits and faults spring from those three heads, and all the side influences which attend its growth and decay may be summed up under these. Thus the whole obligation of the man was connected to the notion of chivalry. Among other things, it involved special treatment that was extended to women by Knights who had sworn to protect women because of s ‘. act in feminine, docile, and subordinates ways. These expectations, it is believed, influence women's criminality and the system’s response to the crimes committed by women offenders. The resulting tendency towards lenient treatment of women offenders may be changing under today’s criminal justice system due to more punitive and determinate-based sentencing. The contemporary literature on gender disparity and sentencing has at times created confusion over the meaning and consequences of chivalry. In the literature, the term paternalism is often used interchangeably with chivalry in order to explain the differential treatment of women offenders (Crew, 1991). The two are closely tied together; however, Moulds (1980) maintains it is important to make the distinction between the two. Because women are viewed as feeble or the weaker gender, the differential treatment resulting from chivalrous attitudes represents an “accompanying power relationship of male domination,” also referred to as paternalism (Moulds, 1980: 280). Paternalism involves “a type of behavior by a superior toward an inferior resembling that of a male parent to his child” (Moulds, 1980: 280). Thus, paternalism by the courts results in lenient treatment for women because judges and their perceived weaknesses (Moulds, 1980). A. ~-‘ if! .- .V v. a» cu nu ’v an, is. ‘. -4- ‘— v... x v other court officials take a fatherly approach in their treatment of women offenders as children, in need of guidance and protection (Parisi, 1982). According to the “evil woman” explanation, women are treated more severely than men for selected similar offenses (Belknap,1996: 70), and they receive harsher sentencing decisions because they violate gender roles and common stereotypes of women. As long as women who come into conflict with the law perform according to their socially accepted gender roles, for instance roles of mother or wife (i.e., family roles), then the criminal justice system has treated them in a lenient manner consistent with the chivalry explanation. This lenient treatment has preserved the family institution and maintained intact bonds between mother and child(ren). However, if women violate their gender roles by committing offenses that are traditionally committed by men (e.g., violent crimes, drug offenses), then the criminal justice system treats them more severely.4 Proponents of both the chivalry and “evil woman” explanations argue that differences in decision outcomes are due to the defendant's gender and are based on cultural images or stereotypes of women. It has been suggested that they are complementary rather than competing explanations ‘ A full discussion will be provided in the next section. 10 .- 0"- ;.-v ...-- hul- -...A -.-- y’- .. Ap- ‘ u» D’- - u..- - s" -5- ~ on- u .4 p- ...‘. v» - v Q. ‘ g _V ‘- ‘ _- ‘ i s_ b- (Crew, 1991; Nagel & Hagan, 1982). In the case of chivalry the woman satisfies her gender role expectations, and in the case of the “evil woman” explanation, the woman violates her gender role expectations in some way or another. For example, several researchers addressed the issue of offense type in relation to gender expectations and found that in some cases sentencing outcome was related to whether or not the offense was a “traditional” feminine crime (Daly, 1987; Spohn & Spears, 1997; Steffensmeier, Kramer, & Streifel, 1993). Feminists advocating for a differential approach to handling women under the law recommend a stance based on the notion “separate but equal.” This means that women and men might receive different treatment or handling, but “women are not placed in a more negative position” (Chesney-Lind & Pollock, 1995: 156). Finally, there is concern that both the equal treatment and different but equal positions are problematic. Both approaches, critics argue, rely on male definitions to determine how women should be treated. In other words, equality translates into rights equal to those of males and differential needs translate into needs different from those of males}5 Males always represent the reference group (Chesney-Lind & Pollock, 1995: 156). 5 Italicized in the original source. 11 .pn v” . . .u u 3. a. ... .a . E. .2 .4 ~ :- .. .u. A. .. 3. v. .'H . .. ”I v. a. .. ~«. A“— v Given significant reforms in sentencing and a “get tough” sentiment in the United States, scholars are left to explain the effects of these changes on the sentencing of women offenders. Have these reforms disproportionately resulted in severe sentences for women in the legal system or do some women continue to receive lenient treatment? This research was designed to examine these and related questions. Some people have argued that the increased number of women in prison is due to changes in the seriousness of women's criminality. Specifically, they feel that women offenders are increasingly violent and that they therefore warrant a more punitive response. Before describing sentencing reforms in more detail, this chapter will consider the pattern and nature of women’s criminality. r'm Arre an Incarceration Pat rns of Women Offenders This section addresses whether or not the increasing female incarcerated population is the result of an increase in the severity of women offenders’ crimes. Some scholars maintain we are not necessarily seeing a new, more violent female offender, and that offending patterns have not 12 np-' ,...A. t'-~- .A vv ...A .u‘ Alla vvo I" I" H' changed (Chesney-Lind, 1997; Chesney-Lind & Pollock, 1995; Nagel & Hagan, 1992; Simon & Landis, 1991; Steffensmeier, 1995). They further argue that the legal system’s response to women has changed so women are more likely to be incarcerated than ever before. Several questions are considered: (1) Have the offending patterns of women changed significantly over the last three decades, and have the offenses committed by women become more serious? (2) Can the offending patterns of women over the last three decades explain the increases in offender populations in U.S. women's prisons? The typical adult female offender is young (oftentimes under the age of 30), a single mother, poor and of color, and lacking education and job skills (Belknap, 1996). Compared to men, she tends to commit less serious offenses that are economic rather than violent in nature. Total arrests rose between 1989 and 1998 by 7 percent with arrests of males increasing 2 percent and arrests of females increasing 28 percent (FBI, 1998). During the 19905 we saw the impact of the war on drugs for women. Over this period, arrests for drug related offenses increased approximately 26% for females (18% for males) (FBI, 1998). Trends in felony convictions of women and men during the first half of the 19903 indicate that convictions for drug-related 13 0» ar‘ CV Q.‘ Dr.‘ I.- ‘Ao . >..,. --~-. .~. A.- ‘9 I... »~_ *1- II! I II A: v‘ I" offenses increased at a higher rate for women when compared to men (37% vs. 25%)(Greenfeld & Snell, 1999). While the arrest of women for violent crimes increased over the decade by 53%, the share of arrests for violent crimes by women still remains quite low (almost 4% of all arrests for women in 1998). As Table 1 shows, women in 1998 continued to be arrested primarily for non—violent offenses such as larceny- theft (i.e., shoplifting)(14.7% of all arrests for women), forgery and counterfeiting, fraud, and embezzlement (total of 6.9%), driving while under the influence (6.5%), disorderly conduct (5%) and drug abuse violations (9%)(FBI, 1998). As the data from the Uniform Crime Reports show, women continue to commit traditional “feminine” offenses. These crimes are those that are committed most often by women. This refers to crimes represented by prostitution, running away, larceny/theft, fraud, and forgery/ counterfeiting (Belknap, 1996). According to Belknap (1996) minor property offenses have been attributed to women more so then men.6 While men do commit the majority of larceny and thefts (e.g., forgery, fraud, and counterfeiting), the proportion of women’s arrests and convictions for these offenses are 6Men commit and are convicted of property offenses more often than women, however these offenses represent a high proportion of arrests and convictions of all offenses for women. 14 tn ' 1 I!) (I! (II “nu- J... (I) 6‘. ’r-. -c.‘ 'l V II- a-.. g‘.‘ :VV‘ “5. I!" u-‘- (I) I (I) r ID sizeable. Chesney-Lind’s review of trends in women’s crime indicates that a significant proportion of larceny thefts are shoplifting. Although men also shoplift it is argued by some that shoplifting is a “prototypical” female offense. Non-traditional feminine offenses include violent offenses such as homicides and assaults. Consistent with the view that women offenders have not become markedly more violent, Steffensmeier (1995) found in his analysis of UCR data over a 30-year period (1960—1990) that while the arrest rate of women increased, female crime patterns (i.e., types of crimes) remained stable over time and women were not becoming more violent. Instead, recent arrest data show women continue to be arrested for minor property crimes (e.g., larceny, fraud, and forgery) and other petty offenses (Steffensmeier, 1995: 89; see also Simon & Landis, 1991). As for drug law violations, Steffensmeier’s (1995: 91) analysis revealed that women were arrested at higher rates across the time period, from 8 of 100,000 in 1960 to 166 of 100,000 in 1990; however, the female share for drug law violations of all arrests remained for the most part stable. 15 Table 1: Total Arrests for Women, 1998 Offense 1998 Percent of Total Arrests, 1998 TOTAL’ 1,950,808 Violent Crime 71,757 3.7% Murder and nonneg. Manslaughter 1,259 b Forcible rape 233 b Robbery 7,874 .4% Aggravated Assault 62,391 3.2% Property Crime 329,132 16.9% Burglary 25,525 1.3% Larceny-theft 287,040 14.7% Motor vehicle theft 14,992 1.0% Arson 1,575 .1% Other assaults 185,178 9.5% Forgery and counterfeiting 27,626 1.4% Fraud 101,194 5.2% Embezzlement 5,122 .3% Stolen property 13,194 .7% Vandalism 28,059 1.4% Weapons; carrying, possessing, etc. 9,443 .5% Prostitution and commercialized vice 36,653 1.9% Sex offensesB 4,964 .3% Drug abuse violations 173,267 8.9% Gambling 652 b Offenses against the family/children 16,465 .8% Driving under the influence 126,781 6.5% Liquor laws 81,762 4.2% Drunkenness 58,067 3.0% Disorderly conduct 98,092 5.0% Vagrancy 4,173 .2% All other offenses (except traffic) 484,874 24.9% Suspicion 605 b Curfew and loitering law violations 37,586 1.9% Runaways 56,767 2.9% ** “b” denotes a percentage less than Source: 1998 Uniform Crime Reports 1998). 7 Does not include suspicion. ' Except for forcible rape and prostitution. 16 a: u. o .3 z. 2‘ 4‘ in Since the 1980s, the number of women incarcerated in the United States has nearly tripled (Chesney-Lind, 1995) and many jurisdictions spent the last decade building facilities to accommodate the larger numbers. Chesney-Lind (1995) contends that much of the increase in the number of imprisoned women can be connected to a change in the system’s response to them, particularly by the courts. While arrest rates for women increased 29% between 1986 and 1990, incarceration figures increased for women by 73% in jails and 77% in prisons during the same time period. Chesney-Lind argues that more women being arrested cannot account for the influx of women into this nation’s jails and prisons. In other words, it is not the behavior of the women which has necessarily changed, but instead, the official response to the women and their activities that has led to the larger number of women entering correctional systems (Krohn, Curry, & Nelson-Kilger 1983; Pollock 1995). Over the last decade Bureau of Justice Statistics (BJS) data show that the number of women incarcerated in this nation’s prisons and jails is increasing at a faster rate than the rate for men (Women in Criminal Justice: A Twenty Year Update Special Report, 1998). From 1985-1995 the number of incarcerated men approximately doubled (691,800 to 1,437,600) and the number of incarcerated women approximately tripled (40,500 to 113,100) over that same time period (Women in Criminal Justice: A Twenty Year Update Special Report, 1998). Chesney-Lind (1997, 1992) attributes 17 no A. Ii vi. I .9? V- "v— I” the system's response to drug law violations as a major catalyst behind more women going to prison than ever before. A comparison between incarcerated women in 1975 and 1995 found that women were primarily incarcerated for larceny, forgery, embezzlement, and prostitution in 1975 whereas they were incarcerated primarily for drug-related offenses and larceny in 1995 (Women in Criminal Justice: A Twenty Year Special Report, 1998). The fact that increased violent criminality does not explain the influx of women into prisons underscores the importance of research to understand the decision-making process that results in the increased rate of incarceration. Also, information on the lives of women involved with the justice system is needed to understand the impact of the decision-making process that the research will examine. Despite the trend that women have not become more violent in their offending patterns, women have been affected by the drug problem in this country. The next section specifically addresses the female drug offender by discussing how women come to use drugs and their involvement in drug—related offenses. The Female Drug Offender The problem of drugs for women offenders has been well documented in the literature (Bush-Baskette, 1998; Inciardi, 18 Lockwood, & Pottieger, 1993; Mahan, 1996; Maher & Curtis, 1995; Pettiway, 1997; Richie, 1996; Sterk, 1999). Several authors examining the issue of gender, drugs, and violence describe the women in their research as being “trapped.” The pathway to the legal system for the typical female drug offender is a sad and tragic one. In addition to her addiction problems, she has typically grown up in a family surrounded by chaos and conflict, experienced physical and sexual abuse, and even sought refuge on the streets turning to prostitution and drug use and other petty forms of crime. To understand why women use drugs and become involved in the criminal justice system requires one to look at the wider social context. According to Maher and Curtis (1996): m[P]revious research [for women] indicates that drug use and violence covary but are not related in a deterministic way. Rather, this relationship is mediated by the complexities of user's social and economic worlds. (p. 152) Drug use, consequently, is a symptom of other underlying problems and concerns in their lives. Women often take drugs as a way of coping with experiences of prior abuse. Mahan's (1996) work involving women and crack cocaine highlights the interconnections between prostitution for drugs, violence, relationships with men, and the environmental context. She found that women working as 19 prostitutes in crack houses lived a continuous cycle of performing sexual services in order to obtain and then use drugs, and then the cycle would repeat itself. While working in the crack houses, women were vulnerable to violence and abuse, as well as contracting HIV/AIDS and other sexually transmitted diseases (Mahan, 1996). Additionally, their relationships with men were dysfunctional and characterized in Mahan's work as relationships based on dependency on and domination by men. Women’s roles in the illegal drug market mirror their position within wider society. Women rarely hold positions of status within the hierarchy and remain on the fringes as drug whores, drug mules, or lookouts. Maher and Curtis (1995) found that many of the women they interviewed gained entry into the drug culture through street-level sex work. Sterk’s “Queens of the Scene” are notable exceptions to this general trend. Converting powder cocaine into crack cocaine is a complicated process (Sterk, 1999). Women who excelled at cooking the rocks of crack cocaine (i.e., the “Queens”) were accorded a certain level of prestige within the drug culture not generally given to women. For the most part, women’s involvement in the drug culture has not been a liberating experience. Instead, their involvement has served to reinforce and remind them of 20 the oppression and suffering they have endured throughout their lives. Women are “trapped” within these realities, with little hope for gaining their own voice and control over their lives. Contributing to this sad picture is the criminal justice’s systems inability to understand and be sensitive to the experiences of women. In constructing sentencing policy, the system seeks to punish and make examples of them. In the end, this is little help to those who remain in the drug culture, trading sex for their habit and working on the fringes of the drug market because of no foreseeable way out. Gender and Drug Conviction Statistics Since the institution of the “war on drugs” policies of the last decade, the percent change in felony convictions for drug offenses were higher for women when compared to men (See Table 2). Overall, drug offenses increased by 27% between 1990 and 1996. However for women, the number of drug felony convictions increased by 37% as compared to 25% for men. 21 Table 2: Percent Change in Felony Convictions of Women and Men, 1990-19969 Felonies Total Females Males All 20% 42% 17% Violent 14% 25% 12% Property 6% 44% -2% Drug 27% 37% 25% Other 46% 65% 44% Table 3 provides more detailed information concerning felony drug convictions for women in the early 1990s after significant drug legislation occurred around the country in the late 19808. Felony convictions for women involving drug trafficking increased 34% from 1990 (24,562 convictions) to 1996 (33,005 convictions). Felony convictions for drug possessions increased 41% between 1990 (18,438 convictions) and 1996 (26,022 convictions). Thus, felony drug convictions for women in state courts have steadily risen between 1992 and 1996. 9 Referenced from W, BJS Publications, December 1999, pg. 6. 22 Table 3: Felony Convictions of Women in State Courts, 1990-199610 1990 1992 1994 1996 Drug Felonies 43,000 42,047 46,468 59,027 Trafficking 24,562 23,529 25,561 33,005 Possession 18,438 18,518 20,907 26,022 Critics of sentencing reforms claim that little if any consideration and therefore understanding has been given to women and the drug issue, and that sentencing reforms fail to recognize how women differ from men in terms of their experiences with drugs, abuse, and illegal activities. While several critics have highlighted the significance of gender in the sentencing and the war on drugs, few have addressed and explored any real impact of such policies for women. The next section reviews the history behind sentencing reforms, the background in Minnesota, and the impact of gender on the development of the new sentencing policies. 1° Referenced from W, BJS Publications, December 1999, pg. 5. 23 Sentencing Background Indeterminate sentencing, the idea that defendants should be reviewed and adjudicated on a case-by-case basis, was the dominant sentencing scheme throughout much of the early twentieth century. During the first two decades of the century, the Progressives and the policies they initiated greatly influenced sentencing policy for decades to come. Progressive reformers (e.g., social workers, psychologists, and psychiatrists, etc.) during this period established indeterminate sentencing and the options of parole and probation as well, and thus introduced a new flexibility in sentencing (Rothman, 1980). Based on the rehabilitative principles, indeterminate sentencing gave broad discretionary power and oversight to judges at the sentencing phase (Miethe & Moore, 1989). Morris and Tonry describe the scope of the discretion enjoyed by court officials, 24 Prosecutors had complete control over charging and plea-bargaining. Judges had little-fettered discretion to “individualize punishment” in deciding who received probation and who was sentenced to jail or prison, and, for those to be confined, to set minimum and maximum terms, and sometimes both. Parole boards, subject only to statutory provisions on parole eligibility, generally when a third of the maximum term had been served, decided who was released from prison prior to the expiration of their terms, when, and under what conditions. (1990: 20) Court officials enjoyed little supervision or review of their decisions under indeterminate sentencing. However, over the next decade this would all change. Determinate Sentencing: Getting Toggh on Crime In 1975, Maine became the first state to abolish their parole system and establish the first determinate sentencing system in the United States (Morris & Tonry, 1990:24). Today 14 states have abolished their parole systems for all or most offenses (Ditton & Wilson, 1999).11 Over the next two to three decades almost every state and the federal government revamped their sentencing systems. Morris and I‘The fourteen states include: Maine (1975), Indiana (1977), Illinois (1978), Minnesota (1980), Florida (1983), Washington (1984), Oregon (1989), Delaware (1990), Kansas (1993), Arizona and North Carolina (1994), Mississippi (1995), Ohio (1996), and Wisconsin (1999). A few other states abolished parole release for specific violent or felony offenses or crimes against a person (Alaska, New York, Tennessee, Virginia and Louisiana). 25 Tonry (1990) believe several factors were instrumental in driving this philosophical and policy shift from indeterminate sentencing to determinate sentencing. In addition to concerns over the perpetual crime problem, observers also criticized the amount of discretion officials held. Unlike other public officials, those in the criminal justice system were oftentimes not held accountable for their decisions. As a result, organizations like the American Friends Service Committee reported that the unfettered discretion of criminal justice officials had led to decisions that were seemingly inconsistent, unjust, and racially biased (Morris & Tonry, 1990). Additionally, empirically based studies also captured the presence of disparity in decision-making by judges and parole boards. Support for indeterminate sentences significantly diminished as a result of a growing dissatisfaction with the merits of rehabilitation as a punishment goal. Evaluations and reviews of treatment and programs for offenders resulted in little if any empirical evidence to suggest that we are effective in our treatment practices (Morris & Tonry, 1990). There were additional reasons for the shift in sentencing philosophy. The courts also played a role in the downfall of indeterminate sentencing systems. Prior to the 19708 courts maintained a “hands-off” policy in relation to corrections. Courts began to take a more careful look at procedural issues, practices that were fair and just. Finally, attacks on the philosophy and principles of 26 indeterminate sentencing came from both liberals and conservatives, albeit for different reasons. For the liberals, unbridled discretion opened the door for discrimination based on race. In the case of conservatives, criticisms were based on the view that crime was out of control and judges were soft on offenders (Morris & Tonry, 1990). Rehabilitation and the principles associated with it (e.g., discretion, parole, etc.) failed to address the crime problem, and treatment programs did not work (Walker, 1993). With support collapsing from both liberals and conservatives, indeterminate sentencing in this country quickly lost favor. In sum, all of these factors played a part in the dramatic shift in sentencing policies: Taken together, these critiques greatly undermined indeterminate sentencing and the practices and institutions that went with it. It is not easy to defend a major set of social institutions that are portrayed as based on unsound empirical, ethical, and psychological premises, as characterized by racial and class bias, by arbitrariness, by lawlessness, and by unfairness, and as conspicuously ineffective at achieving the larger social purposes of reducing crime and rehabilitating offenders - and few tried. (Morris & Tonry, 1990: 24) What was to follow was a period characterized by comprehensive sentencing changes that limited discretion of decision-makers and disparity in sentencing outcomes. Discretion was at the heart of the indeterminate sentencing 27 philosophy. The specific intent of the changes in sentencing legislation was to limit the discretion of judges and courts and place the responsibility in the hands of legislatures (Byrne & Taxman, 1994; Krisberg, 1994; Mauer, 1994; Myers, 1989; Petersilia, 1994; Platt, 1994). Changes were instituted to restrict discretion and sentence defendants based on their offense and criminal history, not on any notions of utilitarian benefits. A shift occurred in the goals of the legal system from focusing on crime causation and rehabilitation goals to retributive and deterrent efforts which resulted in making sentences more swift, certain, and severe (Petersilia, 1994). According to Walker, by the early 19808 reformers believed the best way to limit discretion was with the use of sentencing guidelines. The implementation of sentencing guidelines did not abolish discretion altogether; instead the guidelines suggested a presumptive sentence and allowed for a limited amount of discretion on the part of the judge (Walker, 1993). Discretion was limited through the use of written regulations and formalized procedures to structure decision-making. Sentencing reforms in the 19808 and 19908 involved other important changes as well, for example, flat sentences, mandatory sentences, three strikes initiatives, and truth-in-sentencing laws (Ditton & Wilson, 1999). Individual states revamped their own systems and used either one or a combination of these sentencing strategies to 28 address their crime problem. Sentencing guidelines can be found in over 20 states and mandatory sentences for certain types of offenses can be found in 49 states (Byrne & Brewster, 1993). As of 1994, some form of "three strikes and you're out" initiatives were passed in at least 30 states (Platt, 1994). In 1994 the United States Congress passed the Violent Crime Control and Law Enforcement Act of 1994 which provided additional funds to states to build more prisons and jails so offenders could serve a large portion of their sentence. One year later, 11 states had passed truth-in—sentencing laws that required inmates to serve 85% of their sentence (Ditton & Wilson, 1999). Recent sentencing strategies have attempted to limit the discretion of judges and courts and make offenders more accountable for their crimes. Minnesota’s sentencing guidelines system represents one example of a sentencing strategy that has sought to limit the influence of non-legal factors such as gender into the sentencing decision. Next, the chapter reviews the role of gender in sentencing reforms and addresses concerns for the implications of gender neutral sentencing policies. G n r n ntencin Reform Far—reaching changes in sentencing focused on factors that were considered to be appropriate in the sentencing decision process. Relevant factors, which had previously 29 focused on individual offenders and their rehabilitative needs, now focused instead on aspects of the offense (e.g., severity, prior record, etc.). As a result, factors that were once relevant in the sentencing decisions of women were no longer supported by the changes in the sentencing laws. For example, judges and the courts were no longer supposed to consider whether a woman was pregnant, or if she was a mother and primary caregiver (Nagel & Johnson, 1994). Many of the sentencing changes instituted in states around the country as well as the federal system emphasized the need to eliminate sentencing disparity based on race, gender and class. In turn many jurisdictions establishing new sentencing structures made it a point to restrict the ability of judges and other court officials to consider these factors when making sentencing decisions. For example, at the federal level, the Congress as part of the Sentencing Reform Act of 1994 particularly emphasized the need to eliminate “unwarranted” sentencing disparity. It instructed the Sentencing Commission to “de-emphasize” individual-based factors of the defendants and their case, which had been supported under a rehabilitation philosophy, and to establish sentencing guidelines that were neutral on race, gender, and class grounds (Nagel & Johnson, 1994; Raeder, 1993). 30 Raeder (1993: 908), a legal scholar, has argued that “mthe [Federal Sentencing] Guidelines, which are designed to reduce race, class and other unwarranted disparities in sentencing males, ignores factors that are integral to the lives of female offenders.” In an effort to not penalize impoverished minorities males, the elimination of factors involving family and community ties in the Federal and State Sentencing Guidelines have resulted in a disproportionate negative effect on women, according to critics (Raeder, 1993; Bryne & Taxman, 1994). Instead Raeder suggests that courts should take into account the unique circumstances of women by considering guidelines departures due to situations such as pregnancy, primary care for children, and single parenting. The aim of sentencing, according to Raeder, is for the sentence to be just and reflect gender differences in criminality and parenting responsibilities. Critics (Chesney-Lind & Pollock, 1995) further contend that such practices place women who are for the most part low risk and non-violent offenders behind bars, costing larger sums of money to house and creating difficulties in keeping families together. Edwards (1989: 178) also emphasizes this concern: 31 U (D A... Y;*h~.. -~-..a,_ .. 0 ‘E "‘ .~.. ~ . :" -.. m... Q- ‘. (I) in... ‘ I" ”1 The basic problem is that men and women are not the same and, more importantly, they do not start as equals. mArguing for males and females to be given the same treatment, when their capacities, resources and situations are not the same, means adopting a formal rather than a substantive standard of equality. And what relationship such a concept of equality bears to the notion of justice is a question currently taxing legal philosophers and feministsm. Those advocating for equal treatment under the law fail to recognize the social realities of women and men. There is the false assumption that men and women are on equal footing with one another within society and therefore the law should and will impact them equally. Advocates for women continue to struggle with the predicament of either arguing for equal treatment or for special treatment. Nagel and Johnson address the possible impact from incarceration of women on their dependent children. Several studies (Daly, 1987; Kaukinen, 1995; Steffensmeier et al., 1993) have shown that the courts do consider this issue when determining appropriate sentences for women offenders. However, advocating the inclusion of family ties and parental status into sentencing guidelines has the potential, Nagel and Johnson (1994) warn, to support gender stereotypes and expectations. Proof that differential treatment existed in our sentencing systems was just one of several catalysts influencing the movement to change sentencing decisions, and 32 gender represented one of the subgroups. In an attempt to rid the process of discretion and differential treatment, systems essentially had at their disposal one of three options. As Daly (1994) suggests, we could first, treat men more “like women,” second treat women more “like men,” or third we could “split the difference.” Proband (1997) argues that if a female standard had been used (i.e., using past sentencing patterns of women) to establish presumptive sentences there would have been a public outcry that the new sentencing policies were too lenient. Most policy makers relied on the latter two options, ignoring the possibility of using females as the standard by which to establish sentencing standards (Daly, 1994: 124). Consequently, research analyzing the impact of changes in sentencing philosophy and practices is necessary as well as important in order to understand the likely disproportionate effect of these changes on the sentencing of women offenders over time. To date there have been no studies that have looked specifically at this “disproportionate” question. The studies that have been completed have examined whether or not women received different sentences when compared to similarly situated men under a sentencing guidelines system. The existing studies will be reviewed in the next chapter in the review of the literature. Next, background information and a discussion of the Minnesota sentencing guidelines is presented. 33 Minnesota Sentencing Guidelines History. For much of the 20th century, Minnesota followed many of the national sentencing trends. During the first half of the century, indeterminate sentencing was the predominant system used in handling defendants. In the late 19608 Minnesota instituted mandatory minimum sentences, as was the case in many other state legal systems (Parent, 1988). Beginning in the 19708, however, Minnesota started to move to the forefront of sentencing reform in this country. Crime was a public concern in the state and Minnesota was about to do something about it. Crime in Minnesota mirrored trends occurring nationally for the most part. In state opinion polls, the crime issue was at the top of the public’s concerns. The state prison population in Minnesota was also dramatically increased during the 19708. In addition, a series of determinate sentencing bills were passed in the state legislature. All of these factors contributed to what was the backdrop of the climate in the state that led to the reforms (Parent, 1988). First authorized in 1978 with the establishment of a Guidelines Commission, the state of Minnesota subsequently adopted sentencing guidelines on May 1, 1980 (1978 Minn. Laws 244). The implementation of such guidelines was the culmination of many years dissatisfaction in the state with the previous indeterminate sentencing system in place. Minnesota underwent its reforms by instituting a commission approach. The Sentencing Commission was initially appointed 34 by the governor and chief justice and consisted of people representing trial and appellate judiciary, prosecution, defense, corrections, parole, and the public (Parent, 1988). The Commission originally was charged with creating a guidelines system that would regulate two sentencing decisions: (1) the decision to impose state imprisonment on defendants, and (2) the duration of such imprisonment (Frase, 1994: 11). The Commission also made the decision to eliminate parole, but institute a “good time” related policy with the intention to reduce sentences up to one-third off the specified sentence based on the offender’s behavior while imprisoned, and made the participation in all treatment programs while in prison voluntary (Frase, 1994). As outlined in the Minnesota Sentencing Guidelines Statement of Purpose and Principles, The purpose of the sentencing guidelines is to establish rational and consistent sentencing standards which reduce sentencing disparity and ensure that sanctions following conviction of a felony are proportional to the severity of the offense of conviction and the extent of the offender's criminal history. Equity in sentencing requires (a) that convicted felons similar with respect to relevant sentencing criteria ought to receive similar sanctions, and (b) that convicted felons substantially different from a typical case with respect to relevant criteria ought to receive different sanctions. (MN Sentencing Guidelines Commission, 1997: 1) 35 r-r v-v -0. by V'h §.. .. a. .—~\ ‘~\ §~ ‘ The Commission, thus, crafted a sentencing system based on a “modified just-desserts” (i.e., retribution) philosophy of criminal punishment. Under this type of system, the predominant factor influencing the appropriate sanction involved the severity of the current offense, followed by, to a lesser extent, the person’s criminal history (Moore & Miethe, 1986). As a consequence, the criminal offense became the focus and the individual offender who had previously been the focus under the indeterminate sentencing system became overshadowed. The Commission believed sentencing decisions should be free from influences of the defendant’s race, gender, social or economic status. Quidelinee. Minnesota’s sentencing guidelines represent the model sentencing reform and is probably the most widely recognized of all the sentencing reforms implemented during the 19708 and 19808. The guidelines were introduced to “establish rational and consistent sentencing standards which reduce sentencing disparity and ensure that sanctions following conviction of a felony are proportional to the severity of the offense of conviction and the extent of the offender’s criminal history” (MN Sentencing Guidelines Commission, 1997). According to Walker (1993), the Minnesota Sentencing Guidelines System comprised several “key” elements. First, the guidelines represent a series of “presumptive” sentences which judges are expected to use. They are referred to as “presumptive” according to the Sentencing Commission, since 36 “they are presumed to be appropriate for all typical cases sharing criminal history and offense severity characteristics “ (MN Sentencing Guidelines Commission, 1997: 61). A limited amount of discretion is possible under the system; judges are permitted to depart from the presumptive sentence but must provide written reasons as for the departure decision (Walker, 1993). Second, the state legislature identified clear objectives in their sentencing reforms. The first goal involved reducing or eliminating sentencing disparity based on race, gender, social or economic status. Under the new guidelines system, none of these factors were to be considered during the sentencing phase. The second goal had to do with controlling the use of prison resources, namely prison space. The Commission recognized that prison space is a finite resource and should be reserved for incarcerating the most serious offenders. Finally, what was key about the sentencing reforms in Minnesota was the establishment of a commission to develop the guidelines system. The sentencing guidelines are presented in a two dimensional matrix reflecting the seriousness of the current offense along with the defendant’s criminal history.12 Along the vertical axis, the Commission developed a scale representing the “Severity Level of Conviction Offense” comprising ten distinct levels. The offense severity level is based on the offense of conviction. When the defendant “ See Appendix A to view a copy of the guideline grid. 37 is convicted of multiple offenses, the severity level is determined using the more severe offense. Felony offenses are ordered along this scale ranging from low or Severity Level I to high or Severity Level X.13 First degree murder is not included in the severity scale due to the fact that it results in mandatory life in prison. Offenses included in each severity level are believed to be “generally equivalent” in severity (MN Sentencing Guidelines Commission, 1997: 2). Along the other axis of the sentencing grid, the Commission used a “Criminal History Score” index ranging between “0 points” for no priors and “6 or more points” for a more serious criminal background. A defendant's criminal history index is based on the following measures: (1) prior felony record; (2) custody status at the time of the offense; (3) prior misdemeanor and gross misdemeanor record; and (4) prior juvenile record for young adult felons (MN Sentencing Guidelines Commission, 1997: 5). The guidelines provided a uniform standard in determining a person’s criminal history. Information about the defendant’s criminal background was weighted producing a specific score. By standardizing and attributing weights to a defendant’s previous involvement in the criminal justice system, the Commission was able to increase fairness and equity in sentencing.“ 13See Appendix B for additional information on determining offense severity levels. “ See Appendix C for a review of how to calculate criminal history 38 The presumptive sentence for a defendant is determined by calculating her or his criminal history index score and identifying the severity of the current offense and then locating the corresponding grid cell. The corresponding cell indicates two things about the defendant’s presumed sentence: (1) whether the sentence involves imprisonment in a state penal facility or a community—based disposition (i.e., In/Out decision), and (2) the length of the sentence. According to the Commission, these sentences are “presumed to be appropriate for all typical cases sharing criminal history and offense characteristics” (MN Sentencing Guidelines Commission, 1997: 61). For cells involving a presumptive commitment to state prison (i.e., the region above the bold line) there exists both a single number that denotes the number of months to be served under the presumed sentenced, and a italicized range of numbers within which the judge can sentence the defendant and it not be considered a departure. For cells involving a presumptive stayed sentence (i.e., the region below the bold line or In/Out line) there is single number present.15 This represents the maximum number of months the judge can sentence the defendant to if she or he violates community index scores. “ Certain offenses in this region of the grid always carry a presumptive commitment to a state prison. These offenses include Third Degree Controlled Substance Crimes when the offender has a prior felony drug conviction, Burglary of an Occupied Dwelling when the offender has a prior felony burglary conviction, second and subsequent Criminal Sexual Conduct offenses and offenses carrying a mandatory minimum prison term due to the use of a dangerous weapon (e.g., Second Degree Assault). 39 supervision and probation is revoked. Viewed as a rational sentencing system, Minnesota Sentencing Guidelines attempted to maintain equality and proportionality, while emphasizing just desserts sentencing goals. While the guidelines matrix provides presumptive sentences for corresponding offenses and criminal histories, judges may depart from the recommended sentence if “the individual case involves substantial and compelling circumstances” (MN Sentencing Guidelines Commission, 1997: 22). The Commission requires written justification and that a high standard of departure be met so that the guidelines will be applied consistently and with a high level of regularity to reduce sentencing disparity. Upon deciding to depart from the presumptive sentence, the judge is required to provide written reasons or justifications for the departure, indicating why the sentence selected is more “appropriate, reasonable, or equitable” for the defendant. The Commission developed the guidelines in order that sentencing is neutral with regard to offenders’ race, gender, and income levels (MN Sentencing Guidelines Commission, 1997: 23). In addition to the offender's race, gender, and income level the Commission states that several employment factors (e.g., occupation, employment history, etc.) and social factors (e.g., educational attainment, marital status, etc.) should not be used as reasons for departure.16 Instead, factors related to the victim of the “ See Appendix D for a complete listing of factors to be excluded in 40 'r- V -0...- us. -r-u- ----~. ‘Pn. \— ~w». .. NW .- ‘.i‘_ U) V '9 - '- -- g: . -F offense and the role of the defendant in the commission of the offense can act as mitigating or aggravating factors.17 Since 1980, when the sentencing guidelines system was introduced in Minnesota, the guidelines have been revised to address changes in legislation. During the late 19808, the guidelines went through a significant modification in one particular area, crimes concerning illicit drugs. One of the objectives of this research was to examine the impact of this kind of sentencing reform on women offenders. The next section reviews changes to the guidelines as they relate to drug offenses. Drgg effeneee. Over the last 15 years the Sentencing Guidelines Commission in the state of Minnesota, like most jurisdictions in this country, became concerned about the drug crime problem. Prior to August 1, 1986, the state law only differentiated between the maximum penalties for the sale and possession of controlled substances. No mention was made of amount or type of controlled substance. However, the Minnesota Sentencing Guidelines Commission, in their ranking of drug-related offenses, did distinguish between not only possession and sale, but also type (MN Sentencing Guidelines Commission, 1992).18 From 1986 through August 1989 the state legislature made several more the departure decision. " See Appendix E for a list of factors to be included in making departure decisions. ” See Appendix F for a list of drug-related offenses and their severity level prior to August, 1986. 41 changes to the drug laws in Minnesota.19 On August 1, 1989 the Minnesota State legislature modified the drug laws extensively, creating 5 degrees of drug crimes (MN Sentencing Guidelines Commission, 1992). The degree of drug crime was defined by either possession or sale offenses and by the type of controlled substance and amount. Each of these factors contributed to the degree of the drug offense and also the subsequent changes to the sentencing guidelines as a result. Thus, after the passage of these state laws the commission preserved the degree structure outlined by the legislature, but for a few exceptions.” The exceptions were based on the belief by the commission that possession drug offenses at these levels were more serious due to the fact that they were likely to sell these larger amounts of ” See Appendix F for a list of those additional changes to the state drug laws. " The commission believed that because the legislature had specifically considered the seriousness of drug crimes as it created the new laws, it made sense to retain the degree structure within the severity level rankings of the guidelines. Thus, the commission ranked all drug crimes within each degree at the same severity level with one exception: 1‘t degree at severity level VIII, 2“ degree at severity level VII, most of 3rd degree at severity level VI, 4th degree at severity level IV, and 5th degree at severity level II. The exception the commission made to its ranking decision to keep all drug crimes within a degree at the same severity level was with regard to third degree possession of 3 or more grams of crack or 10 or more grams of cocaine/narcotic. This possession offense was not ranked at severity level VI with all the other 3rd degree drug crimes but was ranked at severity level VII. The commission had been urged by prosecutors and others to rank this possession crime at a level where the guidelines would recommend prison for the first time offender. These possession crimes were believed to be more serious because these individuals likely intended to sell these larger amounts of crack and powered cocaine. 42 drugs (MN Sentencing Guidelines Commission, 1992: 5). Aseeeements. The objectives of sentencing guidelines have not always been met in jurisdictions that have implemented them. In some cases there has been little or no effect on sentencing practices and for many states the result has been a tremendous increase in prison populations. Minnesota’s Sentencing Guidelines has been an exception to this trend. According to research from the early implementation of the guidelines, “During the first 2 years of implementation, Minnesota's guidelines significantly reduced sentencing disparities without putting additional burdens on correctional resources” (Miethe & Moore, 1989: 1). The initial success of the guidelines is believed to have occurred for several reasons. First, the guidelines in Minnesota are presumptive and supported by legal statute. In other states, guidelines are voluntary and in the end fail to reduce sentencing disparity because judges refuse to utilize them. Second, the Minnesota guidelines are “prescriptive” as opposed to “descriptive.” Thus, instead of relying strictly on past practices to suggest appropriate sentences (i.e., “descriptive”), the Commission in Minnesota established their own sentencing standards using a “modified” retributive philosophy (Miethe & Moore, 1989). Third, the Minnesota guidelines structures two key sentencing decisions: (1) whether or not to incarcerate the defendant, and (2) the length of the sentence. Presumptive 43 prison sentences are intended for offenders convicted of serious offenses. This particular guidelines, unlike others, indicates an in/out decision as well as a presumptive range, leaving little room for discretion and potential disparity (Meithe & Moore, 1989). Finally, in developing the guidelines the Commission was concerned with possible increases in incarceration rates and took into consideration the limited resource of prison bed space in the state. All of these factors, it was believed, lead to the successful implementation of guidelines early on. Miethe and Moore were funded in the mid-19808 to study the effects of implementing the guidelines on various aspects of the sentencing process. The questions they addressed included: 0 How did the introduction of guidelines impact charging, plea negotiations, and other sentencing practices from the preguidelines era? 0 Are sentences more uniform, neutral, and predictable? Their research examined sentencing trends during the first four years after implementation of the guidelines. The findings of their study suggest that sentences were in fact more predictable and uniform, particularly for the disposition decision (in/out) (Miethe & Moore, 1989: 3). The rate of departure from the presumptive sentence outline in the guidelines for disposition increased steadily over the four-year period from 6.2 percent in 1981 to 9.9 percent in 1984. On the other hand, the rate of departure for length of sentence decreased slightly over the four years from 8.4 percent in 1981 to 7.6% in 1984. Under the guidelines system social factors are not supposed to play a part in the sentencing process, yet over the four year period these factors (e.g., race, employment) retained some impact on both the disposition and duration sentencing decisions, albeit the impact was minimal. Results from the initial two years of the guidelines system indicated the presence of uniformity, neutrality and proportionality in sentencing. However, the authors found a changing trend over the last two years of the study period, “there has been some movement back to preguidelines levels in both sentencing uniformity and proportionality” (Miethe & Moore, 1989: 4). Additionally, Miethe and Moore used a survey to examine the attitudes of criminal justice officials on the subject of the new guidelines. The authors found that a high proportion of officials believed the guidelines were effective in gaining proportionality (90%), uniformity (92%), and neutrality (88%) in sentencing. Many believed the new sentencing system was an improvement over the older indeterminate one. Despite the fact that officials believed the new guidelines achieved their stated objectives, the survey also revealed the fact that officials “grudgingly” accepted the implementation of the guidelines and over the 45 study period found ways to circumvent the guidelines policies (Miethe & Moore, 1989: 5). These same officials were asked what changes should be made to the guidelines. Both judges and prosecutors responded they would like to see more flexibility and discretion be added to the process. Further, 16 percent of prosecutors and 20 percent of judges called for the discontinuation of the guidelines system. In order to get around their displeasure with what they viewed to be “unreasonable” sentencing procedures, criminal justice officials admitted they were altering their charging and plea negotiating practices (Miethe & Moore, 1989: 5). In comparison to preguideline trends, the study results suggest the Minnesota sentencing guidelines were successful in increasing uniformity, proportionality, and neutrality and consequently reducing disparity. In addition, violent offenders were more likely to be incarcerated than prior to the implementation of the guidelines. Both of these objectives were reached without increasing the overall rate of incarceration in the state. This study suggests that although the guidelines were initially implemented and carried out as intended, as time went on there was some erosion to the guidelines’ effectiveness, and criminal justice officials grew dissatisfied with them and looked for ways to circumvent the policies. This erosion was in part influenced by subsequent changes in Minnesota’s legislation that allowed judges to depart from the presumptive sentences 46 in certain sex offender cases. More importantly, in 1981, the state Supreme Court allowed departure based on a standard referred to as “amenability to probation” (Miethe & Moore, 1989). These changes, along with others, contributed to the degree of discretion allowable under the sentencing system: Each of these changes expanded the discretionary authority of criminal justice officials in relation to the guidelines—precisely at the time when increases in sentencing departures and decreases in uniformity and proportionality became apparent. (Miethe & Moore, 1989: 6) While evidence indicates that early guidelines use diminished the previous existence of disparity sentencing outcomes, modifications after its initial implementation suggest that the uniformity in sentencing and the neutrality, which was once evident, might change over time. Griswold (1987) tested the notion that sentencing guidelines should diminish sentencing disparity for like- situated offenders. To that end, Griswold examined Florida’s sentencing patterns from October 1983 to May 1984 to ascertain whether the newly implemented sentencing guidelines were meeting their goal. Consistent with other sentencing guidelines systems around the country, Florida implemented a system that is supposed to be neutral with 47 regard to a defendant’s race and socio-economic status (Griswold, 1987). Florida’s sentencing guidelines were developed to reflect to a certain extent past sentencing practices. Florida’s guidelines also permit departure using aggravating and mitigating reasons. Griswold’s (1987) findings indicate gender was important in the sentencing outcome for several types of offenses, including robbery, theft/forgery/fraud, and drug offenses. Sentences for women were more likely to be lower than the recommended sentence under the guidelines. Despite the implementation of a sentencing system aimed at reducing disparity, Griswold continued to find differential treatment. Over the past two to three decades many states in this country, along with the federal government, have significantly changed their sentencing systems. In developing new sentencing policies, the specific experiences and circumstances of women were disregarded. Some scholars are concerned that to address the crime problem in this country and respond more punitively and with certainty to young male offenders, the new sentencing strategies will disproportionately impact women. Early research completed by Miethe and Moore (1989) indicates similar sentencing decisions for women and men with comparable cases. However, it appeared that as time went on, officials unhappy with the 48 new sentencing policies found ways to circumvent the guidelines. Griswold’s findings also indicate that differential treatment based on gender can still occur after sentencing guidelines meant to limit discretion are implemented. Each of these studies examined the effect that gender along with race and class have had on the sentencing process after the implementation of sentencing guidelines. Neither of them, however, specifically considered the impact of sentencing changes over time and whether or not this has disproportionately affected women. The present research study fills the gap in research by examining sentencing practices over time before and after sentencing guidelines were implemented in Minnesota. Research Study The research was designed to examine the impact of implementing sentencing guidelines on the sentencing outcomes for men and women offenders convicted of drug offenses (i.e., nontraditional feminine offenses) as compared to property offenses (i.e., traditional feminine offenses). As indicated in this chapter and the chapter to follow, there is some evidence, particularly from research conducted in the 19708 and early 19808, that suggests that 49 some women have been treated more leniently and others more harshly by the legal system. Due to comprehensive changes in sentencing systems around the country, it is very likely that any presence of leniency and chivalry or of bias has diminished greatly. Many of the changes have attempted to restrict the discretion of judges and other court officials in determining the appropriate sentence for defendants. The traditional chivalrous treatment of women, in combination with recent changes in sentencing systems, suggests that this area needs to be revisited in order to understand the impact these policy changes have had, and to understand whether these changes have influenced women disproportionately. To this end, the sentencing outcomes for women and men were examined for three time periods: (1) a “pre-sentencing guidelines era,” (2) a “early sentencing guidelines era,” and (3) a “current sentencing guidelines era. The research addressed the issue of whether or not women are treated any differently from men by the courts for drug-related offenses. Organization Chapter one provided an overview of the research, including the statement of the problem and a brief orientation to explanations of why women have been incarcerated. Chapter one also described the influx of 50 women into prisons and jails and the life circumstances typical for these women, an overview of changes in sentencing with a movement toward determinate—based sentencing systems, including the state of Minnesota, and research questions. In the remaining chapters, the relevant literature and the research plan are outlined. Chapter Two provides a review of the literature on sentencing disparity and gender. Chapter Three provides an overview of the research, including a review of the data used and the analysis used to test the research hypotheses. Chapter four presents the findings and results of the data analysis and Chapter five discusses implications for policy and future research. 51 CHAPTER 2: REVIEW OF THE LITERATURE Chapter two presents a review of the literature on the relationship between gender and sentencing decisions. First, this chapter considers the link between criminological theory and decision-making in court processing. Second, the literature review discusses gender and sentencing in general, including the work of Daly and Bordt (1995), who completed a comprehensive review of studies on the subject. Next, the literature review provides an overview of important variables (legal and extra-legal), that along with gender appear to impact the sentencing decision of convicted offenders. These variables include prior criminal history, pretrial detention, offense severity and/or type, race, socio-economic status, marital status, and number of dependent children. In addition, the review looks at the quality of this research and draws some conclusions as to what is known, and more importantly what remains unanswered about the link between gender and sentencing. Studies regarding each of these variables has played a role in moving the examination of gender and sentencing outcome beyond merely a bivariate level explanation and understanding. 52 a. r. v- ; -... .2 .3 a: Gender and Sentencing The sentencing and legal process has been the subject of much research over the last 50 years. In looking at the research on disparity in sentencing based on gender, scholars have relied on several theories to guide their work including: conflict, labeling and functional theories (Bickle & Peterson, 1992; Nagel & Hagan, 1992)”, as well as social construction feminist theory, sex role theory22 and multiracial feminist theory (Kaukinen, 1995; Lorber, 1998). Additional chivalry, paternalism, familial paternalism, ‘evil woman’ and practicality explanations (Belknap, 1996; Daly, 1989; Flavin, 1995; Spohn & Spears, 1997)23 have influenced the literature in this area, and lastly jurisprudential or legal models (Crew, 1991; Curran, 1983; Flavin, 1995) have been considered. Much of the research on sentencing initially focused on the impact of race and economic standing on the decision-making process and sentence outcome. Conflict theorists maintained that discrimination based on race and economic standing allowed certain weaker segments of the population to be controlled “ See also Chevalier-Barrow 1992; Daly and Bordt, 1995. ” Also referred to as sex role traditionalism (Johnson & Sheuble, 1991). ” See also Nagel and Hagan, 1992; Bickle and Peterson, 1991; Edwards, 1989; Kaukinen, 1995. 53 through the criminal justice system. The role of gender in sentencing did not become a topic of interest in the field until the 19708. Crew (1991) suggests two reasons for the inattention to gender in the sentencing research. First, women were treated more leniently in sentencing decisions and this is inconsistent with conflict and labeling theories, which dominated the sentencing disparity research. Second, researchers have generally ignored women offenders because of their small numbers. Women represent such a small percentage of those people who come into contact with the criminal justice system that officials and scholars alike have not taken their offending very seriously. Not until recently, with the advent of the women’s movement, have women offenders been the focus of criminological theory and research in the criminal justice system. During the 19708, scholars and policymakers began to recognize that female offenders differed from their male counterparts in several ways. What explained crime causation and punishment for male offenders didn’t accurately portray the experiences of women in the system. The study of the relationship between gender and sentencing decisions is important for a number of reasons. First, sentencing represents the final outcome of several 54 decision points in the legal process. Ultimately, someone who stands for sentencing has been arrested, held over for trial, charged, and finally convicted of committing a criminal offense. The sentence is the culmination of this entire process in determining whether or not someone violated the criminal law and the type of punishment they should receive for doing so. Second, the courts as a part of the criminal justice system operate to socially control people. How the courts decide who to most completely control and who not to control as fully in turn is influenced by the discretionary use of this authority and power. Third, research that explores the presence of disparity in sentencing examines the possible misuse of discretion, authority, and power, but also examines the factors or reasons behind why discretion is used in a particular way. Finally, examining the influence of gender can shed light on the ways in which stereotypes and cultural images of women affect legal decisions about type and severity of punishment. When differential sentencing does occur, research can provide understanding of the reasons behind it and of how gender influences those reasons. With some exceptions, research on gender disparity in sentencing generally finds that women have received preferentially lenient treatment at the sentencing stage, 55 particularly in the decision to either incarcerate or grant probation (Daly & Bordt, 1995; Nagel & Johnson, 1994; Steffensmeier, Kramer, & Streifel, 1993). Several authors, however, temper their general conclusions on the lenient treatment of women by drawing attention to what they believe to be flawed research and methodology (Daly & Bordt, 1995; Steffensmeier et al., 1993). This will be explained in more detail later in this chapter. The literature on gender disparity at the sentencing stage has developed over the last 30 years. Factors relevant to the study of gender and sentence disparity can be categorized into one of two types: legal factors and extra—legal factors.24 Early research concentrated on the idea that extra-legal variables led to differential treatment among those that came before the courts. For example, early research often found that the courts treated women more leniently than men. While this pattern was accurate to a certain degree, it certainly did not tell the “ Some factors are referred to as contextual, describing the context in which the court functions and processes criminal cases. These types of variables include measures such as the size of the court (i.e., workload), urban vs. rural, % Black, % Republican, etc. Although these measures are important in describing the court setting and possible influences on the sentencing process, in the current research there is no practical reason for their inclusion. This study involves only two counties in the state of Minnesota. The lack of variation within possible contextual measures would result in the inability to conclude anything meaningful from the findings. See Kruttschnitt and Green (1984), Steffensmeier et al. (1993), and Daly and Bordt (1995) for a discussion on the importance of contextual effects. 56 complete story. Research conducted early on failed to control for factors like prior record and seriousness of the offense-both of which were legally relevant and also related to the defendant’s gender. Daly and Bordt (1995) reviewed and analyzed all published empirical studies (N = 50) looking at the link between gender and sentence outcome and published through the middle of 1990 and found mixed results overall. Their results indicated that approximately 50% of the cases reviewed had gender effects resulting in the lenient sentencing of women and another 25% each found either no gender effects or inconsistent effects (Daly & Bordt, 1995: 145) . In addition, Daly and Bordt (1995) considered the quality of the analysis for each study in relation to whether or not gender effects were found. Their findings imply that quality does in fact matter. As the authors expected, more rigorous studies (i.e., use of control variables, multivariate analyses, consideration of prior record) resulted in diminished evidence of gender effects. In addition, gender effects were more often found in the decision whether or not to incarcerate (i.e., In/Out decision) than in sentence length decisions (Daly & Bordt, 1995: 157). While it appears that gender has been an 57 important determinant of sentencing outcome, the literature indicates that sentencing decisions are more complex. Other variables, either alone or in conjunction with gender, seem to influence sentence outcomes as well. Next, the literature review details the importance of considering other variables besides gender in the sentencing decision. Legal factors are discussed first, followed by extra-legal factors including socio-demographic measures such as race, marital status, employment, and family structure. Recall that according to the thesis that women and men are treated equally in decisions about sentencing, legal factors such as seriousness of offense and prior criminal record should be most important in explaining sentencing outcomes. Legal Factgrs Theories and explanations such as the equal treatment model (Belknap, 1996), the jurisprudential model (Flavin, 1995) and the legal model (Curran, 1983) maintain that male and female offenders are sentenced based on legal factors. Thus, any differences in sentence outcomes (e.g., type of sentence, sentence length) between convicted male and female offenders are not related to their gender, but instead to 58 differences in their criminal history and offense severity. According to theories and explanations consistent with this perspective, legal variables are the most appropriate predictors of sentence outcomes. This would also be consistent with determinant based sentencing systems such as the Minnesota sentencing guidelines, which seek to remove extra—legal factors and sentence convicted offenders based on their offense and criminal history. A defendant’s criminal history has long been a consistent predictor of sentence outcome (Daly & Tonry, 1997). Early studies on gender and sentencing ignored this measure, resulting in what appeared to be leniency in sentencing decisions for women offenders. More recent studies on the subject have included the measure in their analyses and have concluded that criminal history is important in determining the sentence outcome. Researchers have operationalized criminal history in several ways, including prior convictions (Bickle & Peterson, 1991; Chevalier-Barrow, 1992; Spohn & Spears, 1997; Steffensmeier & Kramer, 1998)”, prior arrests (Kruttschnitt, 1985; Spohn & Spears, 1997)“, prior drug convictions (Spohn & Spears, 1997), juvenile record (Kruttschnitt & Green, 1984) and most ” See also Flavin, 1995; Ghali and Chesney-Lind, 1986; Zingraff and Thomson, 1984; Kruttschnitt, 1982. “ See also Kruttschnitt and Green, 1984; Ghali and Chesney—Lind, 1986. 59 serious or type of prior conviction (Bickle & Peterson, 1991; Chevalier-Barrow, 1992; Steffensmeier, Ulmer, & Kramer, 1998).27 Many studies confirm that legal variables are important in predicting the sentence outcome of convicted offenders (Bickle & Peterson, 1991; Chevalier-Barrow, 1992; Crew, 1991; Curran, 1983; Flavin, 1995; Kaukinen, 1995; Spohn & Spears, 1997; Steffensmeier et al., 1998;). For instance, Bickle and Peterson’s (1991) study looked at the sentences of convicted federal forgery offenders in eight federal district courts over a period from 1973 to 1978. For both men and women, they found the sentencing decisions appeared to be influenced by legally relevant factors such as prior record and offense seriousness. Spohn and Spears (1997) and Curran (1983) produced similar findings in their studies. In their analysis of gender and the sentencing of drug offenders, Spohn and Spears (1997) found that legally relevant factors significantly influenced sentence outcomes. Although gender was an important predictor of outcome, legally relevant variables such as crime seriousness (seriousness of the drug offense), being on probation, and prior criminal record (i.e., number of prior felony convictions, prior drug ” See also Kaukinen, 1995; Flavin, 1995. 60 convictions) were the most important indicators of sentence outcome (1997: 20). In the Curran (1983) study, the severity of the offense, along with number of prior arrests and number of total counts, all significantly impacted the sentencing decision. In the sentencing literature prior record is oftentimes measured as a simple dichotomous variable-- whether or not the defendant has a prior record. Chevalier-Barrow (1992) utilized a more comprehensive set of indicators for prior record including total number of adult convictions, number of adult convictions against a person, and prior juvenile convictions. Chevalier-Barrow (1992) found support for the connection of legal-based variables including seriousness of offense and prior criminal convictions. It is expected that legal variables play an important part in sentencing decisions, particularly because many states have moved from an indeterminate sentencing scheme, which allowed the use of discretion by the courts at sentencing, to a determinate sentencing system where discretion is highly discouraged. Consequently, legal variables should have a more profound effect on sentencing outcomes after the guidelines were introduced in 1980 in the state of Minnesota (i.e., time 2 and time 3). Since the guidelines in Minnesota are based upon current offense level 61 and a criminal history score, and are intended to remove extra-legal influences (e.g., race, income, gender) one would expect to find women and men with similar offenses and the same general criminal history be sentenced in the same manner. Given findings from prior research on the importance of extra—legal factors in the sentencing decision, even after the institution of sentencing guidelines, it is important that research also consider extra-legal variables. Extra-Legal Factors This section reviews the importance and relevance of extra-legal factors on sentencing outcomes. Theories and explanations such as conflict, functional, sex role, social construction feminism, multiracial feminism, practicality, ‘evil woman,’ chivalry, paternalism, and familial paternalism all have contributed to the existing knowledge of sentencing disparity and gender. Prior work in this area provides evidence that other extra-legal variables like race, dependent children, and employment act either alone or in combination with gender to significantly impact sentence outcomes. First, a review of the literature involving the relationship between race, gender, and sentence outcome is 62 discussed. Possible interaction effects are considered as well. Second, the literature review examines the significance of socio—economic factors, gender and sentence outcome. Third, the literature on gender, sentencing, and the relationship between family status variables, including the presence of dependent children and the nature of their care, along with marital status and living situation is reviewed. Finally, the review of the literature examines the link between gender and sentencing for drug offenses. Early research looking at extra-legal factors primarily focused on the part race and class had on sentencing outcomes. Much of this research attempted to determine whether or not discrimination was occurring in the courts. According to both the chivalry and the ‘evil woman’ explanations, the important contribution of extra-legal factors are evidence that patterns of disparity occur at sentencing. Age Researchers do not always consistently include age in regression models in the sentencing literature. In studies where age is included it is often entered into the analysis as a control variable with little theoretical interest being attached to the variable. Few studies have directly 63 examined the role of age in the sentencing literature. Steffensmeier, Ulmer, and Kramer (1998: 765) observe, Research findings on the age-sentencing relationship are sparse, and recent research reveals that it is more complex than is usually recognized. On the one hand, most analyses of sentencing merely control for age as a continuous variable and assume a linear effect; these analyses typically report a small or negligible age effect. On the other hand, several studies find when-the data are partitioned into “old” versus “young” subgroups-that elderly offenders (e.g., age 50 and over) are treated more leniently than younger offenders (e.g., offenders in their 208). Steffensmeier et al., (1995) found a slightly different outcome. The authors found a U-shaped distribution with the very young and those age 30 years and older receiving lenient sentences. The 1998 study also found a U-shaped distribution for age and sentencing, with offenders over 50 and under 21 receiving the least severe sentences (Steffensmeier, Ulmer, & Kramer, 1998: 8). Still other researchers have found no age effects in their sentencing research (Bickle & Peterson, 1991; Spohn & Spears, 1997). Race Several researchers have focused on the importance of the interaction between race and gender on sentencing outcome (Bickle & Peterson, 1991; Curran, 1983; Gruhl, Welch, & Spohn, 1984; Spohn et al., 1985; Sphon & Spears, 1997; Steffensmeir, Kramer, & Streifel, 1993). Work looking at the importance of race in sentencing has been framed using a conflict perspective, which rests on the belief that justice is administered disparately in order to protect the power and interests of white males (Chevalier—Barrow, 1992). Other theoretical perspectives have been influential in the work examining race and sentencing, including labeling and functional theories as well as feminist theories supportive of an interactions perspective, or what Lorber (1998) refers to as ‘multiracial feminism.’ Both labeling and functional theories rely on typescripts or stereotypes of groups of people (race in this case) to support power relationships within society (Bickle & Peterson, 1991). An intersections feminist approach acknowledges the fact that there are overlapping layers of power and oppression based on gender, race, and class for example. Young (1986) argues that black women are treated differently by the criminal justice system based on contrary gender expectations. Based on the notions of Lombroso and Pollak’s work, “good” women were viewed as idealizing traditional feminine qualities (e.g., passive, gentle, emotional) and “bad” women were believed to violate this 65 image (e.g., aggressive, deceitful, lacking maternal qualities, masculine, etc.) (Young, 1986). Black females typically were portrayed as bad women and the gender role expectations for black females were different than those for white women: There was little concern with race and class differences in the determination of the good woman, of the woman deserving of “protection." The gender role expectations of black females not only differed from those for good white females; in the case of black females, even the most positive characterization, that of the mammy, had negative implications. (Young, 1986: 311) In sum, there are clear indications that women of color and white women have been treated differently by both the courts and corrections systems.28 Research involving gender and decision-making stages in the criminal justice system must also consider the real possibility that there exists an interaction between gender and race in affecting decision outcomes. Differential treatment or chivalry may be selective based on the woman’s race, which in turn impacts gender role expectations applied to her. 2° In the early part of the 20th century, Rafter (1990) points out, there are two competing ideologies about women offenders; both are dominated by race images. First, there is the image of the ‘fallen woman’ who is likened to a child who is fragile and vulnerable. These offenders who were typically white were sent to reformatories where they would be trained to be better women and to know their place in society. The other image involved a ‘darker side’ of women. These offenders, who were for the most part black, were perceived to be more masculine, independent, assertive, and potentially violent and were treated as such. Their treatment was more severe and more in line with how males were punished by the corrections system. See also Klein, 1995. 66 Empirical research concerning the independent effects and the interaction effect of race and gender provide mixed results. Spohn and Spears (1997), and Curran (1983) failed to find race effects, while Spohn et al., (1985), Bickle and Peterson (1991) and Steffensmeier et al. (1993) discovered varying degrees of some support for race as an important predictor. Using data from felony cases heard in a northeastern city between 1968 and 1979, Spohn et al. (1985) looked at the interaction effect between gender and race in conviction and sentencing decisions. Their findings suggest that while women were treated more leniently than men were, the difference disappeared when race and gender were considered together. According to Spohn et. al. (1985:178), “the analysis reveals an interaction between race and gender that has heretofore gone unnoticed. While black women are less likely than black men to be incarcerated or sentenced harshly, their sentences are comparable to those of white men.” The complexity of the relationship between the gender and race of the defendant and sentencing outcomes in this study supports the intersection perspective that power and oppression are linked to social structures that are interwoven. It appears that black women are treated more harshly than white women which may indicate the occurrence of racial discrimination, and more leniently than black men which may indicate the presence of paternalism (or chivalry)(Spohn, et. al., 1985: 184). Bickle and Peterson (1991) examined whether or not the 67 impact of gender-based family roles varies depending on the race of the defendant. They looked at the interaction between race and the following variables: marital status, source of economic support, emotional support, and living arrangement. The findings from these interactions indicate that family roles usually benefited black women over white women (Bickle & Peterson, 1991). More specifically, being married and providing emotional support for dependent children resulted in a more favorable sentence outcome for black women as compared to white women. The authors suggest that it not enough to just be a mother for black women, they also must perform the role of mother “well” (Bickle & Peterson, 1991: 388). Another common group of measures in the gender and sentencing literature involves that of socio-economic status. This includes information about income, economic dependency and employment. The next section addresses the literature involving these measures. Soeig-eeonomie etatus Drawing upon such theories as the conflict perspective, the social construction of gender, and sex role traditionalism, socio-economic status measures are considered in several sentencing studies. Defendant’s socio-economic status is measured in several ways in the literature, including income (Kruttschnitt, 1980), sources of economic support or economic dependency (Bickle & 68 Peterson, 1991; Kruttschnitt & Green, 1984), and employment (Bickle & Peterson, 1991; Kruttschnitt, 1980). A woman’s economic status is important because whether or not she is employed and therefore independent or economically dependent (for example on her husband) is an indicator of the level of informal social control she experiences within the family setting (Kruttschnitt, 1984). Kruttshnitt (1984) contends that women who are economically dependent on men will be treated more leniently by the courts because of the informal social controls associated with the traditional female roles (e.g., wife, mother) they fulfill. Once again we see traditional gender role expectations impacting the sentencing decision for women, albeit this time in the form of economic dependency. Crew’s (1991) results also indicate an interactive relationship between gender and other factors. For example, consistent with Kruttschnitt’s work involving a dependency measure (i.e., being dependent financially), the combination of being a woman and also being unemployed resulted in more lenient sentences. Kruttschnitt (1980) looked at the significance of a woman's social status on sentencing outcomes. More specifically, she included measures such as economic rank and employment status, along with age and prior criminal record. Race and the income of the defendant were used as indicators for stratification and economic rank. Employment status was measured using temporarily unemployed, retired/health problems, welfare/not looking, housewife, 69 student, and employed, all measured using dummy variables. Kruttschnitt found that the social status of women was influential in the sentencing process. She concluded: ...there does in fact appear to be a significant relationship between the types of sentences accorded women offenders and their degree of social integration. A woman may not be engaged in full-time employment, but if she is perceived as either working toward that goal or fulfilling that goal in the home, she will probably be treated at least as well as, if not better than, she would be if she were employed. (Kruttschnitt, 1980: 259) Therefore, women who were poor and who were not seen as being part of the mainstream social arena, and who had been involved previously with the criminal justice system were treated more harshly by the system. We find in the literature, particularly from the work of Kruttschnitt, that gender role expectations are similarly linked to a woman’s economic dependency within the family setting related to what she refers to as a type of informal social control. From Kruttschnitt's work it appears as if a woman’s employment status and family status are intimately linked with one another. Being unemployed and not contributing to the economic status of the family in no way hurts women as long as they are fulfilling another expected role in society, that of mother, one of the variables discussed in the next section. 70 Femilx gtatgg Over the last decade sex role theory (i.e., gender role expectations) and social construction of gender, in addition to practicality, paternalism and familial paternalism explanations, have informed provocative research on the contribution of family indicators in the relationship between gender and sentencing. Family status is indicated by measures of marital status (Bickle & Peterson, 1991; Chevalier- Barrow, 1992; Crew, 1991; Kruttschnitt, 1980), the presence and number of dependent children (Bickle & Peterson, 1991; Chevalier-Barrow, 1992; Daly, 1987; Kaukinen, 1995; Kruttschnitt, 1980), as well as level or quality of care for dependent children (Bickle & Peterson, 1991; Kaukinen, 1995) and practicality issues stemming from the removal of a mother from her children (Steffensmeier, Kramer, & Streifel, 1993). Statistics indicate that a large percentage of women who are involved in the criminal justice system have children and are the primary care-givers for those children at the time of their arrest. As indicated in the previous section, Kruttschnitt found that the employment status (i.e., employed full time or working towards full time employment) of women along with their role as a full time mother in the home were connected with one another. Her research indicated that the courts believed being a full time mother in the home was as socially acceptable as being employed full time outside of the home. Therefore, women 71 who were unemployed were not penalized if it was shown they were full time mothers and stayed home with their children. They were in a sense being rewarded with lenient treatment by the courts for maintaining a traditional household, fulfilling their role in the family as mother and wife, as opposed to contributing to the economic area of the family. Daly (1987) explored the importance of having dependent children for female offenders and whether or not having children was influential on the type of sentences received in one northeastern jurisdiction. She examined the decision-making process of court officials by interviewing prosecutors, defense attorneys, probation officers, and judges. Many have attributed lenient treatment to the paternalistic views of the courts and judges. Daly developed this idea further by examining the role children play in the differential treatment. Her research indicated that paternalistic views of women were not as simple as previous research had reported. Instead the paternalism or the “protective” concerns toward women were in fact directed at their children or family. Court decisions protected the family in one of three ways(Daly, 1987: 282): 72 (1) keeping families together; (2) maintaining familied defendants’ labor for families, and especially women’s caretaking labor; and (3) protecting those dependent on a defendant’s economic support or care. Daly’s interviews with judges suggest that they use familial paternalism to rationalize differential sentencing. The emphasis on families interacts with the gender of the defendant to impact sentencing decisions. Thus, there is a difference between defendants with or without families and differences between male and female defendants with families (Daly, 1987). Court officials, according to Daly (1987: 284), “think of this differential treatment not as discrimination but rather as legitimate and pragmatic justice.” The significance of Daly’s work is its focus on the reasons behind the decision-making process of court officials concerning the gender differences and sentencing. Others account for this type of treatment on practicality grounds. “The practicality thesis contends that women are treated more leniently because of concerns about the welfare of children if women are incarcerated” (Bickle & Peterson, 1991:373). Kaukinen (1995), like Daly, examined the reasons behind differential sentencing based on gender. From interviews conducted with judges in Ontario Canada, Kaukinen (1995) generated typologies of judges based on their construction 73 of and use of motherhood in the sentencing process. She addressed the following questions in her study, “How is motherhood socially constructed by various judges and how were expectations formed and carried out in the decision to sentence a woman?” Kaukinen found that many of the judges she interviewed held expectations of women that were primarily based on traditional stereotypes of womanhood and limited gender roles. Further, she found that the defense lawyers used motherhood in order to explain women’s criminal actions. By doing this, defense lawyers believed their clients would receive favorable treatment by the courts. Thus court officials other than judges reinforced this particular view of women by also portraying women using traditional images. Kaukinen’s work not only acknowledges the relationships between a woman’s familial role and her sentencing outcome, but moves further beyond this general observation, as Daly did, by using qualitative methods in an attempt to understand how and why judges use the issue of motherhood during the sentencing process. The presence of dependent children was the single most important determinant of the sentencing decision for Kaukinen’s sample of judges. In addition, judges also looked at how the mothers carried out their duties. As a result, Kaukinen found that judges used one of four strategies in sentencing women offenders. First, some judges utilized traditional definitions of womanhood as well as motherhood in their sentencing 74 decisions. Women are assumed to have these family responsibilities and thus receive leniency based on these assumptions (Kaukinen, 1995: 8). The goal of this strategy is to keep families intact and therefore the mother at home with her dependent(s). Within this perspective is the assumption that women are mothers first, and provide the primary care for their children. “Motherhood is assumed to be a normal process within the life of all women” (Kaukinen, 1995: 68). The second strategy used by judges is to determine if the women who come before them and their courts are “good” or “bad” mothers. Just because you have dependent children and are the primary caregiver of those children does not necessarily translate into automatic leniency from this group of judges. Motherhood can actually work against some women if the judge believes the woman is a “bad” mother. As Kaukinen suggests, “mthese judgements often act to marginalize the conditions and experiences of some many womenm..[those] women who are lesbians, single, working outside the home or deviating in other ways” (1995: 68). Therefore, it appears that women are treated leniently only when they are deemed “good” mothers. A third group of judges tried to follow a gender— neutral strategy, and ignore the differential responsibility women typically have in the raising and care of their children. Finally, a group of judges tried to not presume things about the women in their courts, but instead 75 attempted to “sensitize” themselves to the unique and specific circumstances for each woman. Judges in this group attempted to gain an understanding of how different dispositions would affect the lives of these women. Therefore, judges in this group tried to understand the reasons why women were in the legal system and the circumstances around their involvement (Kaukinen, 1995). As a result of these various approaches, Kaukinen suggests: Sentencing strategies utilized by sentencing judges may thereby be seen as the result of the way in which judges identify and construct motherhood for women offenders. The sentencing of women lawbreakers consequently depends on the way in which women’s criminal behaviour and motherhood are socially constructed by the judiciary. (1995: 9) The reinforcement of traditional gender role expectations acts as a method of social control over women. Women are expected to be mothers and have primary responsibilities involving their role as mother. Judges using traditional definitions of womanhood reject the significance of women’s economic responsibilities. As a result judges do not recognize that many women who are involved in the criminal justice system are economically responsible for themselves and their family. In a sense, women are being punished for not meeting traditional expectations of womanhood. Women who do not fit these gender constructions are viewed as “true” deviants and 76 punished as such. Reinforcing these images of womanhood within the legal system legitimizes this particular view of women and maintains women in these prescribed roles and positions within our society. In addition, Kaukinen found that judges also used motherhood to construct and explain women's criminality, especially in the case of property offenses. For example, a traditional feminine offense such as shoplifting is often explained away by the woman's need to provide for her children. Thus, as Kaukinen puts it, “criminal behaviour is often defined as something arising out of the woman’s role as mother” (1995: 77). Constructing female criminality in this way supports the argument that leniency is also offense specific. In other words, women who commit traditional female crimes such as property offenses receive leniency because the criminal behavior falls within the scope of what is expected from them as women or as mothers.29 Judges in Kaukinen’s (1995) study believed women committed different ” According to Kaukinen (1995) judges in her study of traditional images of women and sentencing often tried to explain why a woman might steal or shoplift. She states, Rather than attempting to find explanations grounded in the social and economic conditions experienced by many women, some of the judges in the present research viewed female criminality in terms of the pressures of child care responsibilities. m.. Women offenders are assumed to be parents and their theft offenses are constructed in terms of their role as mothers providing care for their families. Women’s property offenses are identified as arising out of the pressures of motherhood. This judge had identified only one type of female offender, the mother struggling to feed her children. This construction does not describe the majority of women who are involved with the criminal justice system. Motherhood is identified and rationalized as the cause of women committing crime. (Kaukinen, 1995: 78-79) 77 crimes then men and for different reasons. “There appears to be a construction of the types of crimes all women commit and consequently the type of sentencing approach which is appropriate for all women” (Kaukinen, 1995: 42). This would explain other research evidence that finds that judges treat women in a punitive and severe way when they commit offenses that fall outside this rationality. Women who commit nontraditional female offenses such as crimes against persons or drug offenses are likely to find themselves dealt with in a more punitive manner. In committing these types of offenses, women have moved outside of what is expected of them in their roles as women and mothers. Consequently, they become labeled as “bad" mothers and the court feels no obligation to keep these women with their children and families. Other researchers also found positive support for the gender/motherhood and sentencing relationship (Chevalier- Barrow, 1992; Steffensmeier, Kramer, & Streifel, 1993). Chevalier-Barrow (1992) explores the significance of legal variables, socio-demographic characteristics, and family responsibilities in the sentencing decisions for both felonies and misdemeanors in the state of Pennsylvania in 1977. As for the family status indicators, Chevalier-Barrow found women received preferential treatment even after controlling for family status variables. In addition, family status significantly affected the likelihood of receiving a prison sentence. Those offenders who were not married and 78 those offenders who had fewer dependent children were more likely to be incarcerated (1992: 54). Additionally, there were no gender interactions with either of the family status measures in relation to any of the sentencing outcomes. Overall, Chevalier-Barrow (1992) discovered that legal factors were most important in explaining sentence differences between men and women offenders. Still, there was some support for the idea that courts do consider the potential impact of removing mothers from their children. Judges appeared to be unwilling to remove women from their family setting. This was an opinion shared by the judges in Steffensmeier's (et. al., 1993) study as well. Judges rationalized this differential treatment based on their belief that the conditions of prisons were bad and no place for women with dependent children. How judges and other court officials rationalize disparate sentences should provide insight into the presence of paternalism or chivalry, and other reasons for differences in sentencing. This rationalization process, however, can also work to the detriment of women offenders. As we have seen, when women tend to live their lives outside of what is expected from them based on traditional gender role expectations they may in turn be penalized for their behavior with harsher sentences from the courts. The type of offense committed by women is also intricately tied to gender role expectations. In their analysis of convicted federal forgery 79 offenders, Bickle and Peterson’s (1991) examined similar variables as did Kaukinen. Bickle and Peterson (1991) examined the importance of gender-based family roles on sentencing decisions over a period from 1973 to 1978. The study utilized several measures of family status/ role factors, including marital status, the presence of dependents, support for dependents, defendants’ source of economic support, and the defendants’ living arrangement. By determining the level or degree of support to children, the authors were able to consider two possible factors: (1) the nature of the family role the defendant served, and (2) how well they performed their family role (Bickle & Peterson, 1991: 379). The authors found that family roles do play a part in determining sentence outcome and are related to the defendant's gender. Women offenders were less likely to receive a prison sentence than were men offenders (34.7% vs. 50.2 %). Additionally, the factors important in the sentencing decision were different for men and women. 'Unlike Kaukinen and Daly’s results, Bickle and Peterson (1991) found that family role measures had little to do with the sentencing outcomes for women. Instead, important factors involved legally relevant variables (i.e., prior record, seriousness of the offense, and pretrial custody). For men, legal variables were also significant (i.e., number of counts, offense seriousness, prior criminal convictions) as well as employment. Two family status variables, marital 80 status and emotional support for dependents, significantly influenced the sentence outcomes of men. Men who were married30 were more likely to receive a prison sentence, whereas men who did not give significant emotional support to their dependent children were less likely to be incarcerated for their crime (Bickle & Peterson, 1991). Based on the research to date, gender role expectations tied to the family (i.e., wife and mother) are oftentimes influential in the sentencing outcomes for women. Whether this is due to concern for keeping families intact because women tend to be the primary caregivers for their children, or because of practical concerns of where the children would live and who would care for them needs to be studied further. How, if any, would the response from the courts (i.e., lenient sentencing practices) change if the crimes committed by women involved non-traditional feminine offenses such as drug-related crimes? Women fulfilling a gender role expectation of motherhood on one hand, may be viewed as violating another gender role expectation in the type of crime they commit, thus negating any possible leniency they ordinarily experience. In the next section, this question is addressed in the available research on the sentencing of women for drug-related offenses. ” According to Bickle and Peterson this relationships resulted in an unexpected direction. The authors explain, “Perhaps more severe punishment accrues to male offenders who are married because officials perceive them as doubly deviant: they violate the law and, in so doing, jeopardize the well-being of others (i.e., their wives and perhaps children). 81 Drug Offenses Despite the increased numbers of women convicted for drug offenses, little research has considered directly the impact of gender on the sentencing of drug offenders. What research has been completed on gender and the sentencing of drug offenders indicates mixed results, with some studies indicating evidence of preferential sentencing decisions for women (Albonetti, 1997; Chevalier-Barrow, 1992; Spohn & Spears, 1997) and other studies evidence of non-preferential sentencing decisions (Daly, 1987; Steffensmeier et al., 1993) . A prior drug offense is potentially important as a predictor of sentence outcome for women offenders. Both Steffensmeier‘s and Daly’s work provides a hint of evidence that judges oftentimes view women drug offenders as being as culpable and as likely to recidivate as men drug offenders, which is not the case for most offense categories. Daly’s (1987) work suggests that while women with children often receive leniency in their sentencing outcomes, this is not true when the case involves a drug offense. Steffensmeier, Kramer, and Streifel (1993) found a similar outcome using 1985-87 sentencing guidelines data from Pennsylvania. Regarding sentencing for drug offenses, the authors found that gender had a negative effect for drug violations with women receiving sentences which were “slightly longer” than those given to male defendants (p. 82 430). Several judges during the course of follow-up interviews commented that “female drug offenders were unlikely to get a ‘break’ because they are ‘every bit as likely to get into trouble as are male druggies’ ” (Steffensmeier et al., 1993: 434-35). Bush-Baskette (1996) explored Chesney-Lind and Fienman’s contention that the “war on drugs” has been a “war on women,” and that the dramatic increase in the incarceration of women offenders has been the result of the increased punitive focus on drug offenses. Felony incarceration rates for the federal system and the state systems of New York, New Jersey and Florida were studied for the years 1945-1987 (federal level) and 1960—1987 (state level). A longitudinal analysis was completed looking at the effect of the “war on drugs” in changes to federal and state policies had on incarceration rates. Bush-Baskette (1996) concluded that the impact of the “war on drugs” in the federal system and in New York, New Jersey, and Florida did not have significant influence on incarceration rates of women offenders. Spohn and Spears (1997), like Bush-Baskette, recently examined Chesney-Lind’s position involving the “war on drugs” and its effect on women offenders. To that end, Spohn and Spears (1997) indirectly tested31 Chesney-Lind’s contention that the increasing number of women flooding the ” Spohn and Spears acknowledge that they indirectly test Chesney-Lind’s argument since they are using cross-sectional data and not looking at treatment of women drug offenders over time. 83 n: f prison systems around the country is due to the “increased willingness of contemporary judges to sentence female drug offenders to prison” (1997: 2). In addition, the researchers explored the importance, if any, of “familied” defendants and the presence of prior drug offenses. As indicated previously, there is some evidence that shows that chivalry or preferential treatment of women by courts is dependent upon whether or not they have children and are caring for them, or in some cases serving another highly valued gender role (e.g., wife). Using data for convicted felony drug offenses from Cook County, Illinois, Spohn and Spears (1997) built on the works of Chesney-Lind (1995), Daly (1987), and Steffensmeier et al., (1993) and tested two propositions. First, the growth in the female prison population is the result of contemporary judges’ willingness to imprison women convicted of drug offenses, and second, there is an interaction between gender, other offender characteristics, and sentence outcome. In the case of the former proposition, the researchers hypothesized that there would be no gender differences in either the sentencing to prison or sentence length decisions. For the latter, Spohn and Spears made the argument that gender may be influential for “certain types” of offenders, but not others, thus the necessity to explore the specified interaction effects. Spohn and Spears (1997) found that males were more likely than females to be sentenced to prison, even after 84 controlling for legally relevant facts such as prior record and crime seriousness. Thus, it appears that even female drug offenders experience significant levels of preferential treatment. With regard to the possible interaction effects of gender with other influential factors, particularly ” the results dependent children and prior drug offenses, indicated that gender did not influence the sentencing decision for offenders who had dependent children. The same pattern held for the interaction between gender and a previous drug offense. There were no gender differences in whether or not the offender was sentenced to prison. However, for those offenders who had no prior drug offenses, the analysis revealed that men were significantly more likely than women to be incarcerated (Spohn & Spears, 1997). Although limited in its generalizability in testing Chesney- Lind’s position, the results do provide minimal evidence in support of the argument. Spohn and Spears’ work indicates that women still experience a significant amount of preferential treatment by the courts, even for drug offenses. However, as indicated by Daly’s work, women who had dependent children and who were convicted of drug offenses were not given this same leniency. This finding was supported in Spohn and Spears' work as well. This is consistent with the proposition that women who are perceived to be “bad mothers” might not receive the same treatment as women who are perceived to be ” The authors did not consider the interaction between gender and race. 85 “good mothers.” Based on interviews with judges, Spohn and Spears concluded that “women convicted of drug offenses, like women convicted of child abuse or prostitution, may be perceived as inadequate mothers whose children would be better off living with relatives or in foster homes” (1997: 28). Two additional studies also produced results that indicate lenient sentencing for women who committed drug— related offenses. Chevalier-Barrow (1992) found evidence of leniency in her study of gender and sentencing in the state of Pennsylvania. Using data collected prior to the introduction of sentencing guidelines in that state, Chevalier—Barrow found the presence of an interaction between gender and current offense. In the case of drug offenses, women were sentenced more leniently then similarly situated men. Significant differences between sentencing practices involving women and men disappeared when sentence length was considered as the dependent variable (Chevalier- Barrow, 1992: 51). Albonetti (1997) examined sentencing disparity for drug defendants sentenced under the federal sentencing guidelines system. Her research also found that gender, along with other defendant characteristics (e.g., ethnicity, education), significantly influenced sentencing outcomes for federal drug offenses. Exploring the relationship between gender and sentencing for drug-related offenses is worthwhile. Significant sentencing reforms over the last one to two 86 decades in response to “war on drugs” declarations of the 19808 allows for an important examination of the policy implications of such reforms. As was argued earlier, policies in the criminal justice system are oftentimes developed and implemented neglecting women and their involvement with crime. Sentencing reforms, and particularly sentencing reforms in the area of major drug legislation presents us with an opportunity to explore this concern and to address the question of whether or not sentencing reforms, including those in the area of drug offenses, have disproportionately affected women in comparison to men. The next section reviews the literature on existing studies of gender and systems with sentencing guidelines. Stsdies on Gender and Sentencing Guidelines Decisions Research examining the occurrence of gender disparity in sentencing outcomes in jurisdictions having implemented sentencing guidelines is limited to studies conducted in states such as Florida, Pennsylvania, and Minnesota, as well as at the Federal level. Many of the evaluations of guidelines systems to date have focused on disparity in a general sense, looking at a myriad of offender characteristics (e.g., class, race, age, and gender). The research findings indicate that even after the 87 implementation of determinate—based sentencing policies in jurisdictions around this country, the defendant’s gender still appears to be influential in sentencing, whether it involves sentencing to prison (Steffensmeier et al., 1993), reasons for departures from the recommended sentence (Kramer & Ulmer, 1996; Steffensmeier et al., 1993)”, or length of sentence (Albonetti, 1997). F ral n en in uideline Albonetti (1997) analyzed sentencing outcomes in connection with defendant characteristics, guilty pleas, and guideline departures for drug offenses under the federal sentencing guidelines for data from 1991-92. She found that gender, along with race, significantly impacted sentence length with females receiving shorter sentences. In addition, there was evidence of an interaction between gender and race. Black women tended to receive shorter ” Steffensmeier et al., (1993: 433) found five justifications for departure from sentence guidelines that favored women defendants and they included: (1) defendant has a nonviolent prior record (e.g., a high prior record score that consists solely of property offending), (2) defendant has mental or health problems (e.g., jailing would over— burden the jail staff and would harm rather than help the defendant), (3) defendant is caring for dependents or is pregnant (e.g., jailing would not protect the community in the long term and would be inhumane, risky, and possibly costly), (4) defendant played a minor role in the crime or was only an accomplice, and (5) defendant showed remorse (e.g., “felt bad about what she/he had done”). 88 sentences in comparison with similarly situated white women. Pegssylvegia Sesseneing Guidelines Studies by Steffensmeier et al. (1993) and Kramer and Ulmer (1996) examined disparity under the Pennsylvania sentencing guidelines. Steffensmeier et al.’s (1993) data from 1985-1987 indicate “modest support” for the gender disparity model. However, as the authors point out, the effect of gender was influenced by how the dependent variable was defined in the analysis. For instance, when “In/Out” was defined by incarceration in either jail or state prison women were less likely to be incarcerated when compared with men. When “In/Out” was defined by incarceration in state prison only, women and men were equally as likely to be incarcerated (Steffensmeier et al., 1993: 424). Finally, the authors found that gender had no impact on the sentence length decision. In additional analyses of the Pennsylvania sentencing guidelines system, Kramer and Ulmer (1996) considered possible disparity tied to sentencing departures. According to the authors, departures represent a “window of discretion” for judges working within a guidelines system (1996: 81). Departures represent a way for judges to sentence defendants outside of the prescribed ranges and 89 therefore introduce possible disparity related to individual defendant characteristics. A8 in Minnesota, the Pennsylvania system requires judges who depart from the sentencing guidelines to explain the reasons for the departure in writing. Legal factors, particularly criminal history and offense severity, is most influential in dispositional departures. As for gender, women are two times as likely to receive dispositional departures in comparison to men (Kramer & Ulmer, 1996). Kramer and Ulmer (1996) found that gender had little if any impact on durational departures and legal factors contributed the most in explaining departure decisions. Its not enough to simply conclude that gender is influential in departures based on the finding that women are more likely than men to receive a sentence departure. More important are the reasons behind the departures. Kramer and Ulmer concluded that departure reasons could in fact be attributed to a defendant's race and gender. The most common downward departure reasons based on race and gender involve: 9O 1) Defendant is remorseful/good candidate for rehabilitation; 2) Guilty plea/plea bargain; 3) Defendant is caring for dependents, court is unwilling to disrupt family ties; 4) Defendant is employed, court is unwilling to disrupt job ties; 5) Offense or prior record is qualitatively less serious than the guideline scores indicate. (Kramer & Ulmer, 1996: 98-99) The authors report in their findings that the above reasons for departure are often given in the cases of women and white people. Further, interviews conducted with judges indicate that under sentencing guidelines, gender and race stereotypes seem to assist judges in determining who should get a “second chance” or who is likely to not recidivate, and consequently receive a sentence departure (Kramer & Ulmer, 1996: 99). In an attempt to hold on to some level of decision- making and discretion, judges may be finding ways to circumvent sentencing guidelines when they do not agree with the prescribed sentence. For example, several of the departure reasons identified above in Kramer and Ulmer’s study are similar to several extra-legal factors that are related to gender and to lenient sentences for women (i.e., the presence of dependent children, amenability to rehabilitation). Judges use what discretion “windows” they have available to them to sentence in ways they feel are 91 appropriate given case specific situations. Steffensmeier, Kramer, and Streifel’s (1993) work involved sentencing guidelines data from Pennsylvania as well. Using 1985-87 data the authors analyzed the extent to which gender influenced judges’ decision to incarcerate defendants. The authors found that gender influenced the likelihood of the defendant being imprisoned (women being imprisoned less often then men), but no difference was found in the length of the sentence. Steffensmeier et al.’s (1993) result suggests that gender has a slight impact on the in/out decision when compared to the effect from the defendant’s criminal history. Thus it appears from Steffensmeier’s work that gender may not impact both sentencing decisions (In/Out and sentence length) equally. Flerige Senteneing Guidelines Florida’s guidelines system is based upon four factors: (1) the primary offense and additional offenses, (2) prior convictions, (3) legal status at the time of the offense, and (4) the amount of victim injury (Griswold, 1987). In a prior analysis of experimental guidelines in Florida it was determined that women were sentenced more leniently when compared with men.34 Griswold (1987) evaluated Florida’s “ In the state of Florida, sentencing guidelines were developed and 92 sentencing guidelines in order to determine if new sentencing reforms had abolished unwarranted sentencing disparity. The author examined the influence of gender and several other specified independent variables on the dependant variable, sentencing deviation.35 Using this type of dependent variable permitted the author to use a measure that was sensitive to size of deviation from the recommended sentence as opposed to simply whether or not the sentence deviated from the recommended sentence (Griswold, 1987: 322): Griswold's (1987) results showed that gender was related to deviations for three offense categories: robbery, theft/forgery/fraud, and drugs as well as for all offenses.36 Further the direction of deviation was positive, which meant women were sentenced below (i.e., more leniently) the recommended sentences for these offenses. These findings add evidence to the position that even after implementing determinate sentencing reforms aimed at reducing disparity, disparity is still present in sentencing. Griswold’s study also makes the point that continued disparity may be offense implemented in four of twenty circuits. An initial study of the guidelines was conducted eight months after their implementation. After the apparent success of the guidelines in the four circuits, the sentencing guidelines were implemented state—wide. ” Sentence deviation was computed by first subtracting the Actual Sentence from the Recommended Sentence and then dividing that figure by the Recommended Sentence. “ Gender was not significant for the following offense categories: 93 L. TL. 8 V . C.» Z. a: specific or as the author contends, related to the seriousness of the offense. Minnesota Sentencing Guidelines Moore and Miethe (1986) evaluated the Minnesota sentencing guidelines one year after its implementation in order to determine whether or not system objectives were being met (i.e., sentencing based on legally relevant variables). According to their findings, sentencing outcomes substantially complied with the prescribed guidelines. As would be expected under the guidelines system, the seriousness of the convicted offense and criminal history were the “primary determinants” of the presumptive disposition and duration (Moore & Miethe, 1986: 266). As to whether or not departures were used in Minnesota as they were in Pennsylvania, Moore and Miethe found that for the most part exceptions to the guidelines were “situationally specific” and did not appear to be used in a significant way to circumvent use of the guidelines.37 Murder/manslaughter, sexual, violent, burglary, weapons, and other. ” Moore and Miethe (1986) did however qualify their conclusion based on two issues. First, they state, “we cannot with absolute certainty rule out model misspecification as an alternative explanation for the generally low predictive power of our equations” (268). Further, the authors contend that the generally low predictive power of the equations “reflects a generally high degree of compliance with the Commission’s policies on the use of sentencing departures and consecutive sentences. Second, and of more concern is the possibility that departures from the presumptive sentence are used to “adjust” sentences in accordance with 94 Dispositional departures were used in fewer than only 15 percent of all felony sentences (Moore & Miethe, 1986: 269). Knapp (1984) also reviewed the Minnesota Sentencing Guidelines system in her three-year evaluation after the initial implementation of the guidelines. In Knapp’s analysis of relative severity of sanctions across gender subgroups, she found that in 1981 (the first year after the implementation of the guidelines) women offenders received shorter sentences (7.6 months less) than their male counterparts (Knapp, 1984: 67).38 In the following year, women still received shorter sentences; however, the difference decreased somewhat (6 months less in 1982). In 1983 this trend changed, with women receiving slightly longer (3 months) sentences than a comparison group of men (Knapp, 1984: 68). Over a two year period after the guidelines had been implemented, there appears to have been a significant shift in sentencing men and women offenders, at least in regard to sentence duration. what officials feel are appropriate in the specific type of case. ” The severity of sanctions for gender was determined by comparing the severity of sanctions given to women with those given to men. This controls for severity level and criminal history score so that differences can be examined for similarly situated offenders. In the care of gender, male dispositional patterns were applied to women offenders and then compared with what the women actually received. See Knapp (1984) for a further explanation regarding this analysis. 95 Conclgsion A review of the literature concerning gender and sentence outcomes in systems having undergone reforms shows mixed results. Despite the introduction of sentencing reforms around the United States that sought to eliminate differential or biased sentencing practices, the research offers evidence that gender in some cases is still influential in sentencing decisions. Given the number of studies conducted in this area, it is a bit premature to draw any firm conclusions; much more research needs to explore this issue. The lack of research on the role of gender in determinate sentencing systems is one of several criticisms of the literature to date. Criticisms surrounding the literature on gender and sentencing practices suggest some fruitful ground for future research endeavors as well as ways in which research can be improved. These criticisms are outlined in the next section. Cgisisism of the Gender agg gegseneing Reseersh Over the last two decades researchers have examined whether or not a defendant’s gender has any impact on the type of sentence received (Armstrong, 1977; Daly, 1987; Kruttschnitt, 1980; Mann, 1984; Spohn, Welch, & Gruhl, 1985; 96 Steffensmeier, Kramer & Streifel, 1993; Steffensmeier, 1980; Zingraff & Thomson, 1977). While findings have been mixed, it is generally believed that women have experienced a noted amount of preferential treatment from judges and the courts. Findings from early studies on gender and sentencing indicate women were treated more leniently than men were. However, these same studies were later criticized for their weak methodology, including neglect of important control variables (Steffensmeier, 1980; Zingraff & Thomson, 1977). Researchers have identified several common problems with the literature on gender and sentencing. In addition to methodological weaknesses, other criticisms include the use of old datasets, and conceptual limitations tied to explanations of lenient sentencing for women. Several methodological problems, particularly in early studies on gender and sentencing, have clouded our understanding of the relationship between gender and sentencing outcomes (Curran, 1983; Flavin, 1995; Kramer & Ulmer, 1996; Sphon & Spears, 1997; Steffensmeier, 1980; Steffensmeier et al., 1993). First, early studies on sentencing disparity failed to control for relevant legal variables such as prior record and the seriousness of the offense (Steffensmeier, 1980; Steffensmeier et. al., 1993; Zingraff & Thomson, 1977). Studies not controlling for prior record and seriousness of the offense may have inappropriately concluded that women were treated more leniently than men, when in fact the leniency was the result 97 of women committing less serious offenses and having a less extensive criminal record than men had. Also, recent studies examining the relationship between gender and sentencing decisions have included other non- legal variables believed to be influential such as race (Mann, 1984; Spohn, Welch, & Gruhl, 1985), social status (Kruttschnitt, 1980), and familial factors (Daly, 1987; Steffensmeier, 1980). Last, Steffensmeier (1980) points out that many of the early studies used weak analytic designs (i.e., bivariate analyses). In more recent studies the use of multivariate techniques have enabled researchers to control for other possible influential factors such as those of age, employment status, and race (Steffensmeier, 1980). Another area of criticism involves the use of out of date datasets in examining the relationship between gender and sentencing outcomes (Flavin, 1995; Kramer & Ulmer, 1996; Nagel & Johnson, 1994; Spohn & Spears, 1997; Steffensmeier et al., 1993). A significant number of studies looking at gender and sentencing outcomes have been criticized for using old datasets and consequently not being representative of today’s sentencing systems. Nagel and Johnson (1994: 182) suggest: 98 Much of this research contained in these works is based on data collected in the 19608 and 19708. In the 19808, however, significant efforts were made to reform sentencing systems at both the state and federal levels. These reforms were designed to substantially reduce judicial sentencing discretion, to reduce unwarranted sentencing disparities, and to reduce race, gender, and class discrimination. Moreover, these reforms, at least at the federal level, shifted the focus of sentencing from “offender” characteristics, such as family and community ties, education, and employment, to “offense” characteristics and the offender’s criminal history. If successful, these reforms will reduce the favorable treatment previously afforded female offenders, by increasing both their incarceration rate and the length of their sentences. As reported previously in this proposal, scholars writing in the area of gender and crime have discussed the arguable impact that sentencing reforms have had on women offenders. A larger body of literature using more recent data (19808 and 19908) reflecting sentencing policy changes would help shed light on how these reforms have impacted women since their implementation. The research outlined here will look at the sentencing patterns for women before and after the implementation of sentencing reforms in order to gain an understanding of how, if any, sentencing has changed for women in relation to men. A final concern with the available literature relates to the conceptual problems with differential treatment 99 (i.e., leniency for women). Spohn and Spears (1997: 4) indicate, [JJudges who sentence female offenders more leniently than similarly-situated male offenders may be motivated more by beliefs of blameworthiness and by concerns about the social costs of incarcerating women than by paternalism or stereotypes of sex-appropriate behavior. In other words, leniency may not be associated necessarily with one’s gender per se but instead associated with what Daly and Tonry (1997: 232) refer to a as “gender-linked criteria” such as care for dependent children and amenability to treatment and rehabilitation. Summary The results of the literature review on gender and sentencing outcomes provides evidence that both legal and extra-legal variables have been influential in sentencing decisions. This suggests that in some cases the equal treatment position or the legal model (or jurisprudential) is supported. In other cases the differential treatment position or conflict theory, sex role theory, and chivalry as well as familial paternalism explanations are supported. Just how and when this occurs is in need of further um clarification. This study seeks to address several issues that remain unresolved. First, it is evident from the review of the available literature that the role of gender in sentencing is unclear. In the current study, gender was examined at three different time periods: (1) pre-guidelines, (2) early guidelines, and (3) current guidelines. This research ascertained how gender influenced sentencing over a span of almost twenty years, encompassing significant sentencing changes. Very few studies of this nature have been completed to date and those that have been conducted have involved shorter time periods and have included gender as just another variable. Second, a review of the literature shows that leniency as advocated by the chivalry perspective is not universal for all women, but instead is explained by other factors related to gender, race, and type of offense such as femininity, motherhood and marriage. The selective application of the law to women in many cases has been tied to traditional gender expectations, for example traditional ‘feminine’ crimes and care for dependent children. The current research examined the discretionary sentencing of women offenders based upon other gender-linked factors such as caring for dependent children and marital status (i.e., traditional gender roles). Again, unlike other studies, the research explored how the influence of these social Characteristics have changed (if at all) when moving from an 101 H indeterminate sentencing philosophy to a determinate one. Third, the small number of studies specifically looking at the impact of gender differences in the sentencing of drug offenders indicates that this research was greatly needed. Two aspects were important and were pursued within the current research. As the literature suggests, leniency is sometimes denied to women who commit drug offenses. Evidently judges perceive the level of blameworthiness for women drug offenders to be on level or above that for men drug offenders. Therefore the current research used sentencing data for drug-related offenses and property offenses to look at possible interactions between gender and offense type. The study compared gender disparity across two categories of offenses, one representing traditional “feminine” offenses (i.e., property offenses) and the other representing non-traditional “feminine” offenses (i.e., drug offenses). Additionally research on sentencing drug offenses was needed in order to look at the effect of the “war on drugs” on women offenders. Fourth, very few empirical investigations of Chesney- Lind’s notion that the “war on drugs” has been a “war on women” exist. The one identifiable investigation utilized a cross-sectional analytic strategy. This study intended to build upon the previous study by looking at the “war on drugs” and gender using data from a time period before major drug legislation was enacted in the state of Minnesota and using data from a period of time after the implementation of 102 major drug legislation. Research Questions The following research questions guided the research plan: (1) Are convicted women less likely than convicted men to be sentenced to prison in general, and for each of the three time periods, controlling for type of offense? Of those offenders sentenced to prison, are women likely to have shorter sentences than men? Research suggests that women have commonly been sentenced in a lenient manner. However, research has shown that leniency may be crime specific. For example, women who commit traditional “feminine” crimes may continue to enjoy a certain amount of preferential treatment, whereas, women who commit nontraditional offenses (e.g., drug offenses, homicide), may face severe sentencing or at the very least, sentencing that is equivalent to that given to men committing similar offenses. Women who commit nontraditional “feminine” crimes violate their gender roles and in a manner not socially acceptable based on their role in society. It appears that these women receive more severe responses by the courts. 103 U) (1 In 'r! h Will leniency be evident across all time periods? Sweeping changes in sentencing systems and a shift to determinant sentencing philosophies and practices have tried to reduce and even eliminate sentencing disparity that is based on individual—level factors including gender. Thus, we might find leniency under the “pre-guidelines” era (i.e., 1978) when judges had a tremendous amount of discretion in handling cases, but not under the “early guidelines” era (i.e., early 19808) or the “current guidelines” era (i.e., 1994) due to the introduction of determinate-based sentencing structures. (2) Are different factors predictive of the sentence outcomes of men and women for drug and property offenses, in general and for each of the three time periods? Have these predictive factors changed over time for men and women? Leniency may be also situation specific, whether or not a woman is “familied” a8 Daly suggests. For women who fulfill their socially acceptable roles such as mother and caregiver in the family, research has indicated, also experience more preferential treatment, in comparison to women who don’t fulfill these roles. Consequently, women who either do not have children or are not the primary caregivers of them (e.g., do not reside with them) may find themselves treated 104 more harshly by the courts. Again, prior to the sentencing guidelines being created in Minnesota, the judges had expanded discretion and were more at liberty to use different factors in sentencing men and women defendants. After the implementation of the sentencing guidelines, it is conceivable that the judges could no longer use different factors in deciding appropriate sentences, thus the same factors would be used in the decision. (3) Has the “war on drugs” been a “war on women?” In other words has major drug legislation disproportionately affected women in terms of sentencing in comparison with men? Chesney—Lind suggests that the women are the unintended targets of our recent “get tough on drugs” strategies, yet have felt the majority of the punishment for drug—related offenses. She contends the “war on drugs” has translated into a “war on women” and that recently instituted drug laws intended to address young, violent, male drug dealers have instead disproportionately affected women offenders. The sentencing of men and women drug offenders was examined and compared across time periods in order to look at treatment separately for men and women before and after changes in drug legislation during the late 19808. In conclusion, this chapter provided an overview of the 105 theories and explanations in the literature concerning the relationship between gender and sentencing. The chapter began by discussing several general reviews of the literature such as the one completed by Daly and Bordt (1995). The chapter continued with a discussion of the relevance of legal and extra-legal factors to sentencing outcomes stemming from conflict, labeling, and functional theories, along with chivalry, familial paternalism, and sex role explanations. In addition to gender, this included a review of studies on sentencing decisions examining variables such as prior record, offense type, race, socio— economic status, marital status, and motherhood. The chapter next discussed the influence of considering gender in the development of recent sentencing reforms throughout the United States. Prior research studies were reviewed, demonstrating problems with weak methodology and out-of—date datasets. The next chapter outlines the methodology that was used to test the stated hypotheses. 106 CHAPTER 3: METHODOLOGY Hypotheses A review of the literature on gender and sentencing disparity and differential treatment suggests several hypotheses: H1: Under the pre-guidelines sentencing period (1978) women offenders who fulfill traditional gender role expectations (e.g., mother, commit traditional feminine offenses) are significantly more likely to receive lenient decisions at the In/Out sentencing stage, when compared to men in similar situations. Under the pre—guidelines sentencing period (1978), no significant differences will be found between men and women offenders at the sentence length decision stage. Under the early guidelines sentencing period (1981,82 and 84), no significant differences will be found between women and men offenders in either of the two sentencing outcomes (i.e., imprisonment, length of sentence) they receive, even after taking into account measures related to gender role expectations. 107 Under the current guidelines sentencing period (1994), no significant differences will be found between women and men offenders in either of the two sentencing outcomes (i.e., imprisonment, length of sentence) they receive, even after taking into account gender role expectations. As a result of introducing sentencing guidelines and changes in drug legislation in the state of Minnesota, women will have been disproportionately affected in a negative manner (i.e., reflecting a larger gap between sentencing practices before and after the sentencing changes) when comparing sentencing practices over time. As a result of the “war on drugs” era, one would expect that the probability of incarceration would increase more dramatically for women when compared to men. The Study Site Minnesota was selected for the study site for several reasons. First, the state has been a leader in developing and implementing sentencing reforms over the last twenty years. The study offers the chance to examine a model determinate sentencing system that underwent extensive planning and development prior to its implementation. Unlike other sentencing guidelines systems, Minnesota 108 :5 utilizes a presumptive approach that is legally binding. Many other sentencing guidelines systems are voluntary and have been easily thwarted by court officials who do not agree with philosophical changes and/or eliminating their discretion at sentencing. Consequently, Minnesota’s system is a good site to examine the impact of “real” sentencing reforms on women. Second, the Minnesota Sentencing Guidelines Commission (MSGC) which was created with the specific purpose of developing and implementing guidelines is likewise responsible for collecting sentencing information for all convicted felons on an annual basis and makes that data available to researchers around the country. In addition, the researcher was involved in prior research with the study site (Hennepin and Ramsey Counties) and therefore had access to other necessary data (i.e., PreSentence Investigation Reports from 1994). Finally, data were available for offenders sentenced prior to the implementation of guidelines through the commission. Minnesota Sentencing Guidelines Commission collected data on convicted felons prior to implementation for use in their own internal comparison and evaluation study. The existing data enabled the researcher to look at factors of interest pre- and post-guidelines implementation. 109 V4 C\ The Data The data used in the current analysis was obtained from two sources: (1) the Minnesota Sentencing Guidelines Commission, and (2) Ramsey and Hennepin County PreSentence Investigation Reports39 (for 1994 dataset only). Datasets for three specific time periods were used in order to compare the sentencing process and outcomes for men and women for drug-related offenses and property offenses within and across these time periods. The datasets used in the analysis include the following: (1) 1978; also referred to as “Pre-guidelines Data” (2) 1981—82,84; also referred to as “Early-guidelines Data” (3) 1994; also referred to as “Current-guidelines Data” The three time periods were selected in order to achieve a number of objectives. First, data from 1978 represents information on sentencing practices in the state before the establishment and implementation of sentencing guidelines in 1980. Thus, it captures sentencing decisions under an indeterminate sentencing system; hence, there might be more 39Supplemental data for socio-demographic measures were collected for those defendants sentenced in 1994 for drug and property offenses in the counties of Ramsey and Hennepin in order to make comparisons with the other time periods. 110 disparity found in the sentencing decisions between male and female defendants during this time period. Second, sentencing patterns from the 1978 dataset were compared with sentencing patterns from two time periods after the implementation of sentencing guidelines policies (i.e., “early guidelines” or 1981-82, 84 and “current guidelines” or 1994). The 1981—82, 84 dataset documents sentencing decisions immediately after the implementation of the guidelines system. The Sentencing Commission collected in-depth information on the sentencing process as part of a statewide evaluation of the guidelines.40 The information contained in the dataset is comparable to the information collected as part of the 1978 dataset. The 1994 dataset is essential to the analysis since it represents sentencing that reflects policies put in place after the advent of the “war on drugs” era which resulted in major drug legislation in the late 19808 in the state of Minnesota. Therefore, use of these datasets permits an analysis that examines the impact of this “war on drugs” on the sentencing predictors and outcomes for offenders prior to the introduction of major drug legislation in the state and afterwards to gauge the influence of the legislation on sentencing decisions. The Guidelines Commission has collected and continues “ The MN Sentencing Commission for 1983 did not collect in-depth information about the convicted offenders. Per a phone conversation with Anne Wall of the Sentencing Commission no reason was given for the absence of this type of information. 111 to collect sentencing data on convicted offenders for monitoring purposes. The Commission has collected data on the sentencing of convicted felons on an annual basis since the guidelines were implemented in 1980. The information collected by the Commission is generally consistent from year to year; however, specific types of information have been collected reflecting the needs of the Commission and the State. Consequently, it was necessary to match measures in order to make the necessary comparisons for the research plan. For comparison purposes, the Commission prior to the implementation of the sentencing guidelines collected baseline data from 1978. Information contained in the baseline dataset (year, 1978) included the following types of measures: court processing (e.g., case number, county, date of conviction) disposition and plea information, alleged offense information and victimization, offender characteristics, criminal history, and evaluation from the Pre-Sentence Investigation Report. The dataset for years 1981-82 and 1984 was initially collected by the Commission to study the impact of the sentencing guidelines on sentencing patterns. Thus the information collected in this time period is comparable to the information collected in the baseline dataset from 1978. It provides information on offender characteristics, as well as other important crime processing measures. In addition, this dataset documents information on appropriate sanctions and departures under the new 112 f) ("X n- determinate-based system. The third comparison time period consists of a dataset for the year 1994. The information available for this specific dataset is much more limited in comparison to the previous time periods. The Commission did not collect socio-demographic information on the defendants sentenced that year. However, this information was available in the PSI Reports in Ramsey and Hennepin Counties, the two most populated counties in Minnesota. Sample The sample consisted of convicted drug and property offenders sentenced during three time periods: (1) 1978“, (2) 1981-82,84 and (3) 1994 in Hennepin (Minneapolis) and Ramsey (St. Paul) counties. The 1978 dataset included convicted felony offenders who were sentenced for either drug or property offenses between July 1, 1977 and June 30, 1978 in both counties of interest.42 The second dataset (i.e., 1981, 1982, and 1984) consisted of a sample of all convicted felony drug and property offenders sentenced between May 1, 1980 and September 30, 1981, from October 1, 1981 to September 30, 1982, and from October 1, 1983 to “ Fiscal year 1978 (July 1, 1977 to June 30, 1978). “ The entire data set from which the sub-sample of drug and property offenders were drawn consisted of all female offenders and a 42 percent random sample of male offenders who were sentenced for a felony offense during the 1978 fiscal year. The sample contained approximately half of all offenders convicted of a felony in that year. The data were weighted by sex (Wall, 1999). 113 I‘h (W September 30, 1984 from Hennepin and Ramsey counties.43 The final dataset included all females convicted of drug-related or property felony offenses during the calendar year 1994 and a 40% random sample of all males convicted of these same offenses from Hennepin and Ramsey counties. A sample of all convicted male drug and property offenders in these two counties was necessary since supplemental information needed to be collected and merged with the existing data. Because of their small numbers, all women were selected for the study from the 1994 data. Limitations of time and money for the research study made it necessary to draw a sample of males. Two factors were important to consider when selecting an appropriate sample size. Level of confidence refers to the amount of error the researcher is willing to accept. ” The 1981,1982, and 1984 data sets were supplemented with in-depth data collected on samples of cases. The 1981 in-depth data set contained the population of cases committed to the Commissioner of Corrections and samples of cases given stayed sentences from eight of the more populous counties (Anoka, Crow Wing, Dakota, Hennepin, Olmstead, Ramsey, St. Louis, and Washington). The samples of cases with stayed sentences were stratified by gender and race with female and minority offenders sampled at a higher rate than white males to increase the representation of the smaller sub-populations. Data were weighted to indicate the population of cases the samples are designed to represent (Wall, 1999). The 1982 and 1984 in-depth data sets contained the population of offenders committed to the Commissioner from the eight county area mentioned above, as well as samples of stayed cases from each of the eight counties. The samples of stayed cases were stratified by gender and race. Again, data were weighted to indicate the population of cases the samples were designed to represent (Wall, 1999). Extensive data were collected for study purposes from corrections and court files and included information such as offense behavior, victimization, initial charges, plea negotiation, and offender characteristics. The samples of cases were stratified by disposition, county, race, and sex. See Knapp, 1984 for more details. 114 VOA :- nu mtg C. Researchers typically select either a 95 percent level of confidence (5 percent chance of error) or a 99 percent level of confidence (1 percent chance of error) (Rae & Parker, 1992). The second factor, confidence interval, relates to sampling accuracy. In other words, “sample size is directly related to the accuracy of the sample mean as an estimate of the true population mean” (Rae & Parker, 1992: 126). In order to determine the appropriate sample size for variables expressed in proportions (i.e., In/Out sentence outcome) the relationship between confidence interval, level of confidence, and the standard error of the sample were necessary: (I = i Zakn) (3.1) Where c; = confidence interval in terms of proportions, Z: = Z score for various levels of confidence (a). and Up== standard error for a distribution of sample proportions (Rae and Parker, 1992: 129). The standard error for the true population mean proportion is: 09:: ——————- , therefore (3.2) CF12. flu—fl (3.3) 115 The sample size therefore yields: ”=2. _P_(1_-_€)_ (3.4) c, Next, calculating specific sample sizes requires the values of Za,(3 and p be known. The most commonly used value for p. Zais 1.96 for the 95 percent confidence level followed by a value of 2.575 for the 99 percent confidence level (Rae & Parker, 1992: 129). According to Rae and Parker'C% is “typically set not to exceed 10 percent and is more frequently set in the 3—5 percent range depending on the specific degree of accuracy to which the findings must conform” (p. 129). Thus it is recommended that the sample proportion fall within a range of i3, :4, or :5 percent of the true proportion. Since the true proportion (p) is not known, some scholars recommend that a conservative value be applied to handle this uncertainty. The most conservative value would require the largest sample size be drawn. This occurs when p is a value of .5 (Rae & Parker, 1992: 129). Since the population was relatively small with males convicted of drug and property offenses in Hennepin and Ramsey counties (numbering 1,912) an adjustment needed to be made to the standard error to include the finite population correction (a variation from equation 3.4). The equation for calculating the appropriate sample size when the population is small is the following (Rae & Parker, 1992: 116 TIA Fun 17‘ ~11 ~\~ he 131-132): 7,:[20 M. M] 2 (3.5) V C} VIN-1 Equation 3.5 yields the following: I]: 2 Za’ifll-PHN (3.6) 20 [19(1— ,0)] + (N ‘1)Cp Recall that in the above discussion p =.5. The appropriate sample size was therefore calculated as follows: _ (1.96)2(.25)(l,912) — (1 .96)2(.25) + (1,91 1)(.o3)2 The above equation (3.7) yielded the following result: 0:685 Using a 95 percent confidence level and a :3 percent confidence interval (i.e., a conservative range) in terms of proportion, resulted in a sample size of 685 convicted male offenders necessary to obtain accurate sample estimations. A forty percent stratified random sample (stratified based on county and offenses) of male offenders was drawn, taking into account possible missing cases. A 40 percent sampling strategy resulted in a total sample of 825 U7 -1 convicted males in 1994. This was well above the prescribed sample size calculated in the equations above. Supplemental information from Pre-sentence Investigations Reports was collected for the 1994 sub-sample in order that information on key socio-demographic indicators could be merged with the existing Minnesota Sentencing Guidelines dataset and comparisons could be made across time periods.44 The supplemental data included: living status, marital status, number of dependent children, involvement with the children (economic, living), education background, and employment status. Specification of Variables of Interest The Dependent Variables The study examined whether gender alone or in interaction with other factors resulted in a more lenient sentence outcome for women offenders. A list of the variables and their definitions are provided in Table 4. For the purposes of this study, sentence outcome was operationalized using two distinct measures: the decision of whether or not to incarcerate the offender (IN/OUT) and sentence length. While an In/Out distinction may blur variation surrounding sentence outcome (i.e., providing a “ The Minnesota Sentencing Guidelines Commission did not collect this 118 dichotomous outcome versus a more detailed categorization of outcome), Minnesota's sentencing system was designed specifically with this In/Out dichotomy in mind.“5 Recall the sample consists of only convicted offenders, therefore data existed on all sample members for each measure of the dependent variable sentence outcome. The first measure of sentence outcome, the type of sentence (incarcerate vs. community sanction) was coded into a binary variable reflecting an In or Out decision outcome (i.e., 0=out or not incarcerated, and 1=in or incarcerated in prison). The second measure of sentence outcome was defined as sentence length. The data on sentence length were measured in terms of months. Th In en ent Varia le Previous research on gender and sentence outcomes suggested several important variables of interest. The independent variables specified in the analysis were grouped as either control variables or variables of interest. Table 5 identifies each of the independent variables and the expected relationship (i.e., direction) between them and the In/Out sentence outcome. The control variables included information for convicted persons sentenced in 1994. “ Minnesota's sentencing guidelines were designed using an In/Out sentence decision outcome. The Sentencing Guidelines Grid uses an 119 several legal factors such as seriousness of the current offense, prior misdemeanors, prior felonies, and custody status. Each of these factors was believed to have a positive relationship with sentence outcomes across all three-time periods (i.e., lead to a higher likelihood of incarceration). For example, having a prior felony results in a higher likelihood of receiving a sentence of incarceration than the defendants not having this sort of background characteristic. It was expected that each of these measures would affect the sentence outcome and must be controlled for. Control variables also included the county in which sentencing occurred (Hennepin or Ramsey) in order to account for possible differences in the adjudication process. Finally, age and level of education were also included in the study as control variables. A negative relationship between both age and education and sentence outcomes (i.e., older defendants sentenced more leniently) was anticipated for the first time period, while no significant relationship was anticipated for the subsequent time periods. In/Out line to guide sentencing decisions. 120 Table 4: Independent and Dependent Variables Variables: Codes: Variable Definition: Independent Vari abl es Extra-Legal Female 0: No Defendant’s gender 1: Yes Nonwhite 0: No Defendant's race 1= Yes Age Years Defendant’s age at time of conviction Single 0: No Defendant’s marital status 1: Yes Dependent children 0: No Defendant’s dependent 1: Yes children (under 18 yrs) Education level 1: 8t5 and below 2: 9th to 12th The highest education grade completed by the defendant 3: College Employment 0: No Defendant's employment 1: Yes status at the time of the offense Hennepin County 0: No County in which sentencing 1: Yes occurred Independen t Vari abl e8 Legal Drug Offense 0: No Drug or property offense 1: Yes Prior Felony O: No Prior record for any felony 1: Yes offense Prior Misdemeanor = No Prior record for any 1: Yes misdemeanor offense Custody Status 0: No Defendant's custody status 1: Yes at the time of the offense (e.g., probation, parole) Severity Level 1 to 10 Severity level based on Guidelines ranking. Dependen t Vari abl es Sentence (Prison) 0: No Defendant’s sentence 1: Yes includes incarceration? Sentence length Months Defendant’s length of sentence in months 0 Maximum length used for 1978 data 121 Table 5: Expected Relationships between Independent Variables and the In/Out Sentence Outcome, by Time Period variable: Time 1 Time 2 Time 3 variables of Interest Female - None None Nonwhite + None None Single + None None Dependent Children - None None Employment - None None Drug Offense None None + Con trol Vari abl es Age - None None Education - None None Severity Level + + + Prior Misdemeanors + + + Prior Felony + + + Custody + + + Hennepin County None None None Dependent Variables: (1) Incarceration Sentence (In/Out); 1= Prison (2) Sentence Length (months) The literature suggests that in addition to gender, extra-legal variables such as race, marital status, living arrangements, dependent children and employment status are also important predictors. These variables of interest were all expected to have varying amounts of impact on sentencing decisions. More specifically it was anticipated that each of these factors would be influential during the first time period when higher levels of discretion by judges and the courts were condoned. Recall that under a determinate sentencing system such as the one implemented in Minnesota 122 none of these factors were supposed to influence sentencing decisions. Therefore, it was expected that during Time periods 2 and 3, none of these factors would have a relationship with sentence outcome. Offense type (i.e., drug offense or property offense) was also a variable of interest due to the idea that leniency is sometimes connected to the type of offense and whether or not it is a traditional female crime or a non-traditional one. It was believed that an offender with a drug offense would receive more severe sentence outcome based on Daly (1989) and Steffensmeier et. al.,‘s (1993) research. Data Analysis Several types of analytic statistical techniques were used in order to test the hypotheses including descriptive statistics, bivariate associations, Logistic regression analysis, Ordinary Least Squares (OLS) regression analysis, and equivalence of regression coefficients analysis. Desspippive Statistics and Bivariate Relationships First, descriptive statistics were used to summarize and describe general information from the data about the entire sample of property and drug offenders. Second, the analysis looked at the relationship between the specified 123 variables, examining the degree, direction, form and significance, and statistical independence between each variable with the other (Cuzzort & Vrettos, 1996). Depending on the level of measurement for each variable, a t-test, Chi—square or gamma were used to ascertain the nature of the bivariate relationship between variables (Agresti & Finlay, 1986). Logispie Regression Analysis In order to understand the role gender plays in determining sentence outcome for convicted drug offenders two types of regression analyses were used, logistic regression and ordinary least squares regression. Logistic regression analysis was used to determine the significance of gender and other relevant independent variables, such as the presence of dependent children, on the dichotomous dependent variable “sentence outcome (In vs. Out)”. Under Minnesota’s sentencing system, “In” refers to incarceration in a state run correctional facility while “Out” refers to a lesser sentence such as incarceration in a locally run facility and/or probation. Y=a+B.x.+B.x.+B.x.+ B.x.+e (3.8) Logistic regression is a form of Ordinary Least Squares regression (3.8), but is intended for analyses using dichotomous dependent variables such as the In/Out 124 V a Va I: sentencing decision here (Bachman & Paternoster, 1997; Kennedy, 1997; Long, 1997). In this instance, the dependent variable is binary, meaning there are only two possible values for the variable (e.g., yes or no). Logistic regression allows one to analyze the effect of one or more independent variables on a dichotomous dependent variable. The relationship between the variables is nonlinear (Bachman & Paternoster, 1997). The logistic regression model differs from the OLS model because it does not require the same assumptions. For example, the logistic model does not require constant variance. However, there are similarities between both models: both state that the dependent variable is a function of one or more independent variables, data are randomly selected and observations are independent (Bachman & Paternoster, 1997). The logistic model assumes that all predicted probabilities relate to the area under the normal curve. The logistic regression model is estimated using the maximum-likelihood estimation method (MLE). Essentially, “the coefficients are estimated so as to maximize the probability or likelihood of obtaining the observed data” (Bachman & Paternoster, 1997: 572). Estimating coefficients using MLE provides the “greatest probability of obtaining the observed data” (Kennedy, 1997: 21). The logistic model estimates provide information relating to the odds of an event occurring (i.e., probability of an event occurring over the probability of an 125 event not occurring). The estimated equation gives the natural log of the odds that an event will occur. Thus, the regression coefficients can be interpreted as the change in the natural log of the odds of the dependent variable associated with a one-unit change in the independent variable (Bachman & Paternoster, 1997: 574). This does not hold too much intuitive meaning. Consequently, one would want to convert these natural log odds into antilogs, which indicates the type of relationship (positive or negative) and amount of change based on a one-unit change in the independent variable.46 As in the case of OLS regression models, one can examine the significance of the estimated coefficients for logistic models as well. This enables the researcher to consider the individual effects of each predictor on the dependent variable (i.e., sentence outcome) controlling for all other variables in the model. The specified logistic model can also be examined in terms of its “goodness of fit” to the data. It answers the question of how well the model accounts for variation and predicts the dependent variable. Several logistic regression equations were estimated in order to study possible gender effects on sentencing ‘°With a continuous independent variable, the logistic regression coefficient cannot be directly interpreted in probability terms. In addition, because the relationship between the independent variable is nonlinear, so that the effect of x on the probability of y depends on the value of x, the logistic regression coefficient cannot be interpreted in the same manner as an OLS regression coefficient. When the relationship between the variables is nonlinear, there is no constant effect of x on y. (Bachman & Paternoster,1997:57m. 126 decisions for drug offenders. First, an overall logistic regression equation was estimated on the entire sample, including all three-time periods. Dummy variables were created and introduced into this overall regression equation to reflect different time periods. Overall, was gender significantly related to sentence outcome for data on decisions made during almost two decades? Next, logistic regression equations were estimated for each time period. In other words, three separate equations were specified using the same predictors and outcome measures. This permits one to look at possible gender effects and interaction effects with gender on sentence outcomes for drug (i.e., non-traditional female offenses) and property (i.e., traditional female offenses) offenders within each of the three time periods. For each time period, were women and men sentenced in a similar manner? Also, within the three time periods, logistic regression equations were estimated separately for men and for women offenders. Several studies in the literature suggest that while gender may not be a significant predictor overall of sentence outcome, when we look at what contributes to the sentencing decision for men and women separately, we find the factors differ. Thus, by separating men and women and estimating models for each, it could be determined if there are important differences, and further, what those differences entail. 127 Ordinapy Leas; Sgpares Regression Analysis In order to examine the dependent variable, “sentence length” (which is specified in months), ordinary least squares (OLS) regression was utilized (see 3.8).“7 By using OLS it was possible to find out whether two variables (an independent variable and a dependent variable) are linearly related to one another and to calculate the strength of the relationship (Menard, 1995). OLS is appropriate due to the fact that the dependent variable is continuous (i.e., sentence length in months) and an equation is being estimated using multiple independent variables (Bachman & Paternoster, 1997). Linear regression models are evaluated on several grounds: (Menard, 1995: 17) " In order to use Ordinary Least Squares (OLS) regression several assumptions must be met (Menard, 1995:4-5): (1)Measurement: All independent variables are interval, ratio, or dichotomous, and the dependent variable is continuous, unbounded, and measured on an interval or ratio scale. All variables are measured without error. (2)5pecification: (a)All relevant predictors of the dependent variable are included in the analysis, (b) no irrelevant predictors of the dependent variable are included in the analysis, and (c) the form of the relationship (allowing for transformations of dependent or independent variables) is linear. (3)8Xpected value of error: The expected value of the error is zero. (4)HOmoscedasticity: The variance of the error term, is the same, or constant, for all values of the independent variables. (5)Nbrmality of errors: The errors are normally distributed for each set of values of the independent variables. (6)Nb autocorrelation: There is no correlation among the error terms produced by different values of the independent variables. (7)No correlation between error terms and independent variables: The error terms are uncorrelated with the independent variables. an Absence of perfect multicollinearity: For multiple regression, none of the independent variables is a perfect linear combination of the other independent variables; mathematically. 128 (3) How well does the overall model work? Can we be confident that there is a relationship between all of the independent variables, taken together, and the dependent variable, above and beyond what we might expect as a coincidence, attributable to random variation in the sample we analyze? If there is a relationship, how strong is it? If the overall model works well, how important is each of the independent variables? Is the relationship between any of the variables attributable to random sample variation? If not, how much does each independent variable contribute to our ability to predict the dependent variable? Which variables are stronger or weaker, better or worse, predictors of the dependent variable? Does the form of the model appear correct? Do the assumptions of the model appear to be satisfied? Each of these various issues and questions can be appropriately answered in the analysis of sentence length. Recall that the analysis for the sentence length decision included only those offenders sentenced to prison. Sentence Length in months initially could be compared using mean number of months for men and women offenders who are H9 sentenced to state incarceration. Separate OLS equations were specified for each of the time periods to examine possible gender effects, as well as other important predictors of the sentence length.“8 Eggivalence of Regression Coeffieients Analysis Finally, in order to answer the question about whether the “war on drugs” campaign in this country has been a “war on women,” an analysis which included testing for the equality of regression coefficients between two independent equations was used (Brame, Paternoster, Mazerolle, & Piquero, 1998). This type of analysis allows one to look at the magnitude of coefficients between the independent and dependent variables across specific reference groups (e.g., women and men), as well as whether the relationship between two variables remains the same for different time periods. Brame et al. (1998) reviewed criminological studies testing for the significance of coefficient differences and found two statistical testing formulas were used most often. The authors, however, suggest that one of the equations (specifically the denominator portion) that is often used in the criminological literature produces a biased outcome. Specifically, the formula used by some criminologists produces an inaccurate standard deviation of the sampling distribution. According to Brame and his colleagues (1998), “ Due to the small numbers of women sentenced to prison, separate gender OLS models are not estimated. 130 the correct formula for this statistical test is the following:"9 br—bz 2:: x/SEbn’ + SE!»2 (3.9) The comparison of regression coefficients involves two steps (Brame et al, 1998: 247). First, one must estimate the difference between the two coefficients in the independent, randomly selected populations (i.e.,br—bz). Second, one has to estimate the standard deviation of the distribution of differences from repeated samples (i.e., VUHHWL+SEDX ). Using this statistical technique, the regression coefficients for both women and men were tested for equivalence across time periods. In other words, coefficients were tested for convicted women between time 1, time 2, and time 3. The same procedure was used for convicted men as well. Therefore, it was determined if the magnitude of regression coefficients had changed significantly over time for women and/or men convicted of drug offenses (Brame et al., 1998). ” For a more detailed discussion of this statistical formula and its appropriateness in using it to test for the equality of regression coefficients see the works of Brame, Paternoster, Mazerolle and Piquero (1998) and Paternoster, Brame, Mazerolle, and Piquero (1998). Brame et al. (1998) illustrates the problems behind using the statistical formula found in many criminological studies. Type I error occurs 5% of the time when using the appropriate formula, whereas Type I error for the inaccurate formula occurs 20% to 30% of the time. 131 it SE $6 at V .Pwt mu. Ppedieped Probability of Receiving a Prison Sentence In order to understand the impact that certain factors have on the likelihood of receiving a prison sentence, a series of predicted probabilities were computed and compared for specific sub-groups (e.g., gender, offense type, race). The analysis permits one to compare the probability of difference reference groups receiving a prison sentence within time periods as well as changes across time. Thus, one can look at the “war on drugs" issue using predicted probabilities to examine the impact of significant drug legislation had on the likelihood of offenders being sentenced to prison. Sample Selestion Bias Sample selection bias was a concern for this study, as it is a concern with most sentencing studies. Recall that two dependent variables were examined across the three time periods: an In/Out decision and sentence length. Sample selection bias occurs because data for the sentence length variable is contingent upon whether they were incarcerated at the first decision stage. In other words, those offenders not incarcerated for their offenses have missing data for the sentence length outcome. According to Winship and Mare (1992:328), “Sample selection is a generic problem in social research that arises when an investigator does not observe a random sample of a population of interest. 132 Specifically, when observations are selected so that they are not independent of the outcome variables in the study, this sample selection leads to biased inferences about social processes.” Thus, selection bias occurs when there is correlation between the error and the independent variables.50 The criminal justice system represents a series of stages where people are filtered out of the system at various decision points (i.e., arrest, charge, conviction, etc.). A problem occurs when cases that are processed out at earlier stages (e.g., case dismissal, etc.) have certain characteristics that remaining cases do not that reach the sentencing stage (D’Alessio & Stolzenberg, 1993).51 Several problems arise as a result of sample selection bias (Berk, 1983). Sample selection bias in the study might disguise the “true” relationship between gender and sentence outcome. First, if gender discrimination actually occurs at the sentencing stage, sample selection bias may hide the real nature or direction of this relationship (Klepper, Nagin & ” In the Hechman model the correlation occurs between the observed and unobserved factors influencing crime processing in the sampled population (Klepper et al., 1983: 68). ” Klepper et al., (1983: 64) offers the following example for illustrative purposes, prosecutors and judges may possess a great deal of qualitative evidence about a case that the investigator cannot observe from court records. In other instances, the investigator may not observe other, less qualitative types of evidence, such as whether the criminal used a weapon. The combination of screening and incomplete measurement implies that criminals reaching the later processing stages are not representative of the unobservable features of the population of cases entering the system. This introduces the possibility of sample selection bias. 133 QC 2‘ Tierney, 1983). Second, it could be more difficult to find gender effects in the analysis of sentence length because the women who remain are the most serious offenders in this group. Regardless of the potential problem, it is necessary to attempt to account for the sample selection bias so that researchers can confidently make reliable inferences from research conducted on criminal justice processing. Research from sociology and economics has developed ways to handle bias connected to sample selection. Klepper et al., (1983: 78) identify three strategies to respond to selection bias: (1) Measure the relevant factors well enough to eliminate the nonzero covariance between the disturbances in the selection and the regression equations. This will eliminate sample selection bias entirely. (2) The investigator can consider the imposition of an exclusion restriction on the model. (3) If none of the above approaches can be implemented satisfactorily, then the investigator can always resort to the bounding approach. While this approach does not yield a consistent estimate of the regression coefficient vector, it will indicate the potential magnitude of the selection bias. Further, regardless of whether the researcher uses the eXClusion or the bounding approach, he or she will only be ablfi: to take account of bias at processing stages in which datii are available. One will not be able to address sample 134 selection bias that has occurred at prior stages where data are not available on cases (e.g., arrest, conviction stages in this study), that have dropped out. Several correction measures are available to researchers”, including Heckman’s correction for sample selection bias. Heckman’s correction is widely cited and used throughout the literature.53 Heckman’s two step process will be used to analyze the sentence length model. The length of sentence is a function of two factors: (1) a linear combination of regressors and (2) a hazard rate reflecting the influence of the selection equation” (Berk & Ray, 1982: 369). The correction process involved the following steps. First, logistic regression was used to estimate the likelihood of an offender receiving a sentence of incarceration. For each case in the logistic model the predicted probability of exclusion from the sentence length sample was computed (i.e., Hazard rate).54 Second, the hazard rate was entered as an independent variable into the OLS model for sentence length. This procedure controls for the probability of each case not receiving incarceration (Sphon & Spears, 1997). ” See Winship and Mare (1992; Klepper, Nagin, and Tierney (1983); Berk and Ray (1982) for discussions of other available corrections methods. ” Heckman’s work is also the subject of several critical reviews, including that by Stolzenberg and Relles (1990). “ See Heckman (1979: 157) for a detailed explanation of the steps involved in calculating a Hazard rate. 135 W The study examined the role gender plays in the sentencing of drug offenders convicted in the state of Minnesota. In addition, the research explored the question of whether or not sentencing reforms such as the ones implemented in Minnesota disproportionately impacted women. This chapter presented a detailed research plan to study these issues. Finally, several limitations of the design were identified and addressed including how the study handled sample selection bias. The analysis of the data is described in the next chapter. 136 CHAPTER 4: ANALYSIS OF DATA As was indicated in Chapter 1, the purpose of the research was to increase our understanding of the impact of sentencing reforms on female offenders. The analysis of data examined these sentencing reforms in several ways. The analyses included (1) descriptive statistics of the overall sample and sub-samples for both sentence outcomes, (2) logistic and ordinary least squares regression models to determine the relevance of gender and other significant factors have on sentencing practices, and (3) a test for equivalence of coefficients to explore the “war on women” argument and look at the effects of sentencing reforms have on women more generally. First, a descriptive analysis of the sample and respective time periods is presented. Sample Characteristics Full Sample A total of 4,076 offenders spanning the three specified time periods comprised the entire sample. Table 6 details characteristics of the overall sample as well as for the individual time periods. Fifty-eight percent (N=2368) of B7 the offenders were sentenced in Hennepin County, whereas 42% (N=1708) of the offenders were sentenced in Ramsey County. Recall that the research plan involved examining sentencing practices for convicted drug offenders and property offenders (the latter used for control purposes). Just over three-quarters of the sample (77% or N=3159) were involved in property offenses, while the remaining 23% (N=917) were involved in drug offenses. The seriousness of the offense varied from a severity level of one to a severity level of eight. The average severity level for the overall sample was 3.08 (s.d.=1.49). The overall sample was predominantly male (72% or N=2919), almost evenly distributed along the specified racial groups, white (52% or N=2117) and nonwhite (48% or N=1959), and were likely to be unemployed (55% or N=2239 vs. 32% or N=1259 for employed) at the time of their offense. The average age of offenders in the sample at the time of their conviction was 28 years. The education level of the overall sample varies with some offenders having only a grade school or middle school education (10% or N=426), and many more having some amount of high school education (61% or N=2486).5'5 Just over half of the total sample (53% or N=2156) were single, while another 36% were married, ” Includes those who received a GED. 138 cohabiting, divorced, separated, or widowed (N=1465), and the rest had missing information for this measure (11% or N=455). As for the presence of dependent children, 46% (N=1874) of offenders in the sample had at least one dependent child, whereas 42% (N=1710) did not, and 12% (N=492) were unknown. Several legal factors for those in the sample are also detailed in Table 6. Fifty-five percent of the sample had some sort of prior conviction, whereas the remaining 45% did not. The distribution is markedly different depending upon the type of prior conviction, felony or misdemeanor. Only 12% (N=490) of the offenders in the overall sample had a prior misdemeanor conviction (88% or N=3586 had none), while 47% (N=1916) had a prior felony conviction (53% or N=2160 had none). Thirty-three percent (N=1330) of the defendants were under the custody of the criminal justice system at the time of their offense. Finally, 25% (N=1006) of the convicted offenders were sentenced to a state prison facility. And of those offenders sentenced to state prison, they received an average sentence of 23 months. 139 Table 6: Descriptive Information for the Overall Sample and Each Time Period Variable Overall Time 1 Time 2 Time 3 (N - 4076) (N:- 747) (N=- 1958) (N- 1371) # 8 # % fl % # % Female No 2919 72 508 68 1590 81 821 60** Yes 1157 28 239 32 368 19 550 40 Nonwhite No 2117 52 577 77 997 51 543 40" Yes 1959 48 170 23 961 49 828 60 Age mean 28 26 27 30** s.d. 8.27 7.94 7.85 8.53 range 16-71 16-70 16-71 16-71 Education Level 1-9 grades 426 10 95 13 237 12 94 7" 10-12 grades 2486 61 492 66 1350 69 644 47 college 604 15 124 17 224 11 256 19 Missing 560 14 36 S ?? 8 377 27 Single NO 1465 36 311 42 752 38 402 29 Yes 2156 53 402 54 1167 60 587 43 Missing 455 11 34 5 39 2 382 28 Dependents No 1710 42 383 51 1054 54 273 20** Yes 1874 46 324 43 860 44 690 50 Missing 492 12 40 5 44 2 408 30 Employment Status No 2239 55 415 56 1256 64 568 41** Yes part/full 1259 32 271 36 622 32 402 29 Missing 542 13 61 8 80 4 401 29 Hennepin County No 1708 42 280 38 764 39 664 48** Yes 2368 58 467 62 1194 61 707 52 Drug Offense No 3159 77 S77 77 1710 87 872 64** Yes 917 23 170 23 248 13 499 36 Severity Level mean 3.08 2.73 3.01 3.38" s.d. 1.49 1.25 1.37 1.70 range 1-8 1-7 1-7 1-8 Prior Misd. No 3586 88 666 89 1668 85 1252 91** Yes 490 12 81 11 290 15 119 9 Prior Felony No 2160 53 529 71 887 45 744 54*‘ Yes 1916 47 218 29 1071 55 627 46 Custody No 2746 67 629 82 1214 62 917 67** Yes 1330 33 139 18 744 38 454 33 Sentence OUT 3070 75 118 15 627 32 264 19" IN 1006 25 650 85 1331 68 1107 81 Sentence Length x --- 50.57: 23.86” 30.56° s.d. --- 36.40 10.33 23.29 range --— 13.00-240.00 12.10-84.00 12.10-146.00 ' Value is in number of months (N-llS). ” Value is in number of months (N-627). c Value is in number of months (N-264). * Significant at p g .05 ** Significant at p 5 .001 140 Ful m en r Data for all three-time periods were then separated by gender and descriptive information along with significant findings are provided in Table 7. There were significant relationships between gender and several other variables. For instance, there was a significant association between gender and race. The majority of men in the overall sample were white (54% vs. 46%), while the majority of women were nonwhite (53% vs. 47%). In addition women were significantly older than men by an average of three years (27 years of age vs. 30 years of age). As for the relationship between gender and several family related variables, men were significantly more likely to be single (57% vs. 42%) than women and significantly less likely to have dependent children (37% vs. 69%). Overall, there was a significant association between gender and education level, but none with employment status. Also, men were more likely to be convicted and sentenced in Hennepin County (61% vs. 39%) than in Ramsey County. Women, on the other hand, were equally likely to be convicted and sentenced in both counties (50% vs. 50%). 141 Table 7: Descriptive Information for Men and Women, Overall Sample Variable Overall (N = 4076) MEN WOMEN (Na 2919) (N8 1157) 8 % 8 % Nonwhite No 1570 54 547 47** Yes 1349 46 610 53 Age mean 27 30** s.d. 8.30 7.91 range 16-71 17-71 Education Level Grade 9 & below 321 11 105 9** High School 1835 63 651 56 College 369 13 235 20 Missing 394 13 166 14 Single No 950 33 515 45** Yes 1664 57 492 42 Missing 305 10 150 13 Dependents No 1490 51 220 19" Yes 1082 37 792 69 Missing 347 12 145 12 Employment Status No 1589 54 650 56 Yes (part/full) 933 32 362 31 Missing 397 14 145 13 Hennepin County No 1132 39 581 50** Yes 1787 61 576 50 Drug Offense No 2253 77 906 78 Yes 666 23 251 22 Severity Level mean 3.23 2.72** s.d. 1.52 1.33 range 1-8 1-8 Prior Misd. No 2533 87 1053 91** Yes 386 13 104 9 Prior Felony No 1353 46 807 70" Yes 1566 54 350 30 Custody No 1848 63 898 78** Yes 1071 37 259 22 Sentence OUT 2047 70 1023 88** IN 872 30 134 12 p 5 .05 .. p s .01 There was no significant difference between gender subgroups and the type of offense committed. However, several legal measures were related to gender. For instance, men on average committed more serious offenses than did women (3.23 vs. 2.72) and had a more extensive criminal history. Overall, 54% of men had at least one prior felony compared to 30% of women, and 13% of men had at least one prior misdemeanor compared to 9% of women. In addition, 37% of men and 22% of women were under some type of custody status with the criminal justice system (i.e., probation, parole, escapee). Finally, there was a significant difference between men and women and the initial sentencing decision. Overall, men were significantly more likely to receive a prison sentence compared to women (30% vs. 12%). Time Periods Since the analysis involved examining sentencing practices during each of the time periods, descriptive information is provided for each period presented in Table 6 along with significant associations. First, the percentage of women in the sample changes over time from 32% in Time 1 to 19% in Time 2 and 40% in Time 3, resulting in a 143 significant relationship between gender and time period. Changes in the numbers and percentages of women are due in part from offense patterns and criminal justice system responses, but also to the differing sampling strategies used by the Minnesota Sentencing Guidelines Commission for various time periods. The racial make up of the sample significantly changed over time as well. The percentage of nonwhites in the sample continued to increase from 23% in Time 1 to 49% in Time 2 finally to 60% in Time 3. The age of offenders in the three time periods averaged 26 years in Time 1, 27 years in Time 2 and 30 years in Time 3. As indicated in Table 6, there was a statistically significant difference in average age per time period with Time 3 being significantly different than both Time 1 and Time 2. Offenders’ educational level remained relatively constant across time periods, with the exception of a higher amount of missing data in Time 3.56 Offenders were often single (54% in Time 1; 60% in Time 2; 43% in Time 3) and had at least one dependent child. The distributions for both S"Missing information was an issue for four measures. Two strategies were incorporated to handle missing information. If data were missing for 10% or fewer of the cases, missing values were replaced by the mean (e.g., grand sample mean or sub-sample mean). However, if data were missing in more than 10% of the cases for a variable, missing values were recoded and replaced with a zero value. Additionally, a newly created variable reflecting the presence of missing data was introduced into the model in order to control for missing information. 144 Time 1 and Time 2 were similar to one another (43% and 44%), however, the figure increased in Time 3 with 60% of the offenders having had at least one dependent child. The type of offense, property or drug-related differed by time period. Drug offenses were committed by 23% (N=170) of the sample in Time 1, decreased to 13% (N=248) of the sample in Time 2, and then increased to 36% (N=499) of the sample in Time 3 after the proclaimed “war on drugs.” As for the seriousness of the offense, Table 6 provides information on the severity levels for each time period. Average severity levels increased from 2.73 at Time 1 to 3.01 at Time 2 to a high of 3.38 at Time 3. Therefore, the average severity level increased over subsequent time periods. A significant difference in severity levels was found when comparing average scores for each time period. This could be due to changes in sentencing legislation increasing the severity of certain offenses over time, especially drug-related offenses as a result of the “war on drugs” campaign. The distribution for prior convictions of offenders reveals some noteworthy differences. A significant relationship was found between time period and offenders having at least one prior misdemeanor (11% in Time 1; 15% in Time 2; 9% in Time 3). A large percentage of offenders had 145 at least one prior felony conviction and a significant relationship was found between time of sentencing and prior felony convictions. Twenty—nine percent of the sample had a prior felony conviction in Time 1, meanwhile that figure almost doubled (55%) in Time 2, and then decreased somewhat in Time 3 with 46% of the offenders reportedly having a prior felony conviction. Finally, Table 6 details descriptive information for both sentence outcome variables. The decision about whether to incarcerate the offender varied over time (15% in Time 1, 32% in Time 2, and 19% in Time 3) and was significantly related to the time in which the offender was sentenced. Figures for average sentence length were divided according to time of sentencing. Using the maximum sentence length value for Time 1 (i.e., indeterminate sentencing system), the average sentence length was almost 51 months (s.d.: 36.40 and range=13.00 to 240 months).57 The average sentence length for offenders sentenced during Time 2 and Time 3 was approximately 24 months and 31 months respectively.58 ” The current study followed the convention of using the maximum sentenced length under the indeterminate sentencing system, as was done by Stolzenberg and D'Alessio (1994) as well as Moore and Miethe (1986)in their own studies of the Minnesota Sentencing Guidelines. ” Sentence length for the overall sample was not computed due to problems with comparing sentence length in Time 1 with Times 2 and 3. Sentence length in Time 1 (indeterminate sentencing) is defined as the maximum sentence length, while the sentence length for Times 2 and 3 are 146 In conclusion, sample characteristics differ according to sentencing in one time period as opposed to another. A significant relationship was found between time of sentencing and each of the variables, with the exception of marital status. Differences in descriptive information across time may indicate the possibility that influential factors might vary and influence sentencing decisions as a result of different practices being used at different times. Gender Differences within Time Periods Descriptive information for each time period was compared for women and men and is presented in Table 8. Significant relationships between gender and other variables are noted. Several patterns of differences between men and women hold for each of the time periods. For example, women offenders on average, were 2 to 3 years older than men were in each time period. Women also were significantly less likely than men to have been single (never married) in each time period (Time 1 — 38% vs. 61%; Time 2 — 47% vs. 63%; Time 3 - 42% vs. 44%). Additionally, women were significantly more likely to have had at least one dependent child at the time of their offense in comparison to men presumptive sentences (no minimum and maximum values). Different standards of measurement at Time 1 and Times 2 and 3 make it inappropriate to compare figures across time. 147 St; 83? LL) ' 5 .t’ Q) t1”. (Time 1 - 69% vs. 31%; Time 2 - 73% vs. 37%; Time 3 — 65% vs. 40%). Significant differences were found also for employment status and gender for each time period. Men were significantly more likely to be employed either part-time or full-time for Time 1 (38% vs. 32%) and Time 3 (31% vs. 27%). Women, however, were significantly more likely to be employed either part-time or full-time for Time 2 (37% vs. 31%) . Descriptive statistics for legal factors, for the most part, indicated a more extensive criminal history for men than for women. Men were significantly more likely to have at least one prior misdemeanor when compared to women in Time 1 (13% vs. 7%) and Time 2 (16% vs. 10%), however no significant differences were found in Time 3 (8% vs. 9%). Likewise, significant differences were found between men and women and the occurrence of at least one prior felony conviction for each time period (34% vs. 19% in Time 1; 60% vs. 33% in Time 2; 54% vs. 33% in Time 3). As Table 7 illustrates, men were almost twice as likely as women during Time 1 and Time 2 to have at least one prior felony conviction. Again, men were significantly more likely than women to have custody status at the time of their offense for all time periods (20% vs. 13% in Time 1; 42% vs. 22% in 148 Time 2; 37% vs. 27% in Time 3). Clearly, men and women have considerably different criminal backgrounds. 149 an ass «m ass 4 «a 4 mm s as ea Hm magmas: 8m mes Hm mmm em ems AM was an >8 mm ems mos ..Hm omm mm mam .mm man we ones .46 «ms mm mmm oz unmeonmEm mm mms sm mum m m m mm mm s 8 mm magmas: mm mmm ow mmm ms mmm em amm mm mes Hm mma nu» ..ma mm mm can ..mm om as sea ..mm so me mam oz cmupaanu .pcmmoo vs ems om mew 4 os m mm m m 6 mm magmas: ms omm 44 8mm he mes mm mam mm om as man mo» .sm was mm mam ..Hm was GM mom ..om mes mm mos oz msmcam am omH on sea m an s was a m 6 am magmas: mm ova 4H was ms em as ops as as ms mm omoHHou me omm we 4mm mm osm on osHH so Hes mm Hmm NH-6H mosmuo m om m «m as me as ems ms mm «H mm m-H monmuo H0>0H COHUMUDUM os-mH as-ma H8-8H mm-ma mm-ma os-ma omens ..1H8.se mm Asm.me mm ..Lms.sv mm Amm.sc mm ..Aom.se mm Asm.se mm 1.6.me cmoz mmd um mam mm mam mm mmm 64 map mm 66 am «as was .ms mmm an mom ..sm mma 4m Hem .ms mes as 404 oz 0U fi£3GOZ s a s a s a w u s a s u zommuze cosoz Aammuze no: Ammmuze cosoz zommsuze so: .mmmuze cmsoz zmomuze so: stmH uze lemma uze lees uzv moanmaum> m msz n msz H msz Hopsmw >9 UOHHmm mEaB £08m new coHumEHOMcH o>aumwhommn "m canoe 150 AGOamflomU =zH: ..0.av demand Cu poocmucmm mmonu >Hco MOM numsua mmuocuas Hoo. v m um usmufiuwsmflm44 mo. v a no ucmuauadmam 4 law uze Anon nze Lam nze .msm uz. 1mm uze 1mm "2. oo.moa-oa.ma oo.mvauoa.mfi oo.mm-oa.ma oo.vmuoa.ma oo.oma-oo.ma oo.ovm-oo.ma umsmm «4AH.HNvmm.bN Am.mmvmm.dm «4A0.wv eo.mH Am.oav mm.vm Am.wmv mm.mm AH.mmv mo.mm A.O.mv Gov: Amnusos say unumsoa mucoucom HH Hm mm mom «a Hm mm mum m mm ma mm 2H 44mm owe me mam 44mm ram vm «Hoa 44am ham mm mHv Boo mucoucmm hm omH hm vom NN me me mwm ma om om NOH m0» yams 00¢ mm ham same man mm mmm «hm mom om wow 02 xpoumsu mm HmH em mvv mm MNH om mem ma we vm aha mm» %«hw mmm we mom «chm mvm oe mew asHm mmH mm mmm OZ chHmh HOaum m me m on 0H mm ma mmm 5 DH ma em mm? am How mm Hmb «om 0mm em mmma «mm NNN hm eve oz .pmaz Henna m-H m-H >-H 8-H m-a s-H omcmm «4Amm.av mo.m Abb.av Hm.m «4AHH.HV mv.m Ace.av wa.m «alum.v hm.m Amm.Hv mm.m A.p.mv com: Ho>oq >uano>mm mm NmH He bmm OH on ma NHN mm mm mm baa m0» «4dr mmm mm ewe om Nmm hm mend we mmH be Hmm oz omcmwwo msun me mmm hm mmv em mmH mm mmm om eeH em mmm m0» 44pm mam me mmm «we mma hm mmm ov mm mm mma oz >u8300 sammccom w n a a w a w a w a w # Aommuzv GOEOZ AHNmuzv GO: Ammmuzv GOEOS Aommfluzv :02 Ammmuzv GOEOZ Amomuzv GO: Adena uze Ammmfi uze Leas uzv moanmaum> m mzHH m mzHB H mzHB .p.ucoo .Hmpaow %Q poHHmm mash sumo new coHumEHOMGH m>aumauommo "m oHnt 151 Offense Type In order to look at possible relationships of importance within offense categories, the sample was divided into property offenders and drug offenders and descriptive and significant findings are presented in Tables 9 and 10. Interestingly, men and women were equally distributed across offense categories. Men comprised 71% (N=2253) of the property offenders and 73% (N: 666) of the drug offenders in the full sample; and women comprised 29% (N=906) of the property offenders and 27% of the drug offenders (N=251) in the full sample. Additionally, no significant differences were found between drug and property offenders in race or marital status. On the other hand, significant age differences were found, with drug offenders being slightly older than property offenders (29 yrs. vs. 28 yrs.). Education level, employment status, and the presence of at least one dependent child were all significantly related to the type of offense. For instance, drug offenders as compared to property offenders (49% vs. 45%) were significantly more likely to have had at least one dependent child and also be employed (33% vs. 31%) at the time of their offense. 152 Table 9: Descriptive Information by Offense Type Variables PROPERTY OFFENSES DRUG OFFENSES (n = 3159) (n=917 ) # % # % Female No 2253 71 666 73 Yes 906 29 251 27 Nonwhite No 1638 52 479 52 Yes 1521 48 438 48 Age mean 28 29** s.d. 8.14 8.56 range 16-71 16-69 Education Level Grades 1-9 349 11 77 8** Grades 10-12 1969 62 517 56 College 436 14 168 18 Missing 405 13 155 17 Single No 1130 36 335 36 Yes 1711 54 445 49 Missing 318 10 137 15 Dependents No 1386 44 324 35** Yes 1428 45 446 49 Missing 345 11 147 16 Employment Status No 1800 57 439 48* Yes part/full 992 31 303 33 Missing 367 12 17S 19 Hennepin County No 1253 40 455 50** Yes 1906 60 462 50 Severity Level Mean 3.08 3.08 s.d. 1.20 2.21 range 1-7 1—8 Prior Misdemeanor No 2735 87 851 93** Yes 424 13 66 7 Prior Felony NO 1595 50 565 62** Yes 1564 50 352 38 Custody No 2067 65 679 74** Yes 1092 35 238 26 Sentence OUT 2311 73 759 83** IN 848 27 , 158 17 .05 .001 * Significant at p ** Significant at p 153 Although no significant differences were found between the severity levels of property offenders and drug offenders, important associations were found with other legal variables. For example, drug offenders were significantly less likely to have criminal histories, both in terms of prior misdemeanors (7% vs. 13%) and prior felonies (38% vs. 50%), and were less likely to have custody status (e.g., on probation or parole, escapee) at the time of their offense (35% vs. 26%). Finally, those sentenced for drug offenses were significantly less likely than property offenders to receive a prison sentence (17% vs. 27%). Next, the data were split a second time based on gender. Descriptive information, as well as significant findings, is presented in Table 10. Recall that no significant differences were reported in Table 9 for race and offense type; however, when the sample was separated by gender, particular significant relationships were discovered. For property offenses, women were significantly more likely to be nonwhite as opposed to white (56% vs. 44%), whereas men were significantly more likely to be white as opposed to nonwhite (55% vs. 45%). In the case of drug offenses, men were equally likely to be white or nonwhite, 154 while women were significantly more likely to be white (58% vs. 42%). Another noteworthy difference relates to the severity levels for each offense type. For property offenses, men had significantly higher severity levels compared to women (3.28 vs. 2.60). Yet, for drug offenses there were no significant differences in severity levels. The results involving legal factors were similar across offense type for both men and women. For prior misdemeanors and prior felonies, men were significantly more likely than women were to have prior convictions. Fifteen percent of male property offenders (10% of female property offenders) had at least one prior misdemeanor and 9% of male drug offenders (4% of female drug offenders) had a prior misdemeanor. The figures for prior felonies for both property and drug offenders increased for both men and women. Fifty-seven percent of male property offenders (32% of female property offenders) had at least one prior felony conviction. Additionally, 43% of male drug offenders had a prior felony conviction as compared to 25% of female drug offenders. The likelihood of receiving a prison sentence was also related to gender despite differences in type of offense, and this difference was more sizeable for property offenses. Men were three times as likely as women to 155 receive a prison sentence for property offenses (33% vs. 11%). 92W In order to obtain summary information for the offenders in the study, descriptive information was obtained for all offenders, as well as a comparison of men and women on all variables of interest. Significant associations between time period and variables of interest, along with significant associations between gender and variables of interest, were analyzed and provided. In order to look at summary information within time periods, descriptive statistics were used and information was presented. Again, comparisons were made between women and men on all variables of interest and significant associations were noted. This provided a descriptive analysis of the data used in the study, and next the analysis involves examining bivariate relationships between measures of interest. 156 Table 10: Descriptive Information by Offense Type and Gender Variables PROPERTY OFFENSES DRUG OFFENSES (n 3159) (n=923 ) Males Females MEN FEMALE (n: 2253) (n- 906) (n- 666) (n: 251) 8 % 8 % 8 % 8 % Nonwhite NO 1237 55 401 44** 333 50 146 58* Yes 1016 45 505 56 333 50 105 42 Age mean 26.78 29.52** 28.87 30.87** s.d. 8.15 7.78 8.60 8.28 range 16-71 17-71 16-69 18-57 Education Level Grades 1-9 268 12 81 9** 53 8 24 10 Grades 10-12 1461 65 508 56 374 56 143 57 College 254 11 182 20 115 17 S3 21 Missing 270 12 135 15 124 19 31 12 Single No 722 59 408 45** 228 34 107 42 Yes 1334 32 377 42 330 50 115 46 Missing 197 9 121 13 108 16 29 12 Dependents No 1216 54 170 19** 274 41 50 20'* Yes 810 36 618 68 272 41 174 69 Missing 227 10 118 13 120 18 27 11 Employment Status No 1310 58 490 54 279 42 160 64** Yes part/full 696 31 296 33 237 36 66 26 Missing 247 11 120 13 150 22 25 10 Hennepin County No 814 36 467 51** 348 52 114 55 Yes 1439 64 439 49 318 48 137 45 Severity Level mean 3.28 2.60** 3.06 3.13 s.d. 1.24 .93 2.21 2.22 range 1-7 1-7 1-8 1-8 Prior Misdemeanor No 1924 85 811 90** 609 91 242 96* Yes 329 15 95 10 57 9 9 4 Prior Felony No 975 43 620 68** 378 57 187 75** Yes 1278 57 286 32 288 43 64 25 Custody No 1362 60 705 78** 486 73 193 77 Yes 891 40 201 22 180 27 58 23 Sentence OUT 1507 67 804 89** S40 81 219 87* IN 746 33 102 11 126 19 32 13 * Significant at p ** Significant at p MIA .05 .001 157 Bivariate Correlations between Variables of Interest In order to examine the strengths and directions of relationships between each of the variables, bivariate correlations were computed. Table 11 reports correlations between variables in the overall logistic model and the Time 3 logistic model. With few exceptions, most correlations between variables in Table 11 were weak. The incarceration outcome variable was moderately correlated with custody status (r=.435; pg.001) in the expected direction. Having custody status at the time of the offense (e.g., under some type of supervision with the criminal justice system such as probation status, parole, escapee) was significantly associated with the likelihood of receiving a prison sentence. The incarceration outcome variable was also moderately correlated with prior felonies, again in the expected direction (r=.472; pg.001). Offenders with a prior felony record were likely to receive a sentence of incarceration as opposed to an alternative sanction (e.g., probation). Two additional associations are worth noting in Table 11. First, custody status had a positive and moderately strong correlation with prior felonies (r=.583; pg.001). Those offenders on custody status at the time of their 158 offense were more likely to be sentenced to prison than those who did not have custody status. Finally, age at time of conviction was significantly correlated with marital status (r=-.445; pg.001). Single offenders were likely to be younger at the time of their conviction. Table 12 displays the correlations between each of the variables in the models for Time 1 and Time 2. Note the changes in correlations for the variables, marital status, dependent children, educational level, and employment status. In addition to the noteworthy correlations identified previously, several others appear to be of importance as well. Age at the time of conviction was moderately associated with both newly created marital status measures (r=-.503 and r=-.496; pg.001). Younger offenders were likely to single at both time periods (Marital_Ml and Marital_M2). Finally, marital status was moderately correlated with the dependent children measure (r=-.533 and r=-.425; pg.001). Consequently, single offenders were less likely to have dependent children. Conclusion The results from the bivariate correlation analysis highlight several significant relationships between the 159 variables of interest themselves, and with the outcome variables. This provides only part of the picture however. What is the relationship between variables of interest and the outcome variables once controls are introduced? Next, the analysis uses two types of multivariate statistics, logistic regression analysis and ordinary least squares regression analysis in order to address this issue. 160 Table Female (X1) Nonwhite (X2) Age (X3) Single-Z (X4) Dependts-Z(XS) Educ-Z (X6) Employ-Z (X7) Custody (X8) Prior-M (X9) Prior-F (X10) Drug (X11) Severity (X12) Hennepin (X13) Time-1 (X14) (X15) Time—2 Prison (Y1) * p s .05 11.: in Overall Model and Time 3 Model X1 .059** .139** -.l3l** .284** .041** -.007 -.l36*' -.O60** -.213*‘ -.012 -.155*' -.101** .038* .185" -.191** X2 .024 .028 .l39** -.090*‘ -.127** .022 .054** .037* -.003 .044** .026 -.240*' .176b -.037* X3 -.445** .229** -.008 .117** .147** .093** -.009 .011 -.100‘* .176b .035* 161 X4 -.243** .185** -.030 .002 -.053** -.073** -.047** -.033* -.030 .009 -.144** -.029 X5 .184** .O74" -.050** .042** -.021 .029 -.042** -.076** -.025 .062b -.036* X6 .232*’ -.066** -.015 -.068** -.005 -.042** -.037* .095** -.165** -.029 Bivariate Correlations for Variables X7 -.116** -.074** -.105** .015 -.025 -.042** .046** -.o37; -.127** Table 11: Bivariate Correlations for Variables in Overall Model and Time 3 Model, X8 X9 Female (X1) Nonwhite (X2) Age (X3) Single-Z (X4) Dependts-Z(XS) Educ-Z (X6) Employ-Z (X7) mstody (xa) Prior-M (X9) .155** Prior-F (X10) .583** .213** Drug (X11) —.077** -.080** Severity (X12) .025 -.026 Hennepin (X13) .074** .023 Time-1 (X14) -.151** -.017 Time-2 (X15) .007 -.073** Prison (Y1) .435** .170** * p g .05 ** p g .001 X10 -.093* .056** .117** -.169* -.018 .472** X11 .000 -.084** .003 .237** -.093** 162 X12 .116** -.114** .141** .197** cont’d. X13 X14 .042** -.094** -.337** .084** -.102** X15 -.090** Table Female (X1) Nonwhite (X2) Age (X3) Single-M1 Single-M2 (X4) Dependts-Ml Dependts-M2 (X5) Educ-M1 Educ-M2 (X6) Employ-M1 Employ-M2 (X7) Custody (X8) Prior-M (X9) Prior-F (X10) Drug (x11) Severity (X12) County (X13) Prison (Y1) 212 : Bivariate Correlations for Variables in Time 1 and Time 2 Models X1 .059** .139" -.24S** -.124‘* .347** .286** -.003 .036 -.082* .053* -.l36** -.060** -.213** -.012 -.154*‘ -.101** -.191** **p IA X2 .024 -.24S** -.124** .155""‘r .154** -.086* -.016 -.037 -.116** .022 .054** .037* -.003 .044** .026 -.o37* .001 X3 -.503** .496** .337** .240** .084' .102** .162** .097** -.008 .117** .147" .093** -.009 .011 .035* 163 X4 .533** .425** -.045 .067** -.060 .143** .079* .004 .080** .071** .126** .081** .024 .080'* .111** .085** -.059 .012 .013 -.052* X5 -.038 .004 .016 .046* -.055 -.031 .086* .054* .049 -.001 -.081* .015 -.046** -.094** -.034 -.036 .063 -.007 X6 .132** .157** -.065 -.009 -.065 -.050* -.045 .013 .102** .075** -.006 -.070** -.087* .006 -.090* -.008 X7 -.054 -.119‘* -.024 -.O97** -.042 -.llS** .103** .066" -.018 —.048* -.059** -.030 -.088* -.137** Table 12: in Time 1 and Time 2 Models, X8 X9 Female (X1) Nonwhite (X2) Age (X3) Single-M1 Single—M2 (X4) Dependts-Ml Dependts-M2 (X5) Educ-M1 Educ-M2 (X6) Employ-M1 Employ-M2 (X7) Custody (X8) Prior-M (X9) .155** Prior-F (X10) .583** .213** Drug (X11) -.O77** -.080** Severity (X12) .025 -.026 County (X13) .074** .023 Prison (Y1) .435** .170** X10 X11 -.093*' .056** .000 .1l7** -.084** .472** -.093** 164 cont’d. X12 .116** .l97** Bivariate Correlations for Variables X13 .084** Analysis of Sentence Outcome: In vs. Out Full Model The decision whether or not to incarcerate convicted offenders was examined using a series of logistic regression analyses. First, logistic regression was used to estimate a model for the In/Out sentencing outcome decision for the entire sample (i.e., all three-time periods). The results are provided in Table 13, along with estimates for separate gender models. The full model had a significant chi-square value of 1467.54 (pg.001; df=19) and an Rf'value of .30249 for the overall sample (N=4076). Missing information was a concern for several variables included in the analysis. Two strategies were used in handling missing information. First, if data were missing for 10% or fewer of the cases, missing values were replaced by the mean for that unit of analysis (e.g., overall sample mean or sub-sample mean). Second, if data were missing in more than 10% of the cases for a variable, missing values were recoded to a value of zero. In addition, a new variable was created to reflect whether or not the case contained missing information in the original measure, and ‘9 Cox and Snell R2 value. 165 was then introduced into the model to control for missing information (see for example, DeJong and Jackson, 1998; DeJong, 1997). In the full model, gender was a significant predictor of the In/Out sentencing decision. Controlling for all other factors in the model, women were more likely to receive an “Out” sentencing decision (b=-.540; pg.001). Other significant predictors in the full model included education (b=.063; pg.05), employment status (b=-.562; pg.001), and race (b=-.389; pg.001), although race was not related in the expected direction. Therefore, nonwhites were more likely to receive a sentence other than imprisonment (i.e., “Out”) as compared to white offenders. Note that, as was found in previous research, having a dependent child was not a significant factor in the overall model for the imprisonment decision. 166 Table 13: Logistic Regression Results for Full Model and Gender Models Full Model Male Model Fema Mo Variables B S.E. ODDS B .E. ODDS B .E. ODDS RATIO RATIO RATIO Female -.540 .125 .582** -- -- -- -- -- -- Nonwhite -.389 096 .678** -.326 .107 .722* -.651 .240 .522* Age .006 006 1.01 .006 .007 1.01 .002 .016 1.00 Educ (Z) .063 095 1.07; .086 .107 1.09 .012 .219 1.01 Employ (Z) -.562 .111 .570** -.546 .122 .580** -.537 .278 .585* Single (Z) -.118 .119 .889 - 079 .134 .924 -.147 .276 .863 Dependents (Z) 192 .112 1.21 .248 .124 1.28* -.081 .286 .922 Hennepin County .028 .096 1.03 -.005 .107 .995 .135 .229 1.15 Time Period (1) 1978 -.234 .140 .791 -.262 .146 .490** - 130 .325 .878 (2) 1994 -.752 .128 .472** -.713 .375 .018** -.793 .285 .453* Drug Offense -.183 .124 .833 -.259 .140 .771 .117 .292 1.12 custody 1.20 .103 3.33** 1.22 .113 3.39** 1.03 .256 2.81** prior Misd. .495 .122 1.64** 477 .136 1.61** .695 .296 2.00* Prior Felony 1.96 .125 7.07H 2.01 .144 7.49** 1.80 .267 6.02** Severity Level .408 .032 1.51** .395 .035 1.49** .440 .077 1.55** Constant -3.92 .355 .020** -4.00 .375 .018** -4.01 .800 .018** Educ_DM -.403 .269 .668 -.347 .296 .707 -.806 .692 .447 Employ_DM .134 .194 1.14 -.024 .212 .976 1.08 .497 2.95* Marital_DM .215 .279 1.24 .366 .322 1.44 - 172 .618 .842 Dependents_DM .512 .256 1.67* -.024 .212 .976 .399 .639 1.49 -2 LOG LIKELIHOOD: 3087.84 2474.52 598.36 CHI SQUARE MODEL 1467.54** 1085.41** 231.23** R’(Cox & Snell) .302 .311 .181 DF: 19 18 18 N: 4076 2919 1157 pg .05 ** p 5 .001 z denotes the use of the value zero for missing cases. DM denotes the use of a dummy variable to reflect the presence of missing data. 167 The findings in Table 13 also indicate that each of the legal factors significantly predicted the sentence decision, as expected, in a positive direction. For instance, having custody status with the criminal justice system at the time of the offense (b=l.20; pg.001), as well as a prior misdemeanor (b=.495; pg.001), or a prior felony (b=1.96; pg.001) increased the likelihood of being sentenced to prison. Additionally, a higher severity level (b=.408; pg.001) led to a higher likelihood of receiving a prison sentence when holding all other factors in the model constant. The county in which the sentencing occurred and offense type were not key factors in the sentence outcome. Time period indicated mixed results. No significant effect was found for Time 1; however Time 3 (b=-.752; p$.001) was a significant predictor of the In/Out decision. Specifically, those sentenced during Time 2 were more likely to receive a prison sentence than those sentenced at Time 3 in the full model. The entire sample was next divided based on gender, and separate logistic regression models were specified for each group (see Table 13 for results). This step permits one to explore whether different factors were influential in the sentencing decisions for men and women. Both gender models 168 were statistically significant; however, they differed with regard to their explained variance, with the model for men doing a more efficient job (R3=.311 for men and.Rfl=.181 for women). There might be other gender-related factors not considered in the model that would explain the lower explained variance for women (e.g., role in the crime, use of weapon). Many of the same factors significant in the full model were significant in the separate gender models. For instance, both race (b=-.326; pg.05 for men and b=-.651; pg.05 for women) and employment status (b=-.546; pg.OOl for men and b=-.537; pg.05 for women) were significant predictors of sentence outcome in both gender models. As was the case with the full model, legal factors such as custody status, prior misdemeanors, prior convictions, and severity level all significantly influenced, in the expected direction, the decision whether or not to incarcerate the convicted offender. Prior felony was highly influential in the full model for incarceration decisions, and the same holds for both men and women when examined using separate models. There were few differences between the separate gender logistic models. For example, having at least one dependent child for men was significantly related to a decision of incarceration (b=.248; pg.05), whereas having at least one 169 dependent child for women was related to a sentence decision of something other than incarceration, albeit this was not statistically significant. Finally, there were slight differences in the effects of time period on the In/Out decision. In the case of both men and women, being sentenced during Time 3 compared to being sentenced in Time 2 resulted in a greater likelihood of receiving an alternative to prison (b=—.713; pg.001 for men and b=-793; pg.05 for women). In other words, controlling for all other factors in the model, men and women were more likely to be incarcerated if sentenced during Time 2, immediately after the implementation of the sentencing guidelines as opposed to Time 3 under more recent guidelines. Time 1 had a significant effect only for men. Men sentenced during Time 1 as compared to Time 2, again were more likely to receive an “Out” sentencing decision (b=-.262; pg.001). Therefore, controlling for other factors in the model, men were more likely to be incarcerated at Time 2 in comparison with either Time 1 or Time 3 and women were more likely to be incarcerated at Time 2 compared to Time 3, but not Time 1. The findings suggest that sentencing at Time 2 was more punitive in terms of the likelihood of incarceration when compared to sentencing at Time 1 or Time 3. 170 Time 1 Analysis The overall sample was next divided by time period and also by gender. Results for Time 1 are presented in Table 14, first for the overall Time 1 model followed by separate models for men and women. The Time 1 model had a —2 Log Likelihood value of 505.12 and was statistically significant (chi—square model: 136.55; pg.001), with an R3‘value of .167. The goodness of fit measures for the gender models were significant as well (see Table 14 for details). As expected, under the pre—guidelines sentencing system in Time 1, gender had a significant effect on the incarceration decision. Women (b=-.689; pg.05) were significantly more likely, as compared to men, to be sentenced to a sanction other than prison (i.e., “Out”), thereby lending some support to Hypothesis—1. Race (b=-.603; pg.05) and the presence of at least one dependent child (b=1.05; pg.001), likewise, were significant predictors for the initial sentencing decision. However, for race the effects were in the opposite direction than was expected. The results suggest that controlling for all other factors in the model, nonwhite offenders, as compared ‘to white offenders, were more likely to receive an .alternative to incarceration. Thus, white offenders were 171 sentenced more punitively than their nonwhite counterparts. The same results were not supported in the gender models. In both cases race did not have a significant impact on the sentence outcome. Additionally, offenders with at least one dependent child tended to be incarcerated for their crime. When this factor was examined for each gender model, it was found to have a significant effect on the In/Out decision for men (b= 1.19; pg.001), but not for women. Therefore, having dependent children does not appear to produce leniency for women under the indeterminate sentencing system in Minnesota (Time 1), and as such does not provide support for Hypothesis-1. The remaining extra-legal variables did not have a significant impact on the outcome variable. 172 Table 14: Logistic Regression Results for Time 1 Model and Gender Models M denotes mean replacement for missing values. 173 Time 1 Male Model Femele Model Variables s .E. ODDS B 8.8. ODDS B .E. ODDS RATIO RATIO RATIO Female -.689 .301 .502* -- -- -— -- -- —- Nonwhite -.603 .294 .547* -.503 .342 .605 - 901 .639 406 Age -.011 .019 .989 -.018 .021 .983 .023 .044 1.02 Educ (M) -.332 .214 .718 -.312 .247 .732 -.457 .468 .633 Employ (M) - 466 .266 .627 -.345 296 .709 - 938 .695 391 Single (M) .547 .318 1.73 .612 .391 1.84 .723 .658 2.06 Dependents (M) 1.05 .308 2.86** 1.19 .366 3.28** .597 .634 1.82 Hennepin Co. - 029 .241 .972 -.118 .274 .889 .291 .532 1.34 Drug offense -.548 335 .578 -.705 .395 .494 -.290 .647 .749 Custody 1.00 .261 2.73** .924 .299 2.52* 1.22 .577 3.39* Prior Misd. .040 .331 1.04 -.043 .383 .958 .366 .731 1.44 Prior Felony 1.66 .267 5.27H 1.74 .314 5.71** 1.54 .550 4.68' Severity Level .011 .096 1.01 .025 .103 1.03 -.109 .278 897 Constant -2.02 .777 .133* -2.02 .884 .133 -2.96 .92 .052 -2 LOG LIKELIHOOD: 505.12 380.73 120.71 CHI SQUARE MODEL: 136.55** 102.90** 26.16* R2 (Cox & Snell): .167 .183 .104 DF: 13 12 12 747 508 239 [35.05 ** p 3 .001 Two legal factors were found to have significant effects on the incarceration decision. Specifically, one’s custody status at the time of the offense was significantly related to the likelihood of incarceration (b=1.01; pg.001). Prior felonies were also an influential predictor of sentence outcomes (b=1.66; pg.001). Having at least one prior felony resulted in a higher probability of being incarcerated for the offense. The results remain the same in the separate models for men (b=1.74; p$.001) and women (b=l.54; pg.05). In order to examine possible differences for women and/or men, several interactions were introduced into the model at Time 1. The findings for these interactions are provided in Table 15 and as shown there, none of the hypothesized interactions had significant effects on the incarceration decision at Time 1. The final model for the analysis at Time 1 is presented in Table 20. 174 Table 15: Logistic Regression Results for Time 1 Model with Interactions Effects Variables B S.E. ODDS RATIO Female -.376 .586 .687 Nonwhite -.514 .340 .598 Age -.012 .019 .988 Education (M) -.342 .215 .711 Employ (M) -.447 .268 .640 Single (M) .603 .325 1.83 Dependents (M) 1.16 .340 3.19 Hennepin County -.026 .242 .974 Drug Offense -.688 .393 .502 Custody 1.01 .262 2.75** Prior Misd. .060 .333 1.06 Prior Felony 1.66 .270 5.26** Severity Level .010 .096 1.01 Constant -2.06 .782 .127* Race * Sex -.301 .670 .740 Dependents * Sex -.437 .658 .646 Drug * Sex .476 .736 1.61 -2 LOG LIKELIHOOD: 503.82 CHI SQUARE MODEL 137.85** R2 (Cox & Snell) .169 DF: 16 N: 747 5 .05 ** p g .001 M denotes mean replacement for missing cases. 175 Results from the analyses performed for Time 1 suggest limited support for the first hypothesis. Gender did have a significant effect on the initial sentencing decision; however, other factors related to traditional gender roles did not. For example, both marital status and offense type did not have a significant impact on the In/Out decision. In addition, the significant effect of dependent children was in the opposite direction than expected. There was also no support for suspected interactions between gender and key measures such as race, dependent children, and offense type. All of those interactions were tested and failed to reach significant levels at Time 1. In conclusion, Hypothesis-1 had limited support from the analyses performed for Time 1. Time 2 Analysis The next stage of analysis involved estimating logistic regression equations for the In/Out sentencing decision at Time 2. The results are presented in Table 16. The Time 2 model had a —2 Log Likelihood value of 1569.35 and was significant (chi—square model: 886.14; pg.OOl) with an R?mq euaum>om ..oa.o omm. oa.~ ..me.o mom. eH.~ ..em.m eon. oo.a ..eo.e mas. om.a xeoaos scans .eo.H mom. emo. ..ee.H ooa. mom. oo.H Hem. ooo. ..vo.H «NH. moo. .onaz scans .oo.~ Hos. oeo ..ma.o Goa. No.a ..me.~ Hon. oo.a ..mm.m mos. om.a encuuso oee. oom. mom.- .mom. mum. oem.- oem. mmm. oom.- moo. «NH. moH.- omeoouo mono -- -- -- -- -- -- -- -- -- «.meo. oma. mme.- somaxmo -- -- -- -- -- -- -- -- -- Hoe. ova. omm.- oemaxao oofluwm mEfie mm.H ooa. umm. mmo. mma. ema.- mem. Hem. mmo.- mo.H moo. omo. encsoo chooses: -- -- -- oH.H ova. ova. ..oo.m mom. mo.a 12o nosmoeodmo Ho.H «om. oHo. -- -- -- -- -- -- Hm.a was. mmfl. Luv mnemonmdoo -- -- -- moo. ooa. eHH.- me.H mam. evm. is. mameam oem. Hmm. Hmo.- -- -- -- -- -- -- moo. mas. was.. fine mameam -- -- -- ..mmm. ova. omm.- emu. mom. moo.- 1:. e035 .ooo. omm. Ham - -- -- -- -- -- -- ..oem. AHA. mom.- LNG eoHdsm -- -- -- ma.a ems. mas. oae. saw. mmm.- -- -- -- 12o Doom .mo.a ooa. «em. -- -- -- -- -- -- .eo.a moo. moo. luv Odom mom. HHo. Hao.- .mo.a moo. mmo. mom. mHo. HHo.- ao.a woo. woo. mm< eem. mos. moo - ..moo. mma. mom.- .eom. son. moo.- ..oeo. omo. mom.- onarzcoz .oeo. was. mom.- mom. mom. mam.- .mom. son. moo.- ..mom. mma. oom.- masses oHe 30: Mo mm: on» amuocoo ED .mommo mcwmmwe uOu oumn ma~m> may mo mm: on» mouocmo N .mommo meannee uOu unwEoomHQou some neuocoo z H00. W Q «t mo. W Q « Hema mums eve ueoo ”2 ea «a ma ma "no oum. eum. eua. mom. ”iaameu u xooc mm ..mm.~Ho ..oa.vmu .4um.uma ..ou.euva namoo: mmQ mosmbsmm somHHm m mse>flmomm MOM mmfluflaenmnoum pouompwum ”om mHDsH 220 Time 3 Analyeis At Time 3, the results indicate that predicted probabilities of receiving a prison sentence decreased but not as low as the levels at Time 1, with the exception of white male property offenders (Time 1 — .155 vs. Time 3 - .124). The white male property offender had the highest predicted probability or receiving a prison sentence (.124), followed closely by the nonwhite male property offender (.121), the white male drug offender (.098), and the nonwhite male drug offender (.096). The next highest predicted probability involved the white female property offender (.087), closely followed by the nonwhite female property offender (.085), the white female drug offender (.069), and finally the nonwhite female drug offender (.067). 2289128198 The information presented in Table 30 reveals several noteworthy findings. First, the predicted probabilities for being sentenced to prison increases dramatically for all reference groups between Time 1 and Time 2. The implementation of a determinate-based sentencing system 221 between Time 1 and Time 2 appears to have increased the likelihood of incarceration for each of the references groups. Second, all reference groups' predicted probabilities for prison decreased between Time 2 and Time 3, except for nonwhite female drug offenders. This general decrease in the predicted probability of receiving a prison sentence could be the result of any number of factors, including changes in probabilities of arrests and restricted prison resources such as available bed space, or returning to “business as usual” after the initial effect of new sentencing policies. The consistent increase in the predicted probability for nonwhite female drug offenders lends some support to Bush-Baskette’s contention that the “war on drugs” has been specifically a “war on Black women.” The trend across three time periods appears to be different for nonwhite drug offenders. Further research should address this issue particularly in light of the fact that nonwhite offenders were sentenced more leniently at each of the three time periods, yet the probability of going to prison for specific nonwhite subgroups increased over time. Additionally, future research should examine why the predicted probability for nonwhite female drug offenders continued to increase over 222 the three time periods. Finally, the predicted probabilities at Time 3 are closer in range for the reference groups (.124 to .067) as compared to those at the other two time periods (Time 1 — .155 to .028 and Time 2 - .286 to .061). Thus there is less variation in the likelihood of going to prison between the reference groups at Time 3. Note that the predicted probability for white male property offenders was the only figure to decrease at Time 3 below the value at Time 1, while the value for nonwhite female drug offenders increased. The smaller variation between subgroups may result from further attempts by the state to restrict the influences of various extra-legal factors through additional sentencing changes between Time 2 and Time 3. Next, the analysis considers the issues of whether the “war on drugs” has been a “war on women.” Analysis of Equivalence of Regression Coefficients In order to explore the argument that the “war on drugs” has been a “war on women,” a test for comparing two regression coefficients was employed. Borrowing from the work of Brame et al.(l998) a two step procedure was used. Recall equation 3.9: 223 br-bz z = 2 JSEbn’ + SEbz Table 31 provides the results for the testing of equivalence of coefficients of drug offenses for men and women within each time period. Table 31: Equivalence of Regression Coefficients for Men and Women Drug Offenders within Time Periods Coefficients: Men vs. Women Z value Full Model —1.16 Time One -.548 Time Two -.270 Time Three -.530 As indicated in Table 31, no significant differences of magnitude were found between the coefficients for male and female drug offenders for each time period. These findings suggest that the magnitude of coefficients between drug offenses and the decision whether to incarcerate is not significantly different for male and female drug offenders. Next, coefficients were compared for each gender across time periods and changes in sentencing systems (determinate 224 vs. indeterminate; pre and post “war on drugs”). Table 32 provides the results of each test for both genders. Table 32: Equivalence of Regression Coefficients for Men and Women Drug Offenders Across Time Periods Coefficients: Men Z value Coefficients: Women Z value Time 1 v. Time 2 -.276 Time 1 v. Time 2 .074 Time 1 v. Time 3 -.833 Time 1 v. Time 3 -.295 Time 2 v. Time 3 -.767 Time 2 v. Time 3 -.349 Coefficients were compared separately for men and women between the various time periods. Using Chesney-Lind’s argument that the “war on drugs” has been a “war on women,” there should have been a significant difference in the magnitude of coefficients for women when comparing Time 1 with Time 3, and to a lesser extent when comparing Time 2 and Time 3. Again, the findings suggested no significant differences in magnitude of coefficients between comparison groups for drug offenses over time. Thus, the magnitude of the coefficient reflecting the relationship between drug offenses and likelihood of incarceration has not significantly changed over time, moving from an indeterminate system to a determinate one or before and after significant sentencing changes for drug offenses. Chesney—Lind’s argument that the “war on drugs” has 225 been a “war on women” is not supported with these findings. The magnitude of coefficients for drug offense and sentence outcome did not change significantly over time. These results suggest that women have not been disproportionately influenced by sentencing reforms in general, and the “war on drugs” more specifically, in the state of Minnesota. Thus, Hypothesis-5 was not supported from the findings testing the equivalence of coefficients. Qgrlczlnsign The purpose of this study was to ascertain the impact of gender on sentencing outcomes for drug offenders in the state of Minnesota before and after the implementation of sentencing guidelines in May 1980. A series of hypotheses were tested in order to explore the relationship between gender, sentencing reforms, and sentencing decisions. The findings indicated mixed support for the hypotheses. Legal factors were consistently strong predictors of sentencing decisions over the span of the study. With the exception of Time 1, prior felonies and the severity of the current offense were significant in each of the models. As expected, controlling for all other variables in the analysis, gender was a significant predictor of the initial 226 sentencing outcome at Time 1 before sentencing reforms. However, even after the implementation of sentencing guidelines in the state, gender remained influential in sentencing decisions at Time 3 and in combination with race at Time 2. Three sets of interactions were tested in the overall model and at each time period: sex*race, sex*drug offense, and sex*dependent children. Only one interaction was found to be significant, that being between sex and race at Time 2. Nonwhite women were significantly more likely than white women to be sentenced to an alternative to prison. Women in general received leniency from the courts, and this did not result from certain statuses linked to race or gender roles. Findings for the second sentencing decision, sentence length, provided mixed support for hypotheses. As expected at Time 1 and Time 3, gender was not an important predictor in determining the sentence length of those offenders sentenced to prison. However, at Time 2 gender was a significant predictor of the sentence outcome, with women receiving a shorter sentence length. The last chapter provides a brief summary of important findings, along with a discussion of implications for theory, policy and future direction of research on sentencing disparity. 227 CHAPTER FIVE: DISCUSSION AND CONCLUSIONS The final chapter has several objectives. First, a brief overview of the research study and results are discussed for each of the tested hypotheses. Second, limitations of the research study and findings will be discussed in detail. Finally, implications for theory in this area of scholarship will be addressed, and implications for policy reviewed. Summary and Discussion Decieion to Incarcerate The first part of the analysis plan examined the initial sentencing decision, whether to incarcerate the convicted offender. Sentencing decisions across the three periods of time were analyzed together and for each gender group. Many of the same predictors were influential in the full model as well as the respective gender models. When sentencing decisions from the three time periods were considered together, female offenders were significantly more likely to receive an alternative to incarceration when compared to male offenders, controlling for all other 2% predictors in the analysis. Additionally, nonwhite offenders were significantly more likely to receive an alternative to incarceration as well. Time period was also an important factor in determining sentencing outcomes. No significant differences were found in the likelihood of receiving prison sentences during Time 1 compared to Time 2 controlling for all other factors in the model, however significant differences were found between Time 2 and Time 3. Offenders at Time 3 were significantly more likely to receive an alternative to prison when compared to similar offenders who were sentenced at Time 2. Finally, each of the legal variables were significant predictors of the initial sentencing decision in the full model as well as the male and female models, all in the expected direction. For example, having a prior felony or committing a more serious offense resulted in a significantly higher likelihood of receiving a prison sentence. The analysis next considered sentencing decisions for each of the time periods. First, it was hypothesized that gender differences would be found during Time 1 for the In/Out sentencing decision, when sentencing officials had broad levels of discretion, but not during Time 2 or Time 3 after Minnesota developed and implemented sentencing reforms. As expected, gender significantly affected who 229 received a prison sentence during the first time period. At Times 2 and 3 gender should not have had a significant impact on which offenders received prison sentences, however gender did remain a significant predictor of sentencing decisions at Time 3, and contributed to a significant interaction effect at Time 2 with race. Despite reform measures to limit the influence of extra-legal factors such as gender or race on sentencing decisions, they remained influential predictors of the initial sentencing decision even after controlling for key legal variables (e.g., prior felony, custody, severity). Also, it was believed that results would reflect differences among women, for instance due to whether or not they had dependent children. Based on the review of the literature, three particular interactions were of interest: gender and race, gender and dependents, and gender and type of offense. The findings show limited support for this perspective. One could presume that it would be more likely to discover significant interactions at Time 1. However, no significant interaction effects were found for the decisions at Time 1. The one significant interaction that was present involved that between gender and race at Time 2, albeit in the opposite direction of what was expected. White women were significantly more likely to be incarcerated than 230 nonwhite women. The analysis next considered the sentence length decision. Sentence Length Deeieion The sentence length decision was analyzed for those offenders sentenced to prison. It was hypothesized that gender would not be a significant factor in the sentence length decision at each of the time periods. As expected, gender was not an important predictor at either Time 1 or Time 3. Unexpectedly, there was a significant finding at Time 2. Women who were sentenced to prison were significantly more likely to receive a shorter sentence length (by about 3 M months) than men who were sentenced to prison at Time 2. The sentencing county was also a significant predictor in the sentence length decision. For sentence length decisions at both Time 2 and Time 3, the findings suggest that offenders sentenced to prison in Hennepin county as compared to Ramsey county were significantly more likely to receive shorter sentences, by about 2 months at Time 2 and 7 months at Time 3. This result is important considering the fact that this difference in sentence length occurred under a determinate sentencing system that had been implemented 231 state-wide. One would not expect to find such county differences. Further research might consider certain contextual differences among counties in Minnesota and how they may have, in turn, influenced the implementation or application of such reforms. Predicted Probabilitiee of Receiving a Prison Sentence Next, predicted probabilities were calculated for the decision outcome of incarceration in order to determine the likelihood of certain types of offenders receiving a prison sentence. It was hypothesized that the introduction of sentencing reforms and significant drug legislation would have had a disproportionate impact on women over time. The results for this part of the analysis suggest some notable trends. For each time period, white males had the highest probability of receiving a prison sentence followed by white male drug offenders in Time 1 and Time 2, and nonwhite male property offenders in Time 3. White women property offenders had the highest probability of being incarcerated for each of the time periods, followed by white female drug offenders in Time 1 and Time 2, and nonwhite female property offenders in Time 3. Although nonwhite 232 female drug offenders had the lowest predicted probability for receiving a prison sentence for each of the time periods, it is the only predicted probability that continued to increase over the three time periods. These findings suggest that the reforms may be disproportionately impacting certain subgroups of offenders and further research should try to identify and understand this impact. Seneencing Reforms and the “War en ngen” Finally, it was hypothesized that recent sentencing reforms implemented in the state of Minnesota disproportionately impacted women in comparison to men. Some scholars have maintained that in a rush to respond to the drug problems of this country, policymakers implemented sentencing reforms, which have been excessively punitive toward women drug offenders. Instead of looking at possible alternatives to incarceration, community-based programs that are family friendly or other treatment options that recognize the specific circumstances of women drug offenders, officials have adopted determinate measures. Mandatory minimums or sentencing guidelines developed in many jurisdictions around this country have chosen to incarcerate women, and many of 233 them are non-violent. These policies have placed additional burdens on communities and governments to either find care for dependent children, or assist maternal grandparents with public monies. Although one could argue that these sentencing reforms have incarcerated many male and female non-violent offenders alike, the effect on women and children and society, more generally, is substantial and worthy of examination. An analysis of equivalence of coefficients was used to address this question. As shown, results from comparing the effects of gender on sentencing practices for each time period, along with comparing the effects within each gender group across time periods resulted in no significant differences. Furthermore, in all time periods, women were less likely to be sentenced to prison for drug offenses than men. The exception is at Time 2, when this was true for nonwhite women only. Thus, the argument that sentencing reforms and the “war on drugs” have had a disproportionate impact on women was not substantiated by this research. Sentencing reforms have impacted various states and the federal system in specific ways, thus additional research should examine other sentencing systems to ascertain whether the results documented here are replicated or different findings are found. 234 Limitations of the Present Research Several limitations of the present study need to be addressed. First, the data are limited to the extent that only two relevant decision points were examined; the decision to imprison a convicted offender or use a community alternative (In vs. Out). Consequently, this research did not deal with earlier decision points. Differential treatment based on gender or gender-related factors may occur at earlier stages in the legal process. For instance, certain women may be filtered out of the criminal justice system by the police or by the prosecutor, thereby influencing the resulting female offender population. There is varying evidence to suggest that women do experience leniency from criminal justice decision-makers at other stages in the system such as decision to arrest, bring charges, plea-bargain, during the Presentence investigation, and parole (Erez, 1992; Figueira-McDonough, 1985; Frazier, Bock & Henretta, 1983; Wilbanks, 1986). Research continues to examine stages in the criminal justice system where discretion can result in possible differential treatment based on gender. A second limitation involves problems related to the use of existing data. Despite the fact that the same 235 agency, the Minnesota Sentencing Guidelines Commission, collected information about convicted and sentenced offenders on a yearly basis since its inception, there remained several problems with consistency of data over time. First, the same measures were not collected in all three time periods. When information was unavailable for a specific construct in one of the time periods it had to be dropped from the model(s) in order to allow for a true comparison of the results across time. This involved measures such as the type of care for dependent children, the role of the offender in the commission of the crime, whether or not a weapon was used during the commission of the crime, and distinctions between drug offenses and property offenses. Prior research suggests that these measures have been influential in sentencing decisions, but because of their unavailability in the various data periods, they could not be included in the analysis. Nevertheless, many key and important independent measures remained and were included in the logistic and OLS models. Second, the same construct, in some instances, was operationalized using a different level of measurement. In the event that this occurred, information for the variable was collapsed to the lowest level of measurement, for 236 example from ratio level to nominal level (e.g., prior felonies). Consequently, precision was lost for some measures in exchange for their inclusion in the analysis. It is evident that women did receive more lenient sentences at the initial sentencing stage, but why? The inability to answer this question with the current dataset is a third limitation of this study. The available data from Minnesota do not include the information to answer this question. Data on sentencing departures were available, however they were only collected and recorded when judges sentenced outside the presumptive sentence under the guidelines. The answer as to why differential sentencing exists based on gender was outside the scope of this study and remains a fruitful area of exploration in subsequent studies on sentencing decisions. This question is best answered with research using a combination of existing court records and official statistics, along with qualitative research examining the court context and observations or interviews with key court actors such as judges, prosecutors, and defense attorneys. Finally, the generalizability of the results from this study are questionable. First, Minnesota has a unique sentencing guidelines system, thereby reflecting sentencing practices in this state only. One cannot necessarily 237 generalize these findings to other states that do not have guidelines and as much structure in their sentencing systems. Even in states that have established sentencing guidelines, such as Pennsylvania and Washington, there are important distinctions between even these systems and the one in Minnesota. Second, policies and how they impact on women in the Minnesota criminal justice system do not necessarily reflect how women are handled in other states. For example, Minnesota in 1994 had a lower incarceration rate (10 per 100,000 women) than all but two states (North Dakota, 5 per 100,000; Maine, 9 per 100,000), and was below the national average (40 per 100,000). Even after the introduction of sentencing guidelines, Minnesota still maintained one of the lowest incarceration rates in the country (BJS, 1994). As a result, it is recommended that future research examine these same issues in other states, particularly where reforms significantly transformed sentencing practices and systems are dealing with a different offender population, for example larger female corrections systems, and one with a more diverse racial composition. 238 Implications for Theory This section discusses research results and their implications for theories such as multiracial feminism and social construction feminism. Specifically, the implications of findings for explanations based on chivalry and gender role expectations (e.g., concerning motherhood) are discussed. A review of the literature highlights the fact that the relationship between gender and sentence outcome is a complex one. Chivalry As already noted, prior research examining the effect of gender on sentencing practices often discovered lenient treatment of women (sometimes under certain conditions) by the courts. The present research explored the chivalry thesis, which maintains that women are treated more leniently than men are by the criminal justice system. Further, women who come into contact with the legal system are protected, resulting in the differential treatment of women by officials. It is this protection through leniency that is the focus of this study. Controlling for certain legal and other extra—legal factors, women drug and property 239 offenders received more lenient sentences than their male counterparts in the state of Minnesota. Gender was a significant predictor of the In/Out sentencing decision in the overall model and at both Time 1 and Time 3, but not at Time 2 once the interaction between race and gender was introduced in the model. Therefore, gender was important in sentencing decisions under both an indeterminate sentencing system (i.e., Time 1), and the sentencing guidelines system during 1994. Prior work, however, suggests that the impact of chivalry on sentencing decisions occur under certain circumstances, and is not extended to all women. Peeping chivalry ingo gentexg According to Moulds (1980), chivalry oftentimes reveals itself in today’s society through what is deemed appropriate behavior on the part of men and women and in the relationship between both. As a result, women are supposed to carry themselves in feminine, meek and subordinate ways in relation to their male counterparts. Further, appropriate behavior is many times connected to the role women serve as mothers and/or wives. In social construction feminist theory, Lorber (1998) makes the point that gender inequality results when gender is socially constructed in order to continually re-create 240 boundaries between gender categories. Further, gender is a society-wide institution where gender influences the distribution of power, privileges, and monetary resources (Lorber, 1998: 160). While social construction feminism appropriately points out that gender expectations are learned early on in family and school settings, this research examined whether certain gender expectations are reinforced in the justice system at the time of sentencing. Because of sentencing changes discussed previously, the impact of chivalry and gender expectations may not have the same influence on court decisions as they once had. Prior studies discovered leniency results from influential factors often associated with traditional gender role expectations. Consequently, being a woman is simply not enough, but being a woman under a certain set of circumstances warrants leniency on the part of judges and/or courts. For example, prior research indicated that having dependent children is important. The courts in some settings have tried to protect the family unit by not removing the mother, thereby making it less likely to receive a prison sentence. The results from the interaction models, including gender and dependent children, indicate there were no significant effects. Thus, women were sentenced for the offenses regardless of whether or not they 241 had dependent children. It appears that courts were not necessarily worried about protecting the family unit, as much as they were in protecting women more generally. Prior research also suggests that aspects of the conviction offense (e.g., severity, type, role, etc.) are significant predictors of sentencing decisions. For example, the commission of traditional female offenses, as compared to non-traditional female offenses, often results in lenient sentences for women offenders. The level of blame is lower for women who commit offenses that are more consistent with their perceived nature. Results from the analysis of the In/Out sentencing decision do not support such a position. Whether women were convicted of drug related offenses (i.e., non-traditional) or property offenses (i.e., traditional), they received similar prison sentences. Thus, it appears that judges and courts appear to place an equal degree of blameworthiness on property and drug offenses, controlling for seriousness of the offense and other legal measures. Previous research, although not consistent, asserts that chivalry and gender role expectations are shaped by race. Several criminologists have suggested that women of color have not enjoyed the same history of leniency as Caucasian women. Instead, women of color have been treated 242 in a similar manner to men by the criminal justice system. The current research considered the importance of the interaction between race and gender with regard to the likelihood of receiving a prison sentence. With the exception of Time 2, the race and gender interaction was not a significant predictor of the In/Out decision. For Time 2, white women were sentenced more punitively than nonwhite women, and were likely to receive similar sentences to those of nonwhite men. Findings from the current research, for the most part, did not support each of the expected significant interactions. Women regardless of their status as mothers, traditional female offenders, or racial identification were treated leniently compared to male offenders. It appears that either the general status of being a woman or other characteristics or circumstances common to women warrant leniency by the courts in the state of Minnesota, even after the implementation of sentencing guidelines. It is important to understand that although the courts sentenced women more leniently, one cannot necessarily conclude that the lenient treatment found in this research resulted from chivalry on the part of the courts. Continued research into the reasons why women were sentenced differently, particularly under a determinate sentencing 243 scheme would need further investigation. In order to make serious advances in the theories involving the court processing area, it is my belief that future work needs address several issues. First, research must consider the issue of why female offenders are sentenced more leniently than are male offenders. Most sentencing research relies on official data collected by criminal justice agencies. Additionally, most sentencing disparity research focuses on the opinions and decisions of judges. Further research on gender and sentence disparity needs to utilize a qualitative analysis approach such as conducting a case study of a court system, and more importantly the key court officials working within the system. Further research should focus on the opinions and decision-making of defense attorneys, prosecutors, and probation officers in addition to judges. This type of grounded approach would permit more of an exploratory analysis of not only sentencing decisions but decisions of court officials affecting the adjudication process. Second, further research on the use of departures might assist researchers in understanding the relationship between gender and sentencing outcomes under guidelines systems. Prior research from the state of Pennsylvania suggests that departures may be connected to gender. Therefore reasons 244 for differential decisions might be discovered in subsequent research examining departures, or more importantly, in qualitative contributions. Given the research conducted by Moore and Miethe (1986) and Kramer and Ulmer (1996), further research should be completed in order to understand the significance of using departures as a way to circumvent recent sentencing reforms. Research involving less structured determinate sentencing systems such as the one in Pennsylvania suggests that judges might be resorting to the limited discretion that is available to them in order to sentence defendants according to their sense of justice when it differs from that recommended by the guidelines. Implications for Policy Minneeoee Seneencing Guidelines Next, implications from the findings of this research for policy are considered. The findings suggest that the effects of gender have not disappeared from sentencing decisions in Minnesota. Controlling for several important legal measures, gender remained influential in who received prison sentences. This occurred under the indeterminate sentencing system in the late 19703, for nonwhites under the 245 sentencing guidelines systems in the early 1980s, and for all women in 1994. Thus, with the partial exception of white women, sentencing reforms did not affect the relationship between gender and the likelihood of receiving a prison sentence over the span of 15 years in the state of Minnesota. Further, the use of sentencing guidelines in the state did not meet its goal of reducing sentencing disparity due to extra-legal variables such as gender and race for property and drug offenses. While legal factors were consistently significant predictors of the initial sentencing outcome measure, gender, race, and several other extra-legal factors remained significant. The results here are consistent with the findings from Miethe and Moore (1989), who found that criminal justice officials wanted more flexibility and discretion added to the guidelines, and officials had found ways to circumvent the guidelines system. Additional exploration into these findings is necessary in order to surmise the reasons behind such sentencing patterns. Do judges in Minnesota disagree with the guidelines and presumptive sentences for women and not men, or are there other important measures not included in the study that could explain the results? The recommendations 246 outlined in the previous section concerning future theory development may help explain why women continue to be sentenced more leniently in spite of sentencing changes. Justice for Women The concern for what is justice for women was addressed in Chapter 1, in the discussion about the debate over equal versus different treatment of women offenders. Both ‘equality under the law’ and ‘separate but equal’ positions were identified within the scope of the debate. Chesney— Lind and Pollock (1995) appropriately point out that the equalization position has benefited women to a certain degree, specifically in terms of correctional programming, however, for sentencing, the equal treatment of women has translated into an increased reliance on incarceration as a sanction. This study examined the possible implications of implementing an ‘equal’ policy approach for sentencing. It appears that in Minnesota some degree of justice has been preserved through judicial decision-making, which seems to take into account gender or factors related to gender. What is justice for women? This question seems simple from the outset, but in actuality it is complicated and grounded in a more fundamental question of what is justice 247 for all people involved in the criminal justice system. The arguments concerning a just legal system can be made for other marginalized groups of people, nonetheless I have chosen to focus in this study specifically on women and the issue of justice. A just legal system for women is one that is reactive to womens' experiences and realities. Although our legal system assumes that it can neatly categorize ‘like situated' offenders together in order to sentence them to equal and proportionate sanctions, the truth is that women are not on an equal footing with men in society (or with each other for that matter). Race, age, economics, education, family violence, abuse, sexual exploitation, dependent children, and drugs shape the reality for the women involved in the criminal justice system. In an effort to promote what is perceived to be a fair and a just legal system, our sentencing policies ignore the circumstances, experiences, and realities of women. In this study I have tried to describe the unique experiences and realities of women offenders in order to show why recent sentencing reforms make little sense and have unfortunate consequences for them. Despite the evidence that protection or leniency is still present in sentencing practices, there should be 248 continued concern about the impact of sentencing reforms and policies on women. In fact, other state systems should be examined in order to gain a broader perspective on the issue. Very few studies have explored the policy concerns, thus additional research is needed to inform those that make sentencing policy, about the impact and possible unintended consequences of such policies in order to determine what is justice for women. 249 APPENDIX A MINNESOTA SENTENCING GUIDELINES GRID rm >86 UGO cam Hmm> mac 4 oN-mH H unannmndw UOHHOHDEOU ma ea us ma «ma .ma .mH ouumfissam Lo nfimm mm-om HH Loom.mu-oommv >n0muom Russo Hm mm ea ma ms .ma «ma Amnma no oom.mwo umsflno Dunne mm-mm mm-om om-ma HHH mm Hm ma em ma ms .ms oom.mm uo>ov unsano nouns mm-mm um-um um-mm >H om em om mm us ma 4mg xnmsmssm Hmaucooauousoz omlmv mv-Hv ov-mm mmlfim > >Hwnnom THQEflm us no mm mm mm mm oH eumfimusm Hmaucooauom mm-mm mm-mv ev-mu mv-em H> inc u Amy cosmos Em eu Hm me mm mm em Hm .nosocoo Hmsxom Hmcmsaso mem-vom mom-om mm-oo mu-ve me-vu mu-wu mm-sv HH> mos um we we ou om ms summon 8H xssnhom pnum>mnmm< mue-mum mum-moe mma-mmm emH-eHH umm-mom mom-mm Hm-ao HHH> ssnmmo as .Dmsmmm4. 008m0o 6H mum usa ems mmH oHH um um .Dusucoo Hmsxmm Hmcassno uvm-vmm mmm-mem umm-vom Hom-mom uum-wee mem-muH umm-vvs xH immense Hmnofionuncscsv oom mmm omm mas ems mus omH 00nm0o(€m \mnnmno Em .uwousz mmv-mav mam-mmm mmm-mem mem-mum mmm-mmm mmm-msm mam-mmm x inmcauoorm->n-0>asn incense umu uoo uum uum uom umm uom accoaosoneav cosmoo Em .snossz + w m w m N H o AmUAHmbfl as momsmuuo EOEEOUV mmzmmuo mo qm>ma eeHmoomm oEfiu Hash ou uovflndm mum mmusvuswm xsonu usuEsomeumselsos :uw3 muvcsvwwo mmoom %mOHmHm AflZHZHmU .3MH ou mascuouum .wusuummmc m UOEOOU mason mucousmm mnu usonufiz mucouscm >mE umpsm m nofinz Genus: Omsmu Onu Ouosoc pflum mnu Genus: mHOQEDC poneowamuH mausoz Ge mnumsmq ouswucom O>AUQESm0Hm QHMO mmZHAMQHDO UZHUZWBme dBOmmZZHZ I fl NHQmemd 251 emmH .H Dmsms< 0>au00uum .mmosuucmm >u0umcsmz .m.HH pom mucousmm m>euQEDmmum .U.HH msofluumm mum .ADHDMmmd mmummo csoomm ..m.mv cosmos msoummsmp 8 mo mm: on» on was Emma GOmHuQ ESEAGAE kHODmpsms m msflxuumo momsmwuo cam momsmumo unaccoo HmsxOm Hmswsflno usmsvmmnsm pom psoomm .:0euoe>soo xumHmHsn >QOHOM uoflum m mm: umpcowwo on» smnz msfiHHO3o owemsooo Gm wo >umHmusm .sofluofi>soo mono %GOHOM MOAMQ m mun umcswmuo mnu son: moseuo musmumnsm cmaaouusoo summon phase OUDHOGH mumcomwo mmmne .GOmeuQ Uumum 8 Cu usmsueesoo m>eumeswcum m xuumo m>m3am Oeum Unu mo Gofluomm menu Ge momsmwuo Gemuumo .Hm>030m .soflumnosm wo maceuepsoo mm pmmOQEH on smo msofluosmm Hasn-sos Honuo Ho\psm Hash Ge Hmm> m Cu ms .Ompsw can we seeumuowflo 0:» us moosmucmm pmxmum O>HDQESmmHm .GOmaum Eoum pmmmmaou mumcsmwwo xmm H0m sonfl>ummsm wo mcoflumm Eseflses msHUSHose .3mH >3 CUHHOHusOO moosmusom mwonu msepummmu >UHHOQ no“ moosmucmm >u0umocmz .m.HH sofluomm mom .Oosmucmm Owea >u0umpsme m m>m£ Ou mosseusoo cam 3nd >3 wmswamcflsm ecu Eouw UOCDHOXO we mucus: moummo umsflm .uGoESOmflHQEH Oumum ou unmEuHEEoo mm>eumssmmum 252 APPENDIX B OFFENSE SEVERITY LEVEL 2% APPENDIX B - OFFENSE SEVERITY LEVEL Relevant Factors in Determining Offense Severity Level: A. Offense severity is determined by the offense of conviction. The Commission thought that serious legal and ethical questions would be raised if punishment were to be determined on the basis of alleged, but unproven, behavior, and prosecutors and defenders would be less accountable in plea negotiation. It follows that if the offense of conviction is the standard form which to determine severity, departures from the guidelines should not be permitted for elements of offender behavior not within the statutory definition of the offense of conviction. Thus, if an offender is convicted of simple robbery, a departure from the guidelines to increase the severity of the sentence should not be permitted because the offender possessed a firearm or used another dangerous weapon. The date of the offense is important because the offender's age at the time of the offense will determine whether or not the juvenile record is considered, the date of the offense might determine whether a custody status point should be given, and the date of the offense determines the order of sentencing with multiple convictions. For those convicted of a single offense, there is generally no problem in determining the date of the offense. For those convicted of multiple offenses when theft and damage to property aggregation procedures are used for sentencing purposes or when multiple offenses are an element of the conviction offense, the following rules apply: 0 If offenses have been aggregated under Minn. Stat. § 609.52, subd. 3(5), or § 609.595, the date of the earliest offense should be used as the date of the conviction offense. 0 If multiple offenses are an element of the conviction offense, such as in Subd. 1(h) (iii) of first degree criminal sexual conduct, the date of the earliest offense should be used as the date of the conviction offense. 254 APPENDIX C CALCULATION OF CRIMINAL HISTORY INDEX SCORE 2% APPENDIX C - CALCULATION OF CRIMINAL HISTORY INDEX SCORE The offender’s criminal history index score is computed in the following manner: 1.Subject to the conditions listed below, the offender is assigned a particular weight for every extended jurisdiction juvenile conviction and for every felony conviction for which a felony sentence was stayed or imposed before the current sentencing or for which a stay of imposition of sentence was given before the current sentencing. For the purposes of this section, prior extended jurisdiction juvenile convictions are treated the same as prior felony sentences. a. The weight assigned to each prior felony sentence is determined according to its severity level, as follows: Severity Level I — II = % point; Severity Level III - V = 1 point; Severity Level VI — VII = 1 % point; Severity Level VIII - X = 2 points; and Murder 18t Degree = 2 points. The severity level to be used in assigning weights to prior offenses shall be based on the severity level ranking of the prior offense of conviction that is in effect at the time the offender commits the current offense. b. When multiple sentences for a single course of conduct were imposed pursuant to Minn. Stats. §§ 609.585 or 609.251, only the offense at the highest severity level is considered; Only the two offenses at the highest severity levels are considered for prior multiple sentences arising out of a single course of conduct in which there were multiple victims; When a prior felony conviction resulted in a misdemeanor or gross misdemeanor sentence, that conviction shall be counted as a misdemeanor or gross misdemeanor conviction for purposes of computing the criminal history score, and shall be governed by item 3 below; an e. Prior felony sentences or stays of imposition following felony convictions will not be used in computing the criminal history score if a period of fifteen years has elapsed since the date of discharge from or expiration of the sentence, to the date of the current offense. Calculation of Score: The basic rule for computing the number of prior felony points in the criminal history score is that the offender is assigned a particular weight for every felony conviction for which a felony sentence was stayed or imposed before the current sentencing or for which a stay of imposition of sentence was given before the current sentencing. Prior felony convictions for an attempt or conspiracy for which a felony sentence was stayed or imposed before the current sentencing are weighted the same as completed offenses. The felony point total is the sum of these weights. No partial points are given - thus, a person with less than a full point is not given that point. For example, an offender with a total weight of 2 % would have 2 felony points. 257 APPENDIX D FACTORS TO BE EXCLUDED IN MAKING DEPARTURE DECISIONS 2% APPENDIX D - FACTORS TO BE EXCLUDED IN MAKING DEPARTURE DECISIONS Factors that should not be used as reasons for departure: The following factors should not be used as reasons for departing from the presumptive sentences provided in the Sentencing Guidelines Grid: A. B. (A) :bUJNH Race Sex Employment factors, including: occupation or impact of sentence on profession or occupation; employment history; employment at the time of offense; employment at time of sentencing. Social Factors, including: educational attainment; living arrangements at time of offense or sentencing; length of residence; marital status. The exercise of constitutional rights by the defendant during the adjudication process. 259 APPENDIX E FACTORS TO BE INCLUDED IN MAKING DEPARTURE DECISIONS um APPENDIX E - FACTORS TO BE INCLUDED IN MAKING DEPARTURE DECISIONS Factors that may be used as reasons for departure: The following is a nonexclusive list of factors, which may be used as reasons for departure: Mitigating Factors 1. 2. 5. The victim was an aggressor in the incident. The offender played a minor or passive role in the crime or participated under circumstances of coercion or duress. The offender, because of physical or mental impairment, lacked substantial capacity for judgement when the offense was committed. The voluntary use of intoxicants (drugs or alcohol) does not fall within the purview of this factor. The offender’s presumptive sentence is a commitment to the commissioner but not a mandatory minimum sentence, and either of the following exist: The current conviction offense is at severity level I or II and the offender received all of his or her prior felony sentences during less than three separate court appearances; or The current conviction offense is at severity level III or IV and the offender received all of his or her prior felony sentences during one court appearance. Other substantial grounds exist which tend to excuse or mitigate the offender’s culpability, although not amounting to a defense. Aggravating Factors 1. The victim was particularly vulnerable due to age, infirmity, or reduced physical or mental capacity, which was known or should have been known to the offender. The victim was treated with particular cruelty for which the individual offender should be held responsible. 261 The current conviction is for a Criminal Sexual Conduct offense or an offense in which the victim was otherwise injured and there is a prior felony conviction for a Criminal Sexual Conduct offense or an offense in which the victim was otherwise injured. The offense was a major economic offense, identified as an illegal act or series of illegal acts committed by other than physical means and by concealment of guile to obtain money or property, to avoid payment or loss of money or property, or to obtain business or profession advantage. The presence of two or more of the circumstances listed below are aggravating factors with respect to the offense: the offense involved multiple victims or multiple incidents per victim; the offense involved an attempted or actual monetary loss substantially greater than the usual offense or substantially greater than the minimum loss specified in the statutes; the offense involved a high degree of sophistication or planning or occurred over a lengthy period of time; the defendant used his or her position or status to facilitate the commission of the offense, including positions of trust, confidence, or fiduciary relationships, or the defendant has been involved in other conduct similar to the current offense as evidenced by the findings of civil or administrative law proceedings or the imposition of professional sanctions. The offense was a major controlled substance offense, identified as an offense or series of offenses related to trafficking in controlled substances under circumstances more onerous than the usual offense. The presence of two or more of the circumstances listed below are aggravating factors with respect to the offense: the offense involved at least three separate transactions wherein controlled substances were sold, transferred, or possessed with intent to do so; or 262 b. the offense involved an attempted or actual sale or transfer of controlled substances in quantities substantially larger than for personal use; or c. the offense involved the manufacture of controlled substances for use by other parties; or d. the offender knowingly possessed a firearm during the commission of the offense; or e. the circumstances of the offense reveal the offender to have occupied a high position in the drug distribution hierarchy; or f. the offense involved a high degree of sophistication or planning or occurred over a lengthy period of time or involved a broad geographic area of disbursement; or g. the offender used his or her position or status to facilitate the commission of the offense, including positions of trust, confidence or fiduciary relationships (e.g., pharmacist, physician or other medical professional). 6. The offender committed, for hire, a crime against the person. 7. Offender is a “patterned sex offender” (See Min. Stat. § 609.1352). 8. The offender committed the crime as part of a group of three or more persons who all actively participated in the crime. NOTE: The Commission provided a non—exclusive list of reasons, which may be used as reasons for departure. The factors are intended to describe specific situations involving a small number of cases. The Commission rejected factors which were general in nature, and which could apply to large numbers of cases, such as intoxication at the time of the offense. The factors cited are illustrative and are not intended to be an exclusive or exhaustive list of factors, which may be used as reasons for departure. Some of these factors may be considered in establishing conditions of stayed sentences, even though they may not be used as reasons for departure. For example, whether or not a person is employed at time of sentencing may be an important factor in deciding whether restitution should be used as a condition of probation, or in deciding on the 263 terms of restitution payment. 264 APPENDIX F HISTORY OF MINNESOTA’S CONTROLLED SUBSTANCES LAW flfi APPENDIX F - HISTORY OF MINNESOTA’S CONTROLLED SUBSTANCES LAW Initial Sentencing Guidelines Handling of Drug Offenses Table 1: DRUG-RELATED OFFENSES AND SEVERITY LEVELS (Sentencing Guidelines Prior to August 1, 1986) Severity Level VI IV III II Offense Sale of Hallucinogens, PCP, Heroin, and Remaining Schedule I & II Narcotics Sale of Cocaine (Thie offense was ranked et eeveriey level III prior Lo 1982) Sale of Remaining Schedule I, II, & III Non-narcotics Possession of Hallucinogens, PCP, Heroin, and Remaining Schedule I & II Narcotics Sale of Marijuana/ Hashish/ Tetrahydrocannobinols, and Schedule IV substances Sale of Simulated Controlled Substance Possession of Cocaine, Marijuana/ Hashish/ Tetrahydrocannabinols, Remaining Schedule I, II, & III Non-narcotics, and Schedule IV Substances Changes to the guidelines for drug related offenses between the startup of the guideline policy up through August 1, 1986: ** Effective August 1, the commission added the factor of “major controlled substance offense” to the nonexclusive list of aggravating factors for departure. NM ** Effective August 1, 1985, the commission added guideline language to II.C. to presume a prison sentence for persons convicted of sale of cocaine or sale of a controlled substance that was ranked at severity level VI if there had been a previous adjudication of guilt for sale of cocaine or sale of a severity level VI drug. (Source: MN Sentencing Guidelines Commission Report, 1992: 3) Aggravated Departure Categopy - “Major Controlled Substance Offense” Because the amount of the controlled substance which was sold did not influence the severity level nor the offense, the Sentencing Commission established an aggravated (or upward) sentencing departure category for what they referred to as a “major controlled substance offense.” Under Minnesota Sentencing Guidelines § II.D.2.b.(5), a court was allowed an aggravated departure from the presumptive guidelines sentence when: (1) the offense involved at least three separate transactions wherein controlled substances were sold, transferred, or possessed with intent to do so; (2) the offense involved an attempted or actual sale or transfer of controlled substances in quantities substantially larger than for personal use; or (3) the offense involved the manufacture of controlled substances for use by other parties; or (4) the offender knowingly possessed a firearm during the commission of the offense; or (5) the circumstances of the offense reveal the offender to have occupied a high position in the drug distribution hierarchy; or (6) the offense involved a high degree of sophistication or planning or occurred over a lengthy period of time or involved a broad geographic area of disbursement; or 267 (7) the offender used his or her position or status to facilitate the commission of the offense, including positions of trust, confidence or fiduciary relationships (e.g., pharmacist, physician, or other medical professional). (Source: Memorandum to Members of the MN Sentencing Guidelines Commission, July 14, 1999: 2-3.) 1986 Legislative and Senpencing Guidelines ghangee The 1986 Minnesota Legislature modified the drug laws to provide greater statutory maximum penalties for offenders convicted of the sale of 7 or more grams or 10 or more dosage units of any narcotic classified in schedule I or II, or PCP, or hallucinogens (other than marijuana). (1) The commission ranked these new controlled substance offenses at severity level VII. At that time, a severity level VII carried a sentence of 24 months—but this sentence was presumed to be executed; for the first time, the guidelines presumed imprisonment for a drug offense. (2) The sale of smaller amounts of most drugs remained at severity level VI-21 months stayed. The severity level of sale of a small amount of cocaine was increased. It had been a severity level IV offense since 1982, but was increased to a severity level VI offense (21 months - stayed). (3) The Sentencing Guidelines Commission increased the severity level for cocaine possession from I to III. (Source: MN Sentencing Guidelines Commission Report, 1992; Memorandum to Members of the MN Sentencing Guidelines Commission, July 14, 1999: 3-4.) 268 1987 Legislative end Senteneing Quidelinee Qhengee During 1987, the Minnesota Legislature implemented different threshold levels for harsher penalties for powder and crack cocaine sales. The threshold for the higher penalty was set at 3 grams for crack and 10 grams (sold on one or more occasions within a 90 day period) for powder. (Source: Memorandum to Members of the MN Sentencing Guidelines Commission, July 14, 1999: 4). 1989 Legislaeive and Sen§encing Guidelinee ghengee There were significant Changes made to the drug laws by the Minnesota Legislature in 1989. The Legislature created several levels of controlled substance offenses - first, second, third, fourth, and fifth degree offenses - in decreasing order of severity. See 1989 Minn. Laws Ch. 290, art. 3, §§8-12, codified at Minn. Stat. 55 152.01-.028. Two critically important changes occurred with new statutes. First, ell people who either possessed or sold drugs at the levels indicated for first, second, or third degree offenses were presumed to be drug dealers. Second, because of this, the definition of “sale” of drugs no longer included “possession with intent to sell;” because of the presumption that a person was a drug dealer if he or she possessed a certain amount of drugs, defining “sale” to include “possession with intent to sell” would have been redundant (and would have improperly increased the punishment of certain offenders). The Commission increased penalties for new offenses: (1) First degree offenses. Ranked at severity level VIII — these were, in the words of the legislative history, the true drug kingpins, the drug wholesalers. These people -who possessed 500 grams of powder cocaine (roughly one pound), or were selling 50 grams of powder cocaine (roughly two ounces) at a time - were viewed to be similar to a person who raped someone using a threat of serious bodily injury. Severity level VIII offense punishments were increased during this same period, so this offense carried a presumptive sentence of 86 months in prison. 269 (2) Second degree offenses. People who possessed 50 grams (roughly two ounces), or who sold 10 grams, were guilty of a second degree offense. This offense was ranked at severity level VII - which carried a newly increased sentence of 48 months in prison. (3) Third degree offenses. People who possessed 10 grams of cocaine, and who sold any amount of cocaine, were guilty of a third degree offense. The presumed sentence was 21 months (stayed). The 1989 statutory changes to the drug laws in Minnesota mirrored those sanctions established at the federal level in regards to differentiating between crack cocaine and powder cocaine offenses. The thresholds established by the Legislature for First, Second and Third degree cocaine offenses include: Sale Possession Degree Crack Powder Crack Powder First 10 grams 50 grams 25 grams 500 grams Second 3 grams 10 grams 6 grams 25 grams Third Any amount Any Amount 3 grams 10 grams (Source: Memorandum to Members of the MN Sentencing Guidelines Commission, July 14, 1999: 6.) The result of these changes was disparate impact on minority (primarily African American) offenders. The bulk of people prosecuted for crack cocaine were African Americans received prison sentences and the bulk of people prosecuted for powder cocaine offenses were white did not. 2m The Minnesota Supreme Court in State v. Russell (477 N.W.2d 886, Minn. 1991) declared the disparate treatment of powder cocaine and crack cocaine offense were unconstitutional under the Minnesota Constitution. The court decreased penalties for crack cocaine offenders to be equal to those of powder cocaine offenses. The legislature moved quickly in response to the MN Supreme Court decision and increased penalties for powder cocaine to those formerly set for crack cocaine. To deal with the question of disparate handling of cocaine offenses, the legislature simply set up a new system which punished both cocaine offenses the same - reverting back to the system which was originally set up to sanction crack cocaine. (Source: Memorandum to Members of the MN Sentencing Guidelines Commission, July 14, 1999: 6-7). 271 Drug-related Offenses and Severipy Levels (Effective August 1, 1993) fiev, 22221 821382 8 2 Q2str211ss_8nh1fsnss_Qrins_in_tss_zirs§_nssrssi M-S- § 152.021) a1 P SS 8 i n wi I t n : v r da Esrigd(subd. 1) (1) 10 or more grams Cocaine (2) 50 or more grams Narcotic other than Cocaine (3) 50 grams or 200 or more dosage units PCP/ Hallucinogens/ Methamphetamine (4) 50 kilograms or more Marijuana p; 25 kilograms or more Marijuana in a School, Park, or Public Housing Zone Possessign (subd. 2) (1) 25 or more grams Cocaine (2) 500 or more grams Narcotic other than Cocaine (3) 500 grams or 500 or more dosage units PCP/ Hallucinogen/ Methamphetamine (4) 100 kilograms or more Marijuana 7 1.5 n r : (M.S. § 152.022) al Po 8 i n wi In :A a v 9 a Esrigd(subd. 1) (1) 3 or more grams Cocaine (2) 10 or more grams Narcotic other than Cocaine (3) 10 grams or 50 or more dosage units PCP/ Hallucinogen/ Methamphetamine (4) 25 kilograms or more Marijuana (5) Cocaine/ Narcotic to minor or employs minor (6) Any of the Following in a School, Park, or Public Housing Zone: (i) Schedule I & II Narcotics or LSD (ii) Methamphetamine/ Amphetamine (iii)5 kilograms or more Marijuana Poseession (subd. 2) (1) 6 or more grams Cocaine (2) 50 or more grams Narcotic other than Cocaine (3) 50 grams or 100 or more dosage units PCP/ Hallucinogen/ Methamphetamine (4) 25 kilograms or more Marijuana 6 1.5 n r h T : (M.S. § 152.023) 272 gev a Level Weigh; SaleZPesseesion wiph Ippenp (subd. 1) (1) Cocaine/ Narcotic (2) 10 or more dosage units of Hallucinogen/ PCP (3) Schedule I, II, III to minor - Not Narcotics (4) Schedule I, II, III employs minor — Not Narcotics (5) 5 kilograms Marijuana Poeeeeeiep (subd. 2) (1) 3 or more grams Cocaine (2) 10 or more grams Narcotic other than Cocaine (3) 50 or more dosage units of Narcotics (4) Sch. I & II Narc./ 5 or more d.u. LSD in a School, Park, or Public Housing Zone (5) 10 kilograms Marijuana (6) Methamphetamine/ Amphetamine in a School, Park, or Public Housing Zone QQQEIQJJEQ EHRIEBDQQ $11!! in sh; IQQEEh pegged: (M.S. § 152.024): SaleZPosseseign wiep lnten; (subd. 1) (1) Schedule I, II, III (except for Marijuana) (2) Schedule IV, or V to minor (3) Employs minor to sell schedule IV or V (4) Marijuana in a School, Park, or Public Housing Zone Possessign (subd. 2) (1) 10 or more dosage units of Hallucinogen/ PCP (2) Schedule I, II, III (except Marij.) w/ intent to sell n r F1 : (M.S. § 152.025): SaleZPossession wiph Intent (subd. 1) (1) Marijuana (2) Schedule IV Essssssign (subd. 2) (1) Possession of Schedule I,II,III,IV - Includes Marijuana. Also includes: Crack/Cocaine/ Narcotics/ PCP/ Hallucinogen (2) Marijuana with intent to sell (3) Procurement by fraud 273 APPENDIX G OFFENSE SEVERITY REFERENCE TABLE 2% APPENDIX G- OFFENSE SEVERITY REFERENCE TABLE Severity Offense: State Statute: Level: Drug Offenses: VIII Controlled Substance Crime in 152.021 the First Degree VIII Importing Controlled 152.0261 Substances Across State Borders VII Controlled Substance Crime in 152.022 the Second Degree VI Controlled Substance Crime in 152.023 the Third Degree (non-aggregated offenses) IV Controlled Substance Crime in 152.024 the Fourth Degree II Controlled Substance Crime in 152.025 the Fifth Degree I Sale of Simulated Controlled 152.097 Substance Property Offenses: VII Arson I 609.561 VII Burglary I 609.582, 1(b) / (c) VI Bringing Stolen Goods into 609.525 State (over $2,500) VI Burglary I 609.582, subd. 1(a) VI Precious Metal Dealers, 609.526, (1) Receiving Stolen Goods (over $2,500) VI Precious Metal Dealers, 609.526, Receiving Stolen Goods Zm’or subs. (over $300) Viol. VI Theft over $35,000 609.52, subd. 3 (1) V Arson 2 609.562 TH IV Bringing Stolen Goods into State ($1,000-$2,500) Burglary Bringing Stolen Goods into State ($301-$999) 609.525 609.582, subd. 2(a) (b) 609.525 / 2M REFERENCES 277 REFERENCES Agresti, A., & Finlay, B. (1986). Statistical methods for the social seiences. San Francisco, CA: Dellen Publishing Company. Albonetti, C.A. (1997). Sentencing under the Federal sentencing guidelines: Effects of defendant characteristics, guilty pleas, and departures on sentence outcomes for drug offenses, 1991-1992. Law & Socieey Review, ;l(4), 789—822. Armstrong, G. (1977). Females under the law--“protected” but unequal. grime e Delingpency, ge,09-120. Bachman, R., & Paternoster, R. (1997). ° ti m t od for eriminology end eziminel jdetiee. New York: The McGraw-Hill Companies, Inc. Belknap, J. (1996). The Invieible woman: gender, erlme, end justice. Belmont: Wadsworth Publishing Company. Berk, R.A. (1983). An Introduction to sample selection bias in sociological data. Americap Seclelegieal Revlew, ge, 386—398. Berk, R.A., & Ray, S.C. (1982). Selection bias in sociological data. Spelel Seience Reeearch, ll, 352- 398. Bickle, G.S., & Peterson, R.D. (1991). The Impact of gender- based family roles on criminal sentencing. Seeiel Erppleme, ;§(3), 372-394. Brame, R., Paternoster, R., Mazerolle, P., & Piquero, A. (1988). Testing for the equality of maximum-likelihood regression coefficients between two independent equations. Journal pf Quantipapive Criminology, lg(3), 245-261. Bureau of Justice Statistics. (1994). Women Offenders. Washington, D.C.: U.S. Department of Justice. Bureau of Justice Statistics. (1995). Drugs and crime facts, 1994. Washington, D.C.: U.S. Department of Justice. 2m Bush-Baskette, S.R. (1998). The War on drugs as a war against black women. IN S.L. Miller (ed.), grime control and women (pp. 113-129). Thousand Oaks, CA: Sage Publications. Bush-Baskette, S.R. (1996, July). nve i in h im a t of the ‘war on drugs’ en the incarceraeion pf females. Paper presented at the Joint Meetings of Law and Society Association and Research Committee on the Sociology of Law of the International Sociological Association, Glasgow, Scotland. Byrne, J.M., & Brewster, M. (1993). The Future of American corrections: Punishment or reform? Federel Propegign, 2(4)! 3'9- Byrne, J.M., & Taxman, F.S. (1994). Crime control policy and community corrections practice: Assessing the impact of gender, race, and class. Evaluapien end Pregrem Plenning, l1(2), 227-233. Chesney-Lind, M. (1997). Th emal ff - ' w m n end prime. Thousand Oaks: Sage Publications. Chesney—Lind, M., & Pollock, J.M. (1995). Women’s prisons: Equality with a vengeance. In A.V. Merlo & J.M. Pollock (Eds.), Wemen, lew, e epelal eonerol (pp. 155-175). Boston, MA: Allyn and Bacon. Chesney-Lind, M. (1992). Rephinkipg wemen'e impplepnmene; A '.i -_ ‘ -mi . ion 0f _r- 0; i -m.l: in -7 -r- ion. Unpublished manuscript, University of Hawaii at Manoa. Chevalier-Barrow, M. (1992). er a en n in - An nv-: i0. 7-n o 9 io _- . . _-n of.‘ ;- -mil us n Ci m r 'c ch ri i Unpublished doctoral dissertation, Fordham University. Cornish, F.W. (1908). ghivalpy. New York: The Macmillan Co. Crew, B.K. (1991). Sex differences in criminal sentencing: Chivalry or patriarchy? dpetice Qparterly, §(1), 59- 83. Curran, D.A. (1983). Judicial discretion and defendant’s sex. Criminology, gl(1), 41-58. 279 Cuzzort, R.P., & Vrettos, J.S. (1996). The Elemenpepy forms pf staeietical reason. New York: St. Martin’s Press. D’Alessio, S.J., & Stolzenberg, L. (1993). Socioeconomic status and the sentencing of the traditional offender. dodpnal pf gpiminal dueeiee, gl, 61-77. Daly, K. (1987) Structure and practice of familial-based justice in a criminal court. Law & Sociepy Revlew, 21(2), 267-290. Daly, K. (1989). Rethinking judicial paternalism: Gender, work-family relations, and sentencing. gender g Society, 1(1), 9-36. Daly, K. (1994). Gender and punishment disparity. In G.S. Bridges & M.A. Myers (Eds.), ne 1' 'm eocial control. Boulder, CO: Westview Press. Daly, K., & Tonry, M. (1997). Gender, race, and sentencing. In M. Tonry (ed.), Crime and justice: A Review pf research, Vol. 22. Chicago: University of Chicago Press. Daly, K., & Bordt, R.L. (1995). Sex effects and sentencing: An Analysis of the statistical literature. deeplee Quarterly, l;(1), 141-175. DeJong, C. (1997). Survival analysis and specific deterrence: Integrating theoretical and empirical models of recidivism. Criminelogy, 15(4), 561-575. DeJong, C., & Jackson, K.C. (1998). Putting race into context: Race, juvenile justice processing, and urbanization. Jeepice Quarterly, l§(3), 487-504. Ditton, P.M., & Wilson, D.J. (1999). Truth in §entencing in State Prieene. Washington, DC: Bureau of Justice Statistics. Office of Justice Programs, U.S. Department of Justice. Edwards, A.R. (1989). Sex/Gender, sexism and criminal justice: Some theoretical considerations. Internagionel Journal of the Sociology of Law, l1, 165- 184. um Erez, E. (1992). Dangerous men, evil women: Gender and parole decision-making. duetice Quarteply, 2(1), 105- 126. Federal Bureau of Investigation. (1998) Uniform Crime Reports, 1998. Washington DC: Federal Bureau of Investigation. Feinman, C. (1994). Women ip phe eriminal jdetiee eyseem. Westport, CT: Praeger. Figueira-McDonough, J. (1985). Gender differences in informal processing: A look at charge bargaining and sentence reduction in Washington, D.C. dgdrnel pf Reeeapgh lp grime end Delingpency, ee(2), 101-133. Flavin, J.M. (1995). Qf Punishmen; and parenthgod; The Rel ti n hi b we f mil - r 1 an eentencing. Unpublished doctoral dissertation, American University. Fox, J. (1991). Re e ' ' ' . Newbury Park, CA: Sage Publications. Frase, R.S. (1994). Purposes of punishment under the Minnesota Sentencing Guidelines. griminel Jueeice Ethics, 13(1), 11-20. Frazier, C.E., Bock, E.W., & Henretta, J.C. (1983). The Role of probation officers in determining gender differences in sentencing severity. The Socielggieal Qparperly, g5, 305-318. Ghali, M., & Chesney-Lind, M. (1986). Gender bias and the criminal justice system: An Empirical investigation. Soeiology and Seeial Reseereh, 1Q, 164-171. Greenfeld, L.A., & Snell, T.L. (1999). Wemen effenders. Washington, DC: Bureau of Justice Statistics. Office of Justice Programs: U.S. Department of Justice. Griswold, D. (1987). Deviation from sentencing guidelines: The Issue of unwarranted disparity. Journal of Qriminal dpeeice, 15, 317-329. 281 Gruhl, J., Welch, S., & Spohn, C. (1984). Women as criminal defendants: A Test for paternalism. Wespern Pplitieal Quarterly, 11, 227-249. Heckman, J.J. (1979). Sample selection bias as a specification error. Ecpnemeerlga, 11(1), 153-161. Heckman, J.J. (1976). The Common structure of statistical models of truncation, sample selection, and limited dependent variables and a simple estimator for such models. Annals of Eeonomig and §ocial Meaedremenp, 1, 475-492. Inciardi, J.A., Lockwood, D., & Pottieger, A.E. (1993). Women end ereCk-eeeaine. New York: MacMillan Publishing Company. Johnson, D.R., & Scheuble, L.K. (1991). Gender bias in the disposition of juvenile court referrals: The Effects of time and location. Qriminglggy, 11(4), 677-699. Kaukinen, C.E. (1995). Women lewbreekere eonetructed in erms o radi ional fini i n f f minini - h Senpeneing pf wemep ln eenflle; WILD Lee leg. Unpublished master's thesis, University of Windsor, Ontario, Canada. Kennedy, P. (1997). A lede pg econpmepzlcs. Cambridge, MA: The MIT Press. Klein, D. (1995). The Etiology of female crime: A Review of the literature. In B.R. Price & N.J. Sokoloff (Eds.), he 'm'n ' i m n w men- n e s vieeime, end wgrkere (pp. 30-53). New York, NY: McGraw-Hill. Klepper, S., Nagin, D.S., & Tierney, L. (1983). Discrimination in the Criminal Justice System: A Critical Appraisal of the Literature. In A. Blumstein, J. Cohen, S. Martin, & M. Tonry (Eds.), Reeeareh pp eeppenelng; The Seereh fpp reform (Vol., 2: pp. 55— 128). Washington, D.C.: National Academy Press. Knapp, K.A. (1984). The Impaet ef the Minnesota Senpencing Qpidelines; Three year evaldepion. Minnesota Sentencing Guidelines Commission: St. Paul, MN. 282 Kramer, J.H., & Ulmer, J.T. (1996). Sentencing disparity and departures from guidelines. duepiee Qparperly, 11, 81-106. Krisberg, B. (1994). Distorted by fear: The Make believe war on crime. Soeial dustiee, 11(3): 38-49. Krohn, M., Curry, J.P., & Nelson-Kilger, S. (1983). Is Chivalry dead? An Analysis of changes in police dispositions of males and females. Qriminelpgy, 11(3), 417-437. Kruttschnitt, C (1984). Sex and criminal court dispositions: The Unresolved controversy. Journal pf Reeeereh in Qrime end Delingpency, 11, 213-232. Kruttschnitt, C. (1980). Social status and sentences of female offenders. Lew & Seciety Review, 11(2), 247- 265. Kruttschnitt, C. (1982). Respectable women and the law. The Socielogieel Quarterly, 11: 221-234. Kruttschnitt, C. (1985). Legal outcomes and legal agents: Adding another dimension to the sex-sentencing controversy. Law and Hpman Behevipp, 1(3), 287-303. Kruttschnitt, C., & Green, D.E. (1984). The Sex-Sanctioning issue: Is it history? Amerieen fieeielegieal Review, 12: 541-551. Long, J.S. (1997). Regression modele for categgricel end limited dependent variablee. Thousand Oaks, CA: Sage Publications. Lorber, J. (1998). Gender inegpality; Feminie; theoriee end pelipiee. Los Angeles: Roxbury Publishing Company. Mahan, S. (1996). Qraek eoceine, grime, end wemen. Drugs, Health, and Social Policy Series, Volume 4. Thousand Oaks, CA: Sage Publications. Maher, L., & Curtis, R. (1995). In Search of the female urban ‘gangsta’: Change, culture, and crack cocaine. In Price, B.R. & N.J. Sokoloff (Eds.), The Criminal juepiee eyspem end women (pp. 147-166). New York: McGraw-Hill, Inc. NB Mann, C.R. (1984). Female crime end delingpeney. University, AL: The University of Alabama Press. Mauer, M. (1994). The Fragility of criminal justice reform. Social duseice, 11(3), 14-29. Memorandum to Members of the Minnesota Sentencing Guidelines Commission (from Scott Swanson), July 14, 1999. Subject: History of controlled substance statutes in Minnesota as it relates to sentencing. Menard, S. (1995). Applied lggieeig regpeeeign enelyeis. Thousand Oaks, CA: Sage Publications. Miethe, T.D., & Moore, C.A. (1989). Sentencing guidelines: Their effect in Minnesota. Researeh in Brief. U.S. Department of Justice. Washington, D.C. Minnesota Sentencing Guidelines Commission. (1997). Minnesgpa Senteneing Quidellnes end ngmeppepy. St. Paul, MN. Minnesota Sentencing Guidelines Commission. (1992). Repprt pp phe Legielatpre pp Qpnppglled gdpepenee foepeee. St. Paul, MN. Moore, C.A., & Miethe, T.D. (1986). Regulated and unregulated sentencing decisions: An Analysis of first-year practices under Minnesota’s felony sentencing guidelines. Lew e Sgeiepy Review, 1Q(2), 253-277. Morris, N., & Tonry, M. (1990). Between priegp end probaeien: Intermediege pdniehmenpe in e raeienel eenpenelng eyepem. New York: Oxford University Press. Moulds, E.F. (1980). Chivalry and paternalism: Disparities of treatment in the criminal justice system. In S.K. Datesman & F.R. Scarpitti (Eds.), ngen, grime, and justice (pp. 277-299. New York: Oxford University Press. Myers, M.A. (1989). Symbolic policy and the sentencing of drug offenders. Law 1 Society Review, 11(2), 295-315. NM Nagel, I.H., & Johnson, B.L. (1994). The Role of gender in a structured sentencing system: Equal treatment, policy choices, and the sentencing of female offenders under the United States Sentencing Guidelines. The deurnal of Qriminal Law 1 Qriminology, 11(1), 181-221. Nagel, I.H., & Hagan, J. (1992). Gender and crime: Offense patterns and criminal court sanctions. In N. Morris & M. Tonry (Eds.), rime n t' e. (pp.91-144, V. 4). Chicago: University of Chicago Press. National Association of Drug Court Professionals, Drug Court Standards Committee (1997). D f'nin r : Th Key eempgnenes. Drug Courts Program Office, Office of Justice Programs, U.S. Department of Justice. Parent, D.G. (1988). Strueedrlng criminel eeneeneee; The Evelptien of Minneseea’e Senpeneing Qpidelinee. Stoneham, MA: Butterworth Legal Publishers. Parisi, N. (1982). Are Females treated differently? A Review of the theories and evidence on sentencing and parole decisions. In N. Rafter & E. Stanko (Eds.), dddge, lapyep, vleelm, thief (pp. 205-220). Boston, MA: Northeastern University Press. Paternoster, R., Brame, R., Mazerolle, P., and Piquero, A. (1998). Using the correct statistical test for the equality of regression coefficients. Qriminglogy, 11(4), 859-866. Petersilia, J. (1994). Debating crime and imprisonment in California. Evelpepiep end Planning, 11(2), 165-177. Pettiway, L.E. (1997). Werkin’ it. Philadelphia: Temple University Press. Platt, A.M. (1994). The Politics of law and order. Social Justice, 11(3), 3-13. Pollock, J.M. (1995). Gender, justice, and social control: A Historical perspective. In A.V. Merlo & J.M. Pollock (Eds.), meen, law, end social control (pp. 3-35). Boston, MA: Allyn and Bacon. Pollock-Byrne, J. (1990). Women, prison, and crime. Pacific Grove, CA: Brooks/Cole Publishing Company. 285 Proband, S.C. (1997). Rapid increase in women prisoners. Overcrowded Timee, 1(2), 4-5. Raeder, Myrna S. (1993). Gender and sentencing: Single moms, battered women, and other sex-based anomalies in the gender-free world of the federal sentencing guidelines. Pepperdine Law Review, 11(3), 905-990. Rafter, N. (1990). Partial justice: Women, prispns, and eocial generol. New Brunswick, NJ: Transaction Publishers. Rae, L.M., & Parker, R.A. (1992) Designing end eondpeeing surv r e r h- m r h n 'v 1 . San Francisco: Jossey-Bass Publishers. Ritchie, B.R. (1996). Compelled go erlme. New York: Routledge. Rothman, D.J. (1980). Conscience and convenience; The Asylum and its alternatives in progressive America. Boston: Little and Brown. Simon, R.J., & Landis, J. (1991). The Crimee wemen eemmie, the punishmente Lhey peceive. Lexington, MA: Lexington Books. Simpson, 8.8. (1989). Feminist theory, crime and justice. Criminology, 11(4), 605-631. Spohn, C., & Spears, J. (November, 1997). Gender and Sentencing of drug offendere. Paper presented at the 1997 Annual Meetings of the American Society of Criminology. San Diego, CA. Spohn, C., Welch, S., & Gruhl, J. (1985). Women defendants in court: The Interaction between sex and race in convicting and sentencing. §ociel scienee Qpezperly, 55(1), 178-185. Steffensmeier, D.J. (1980). Assessing the impact of the women’s movement on sex-based differences in the handling of adult criminal defendants. Crime 1 Delingpency, 11,344-357. Steffensmeier, D., Kramer, J., & Streifel, C. (1993). Gender and imprisonment decisions. Criminology, 11(3), 411-446 . 286 Steffensmeier, D., Ulmer, J., & Kramer, J. (1998). The Interaction of race, gender, and age in criminal sanctioning: The Punishment cost of being young, black, and male. Criminology, 11(4), 763-793. Steffensmeier, D. Kramer, J.H., & Ulmer, J. (1995). Age differences in sentencing. Justice Quarterly, 11(3), 701-719. Sterk, C.E. (1999). Fast lives: Women who use crack cocaine. Philadelphia: Temple University Press. Stolzenberg, R.M., & Relles, D.A. (1990). Theory testing in a world of constrained research design. Sociological Methods and Eeseereh, 11(4), 395-415. Tonry, M. (1995). Malign neglect-race, grime, and punishment in America. New York: Oxford University Press. Walker, S. (1993). Tamin the stem: The ntrol o discretion in Qriminal Justice, 1950-1990. New York: Oxford University Press. Wall, A. Personal communications through electronic mail. October 21, 1999. Wilbanks, W. (1986). Are Female Felons Treated More Leniently By the Criminal Justice System? Justice Quarterly, 1(4), 517-529. Winship, C., & Mare, R.D. (1992). Models for sample selection bias. Annual Review of Sociology, 11, 327- 350. Women in criminal justice: A Twenty year update special report. (October, 1998). Office of Justice Programs, Department of Justice. Washington, D.C. Young, V.D. (1986). Gender expectations and their impact on black female offenders and victims. Justice Quarterly, 1(3), 305-327. Zingraff, M., & Thomson, R. (1984). Differential sentencing of women and men in the U.S.A. Internation Journal of ehe gociology of Law, 11, 401-413. NW "llllllllflllllflli